diff --git a/clang/include/clang/Basic/riscv_vector.td b/clang/include/clang/Basic/riscv_vector.td
index bba2f20d06b96..3aa69b124a97a 100644
--- a/clang/include/clang/Basic/riscv_vector.td
+++ b/clang/include/clang/Basic/riscv_vector.td
@@ -371,10 +371,18 @@ let HasMaskedOffOperand = true in {
     defm "" : RVVOutOp0BuiltinSet<NAME, "xfd",
                                   [["vs", "vSv", "SvvSv"]]>;
   }
+  multiclass RVVFloatingReductionBuiltinRoundingMode {
+    defm "" : RVVOutOp0BuiltinSet<NAME, "xfd",
+                                  [["vs", "vSv", "SvvSvu"]]>;
+  }
   multiclass RVVFloatingWidenReductionBuiltin {
     defm "" : RVVOutOp0BuiltinSet<NAME, "xf",
                                   [["vs", "vSw", "SwvSw"]]>;
   }
+  multiclass RVVFloatingWidenReductionBuiltinRoundingMode {
+    defm "" : RVVOutOp0BuiltinSet<NAME, "xf",
+                                  [["vs", "vSw", "SwvSwu"]]>;
+  }
 }
 
 multiclass RVVIntReductionBuiltinSet
@@ -2469,12 +2477,65 @@ let HasMaskedOffOperand = true in {
 // 15.3. Vector Single-Width Floating-Point Reduction Instructions
 defm vfredmax : RVVFloatingReductionBuiltin;
 defm vfredmin : RVVFloatingReductionBuiltin;
-defm vfredusum : RVVFloatingReductionBuiltin;
-defm vfredosum : RVVFloatingReductionBuiltin;
+let ManualCodegen = [{
+  {
+    // LLVM intrinsic
+    // Unmasked: (passthru, op0, op1, round_mode, vl)
+    // Masked:   (passthru, vector_in, vector_in/scalar_in, mask, frm, vl, policy)
+
+    SmallVector<llvm::Value*, 7> Operands;
+    bool HasMaskedOff = !(
+        (IsMasked && (PolicyAttrs & RVV_VTA) && (PolicyAttrs & RVV_VMA)) ||
+        (!IsMasked && PolicyAttrs & RVV_VTA));
+    bool HasRoundModeOp = IsMasked ?
+      (HasMaskedOff ? Ops.size() == 6 : Ops.size() == 5) :
+      (HasMaskedOff ? Ops.size() == 5 : Ops.size() == 4);
+
+    unsigned Offset = IsMasked ?
+        (HasMaskedOff ? 2 : 1) : (HasMaskedOff ? 1 : 0);
+
+    if (!HasMaskedOff)
+      Operands.push_back(llvm::PoisonValue::get(ResultType));
+    else
+      Operands.push_back(Ops[IsMasked ? 1 : 0]);
 
-// 15.4. Vector Widening Floating-Point Reduction Instructions
-defm vfwredusum : RVVFloatingWidenReductionBuiltin;
-defm vfwredosum : RVVFloatingWidenReductionBuiltin;
+    Operands.push_back(Ops[Offset]); // op0
+    Operands.push_back(Ops[Offset + 1]); // op1
+
+    if (IsMasked)
+      Operands.push_back(Ops[0]); // mask
+
+    if (HasRoundModeOp) {
+      Operands.push_back(Ops[Offset + 2]); // frm
+      Operands.push_back(Ops[Offset + 3]); // vl
+    } else {
+      Operands.push_back(ConstantInt::get(Ops[Offset + 2]->getType(), 7)); // frm
+      Operands.push_back(Ops[Offset + 2]); // vl
+    }
+
+    IntrinsicTypes = {ResultType, Ops[Offset]->getType(),
+                      Ops.back()->getType()};
+    llvm::Function *F = CGM.getIntrinsic(ID, IntrinsicTypes);
+    return Builder.CreateCall(F, Operands, "");
+  }
+}] in {
+  let HasFRMRoundModeOp = 1 in {
+    // 15.3. Vector Single-Width Floating-Point Reduction Instructions
+    defm vfredusum : RVVFloatingReductionBuiltinRoundingMode;
+    defm vfredosum : RVVFloatingReductionBuiltinRoundingMode;
+
+    // 15.4. Vector Widening Floating-Point Reduction Instructions
+    defm vfwredusum : RVVFloatingWidenReductionBuiltinRoundingMode;
+    defm vfwredosum : RVVFloatingWidenReductionBuiltinRoundingMode;
+  }
+  // 15.3. Vector Single-Width Floating-Point Reduction Instructions
+  defm vfredusum : RVVFloatingReductionBuiltin;
+  defm vfredosum : RVVFloatingReductionBuiltin;
+
+  // 15.4. Vector Widening Floating-Point Reduction Instructions
+  defm vfwredusum : RVVFloatingWidenReductionBuiltin;
+  defm vfwredosum : RVVFloatingWidenReductionBuiltin;
+}
 }
 
 // 16. Vector Mask Instructions
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index 693ff523f822d..d0680b13c294c 100644
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -4842,6 +4842,10 @@ bool Sema::CheckRISCVBuiltinFunctionCall(const TargetInfo &TI,
   case RISCVVector::BI__builtin_rvv_vfrdiv_vf_rm:
   case RISCVVector::BI__builtin_rvv_vfwmul_vv_rm:
   case RISCVVector::BI__builtin_rvv_vfwmul_vf_rm:
+  case RISCVVector::BI__builtin_rvv_vfredosum_vs_rm:
+  case RISCVVector::BI__builtin_rvv_vfredusum_vs_rm:
+  case RISCVVector::BI__builtin_rvv_vfwredosum_vs_rm:
+  case RISCVVector::BI__builtin_rvv_vfwredusum_vs_rm:
   case RISCVVector::BI__builtin_rvv_vfsqrt_v_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfrec7_v_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfcvt_x_f_v_rm_tu:
@@ -4889,6 +4893,10 @@ bool Sema::CheckRISCVBuiltinFunctionCall(const TargetInfo &TI,
   case RISCVVector::BI__builtin_rvv_vfrdiv_vf_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfwmul_vv_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfwmul_vf_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfredosum_vs_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfredusum_vs_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwredosum_vs_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwredusum_vs_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfmacc_vv_rm:
   case RISCVVector::BI__builtin_rvv_vfmacc_vf_rm:
   case RISCVVector::BI__builtin_rvv_vfnmacc_vv_rm:
@@ -4957,6 +4965,10 @@ bool Sema::CheckRISCVBuiltinFunctionCall(const TargetInfo &TI,
   case RISCVVector::BI__builtin_rvv_vfrdiv_vf_rm_tama:
   case RISCVVector::BI__builtin_rvv_vfwmul_vv_rm_tama:
   case RISCVVector::BI__builtin_rvv_vfwmul_vf_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfredosum_vs_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfredusum_vs_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwredosum_vs_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwredusum_vs_rm_tama:
   case RISCVVector::BI__builtin_rvv_vfsqrt_v_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfrec7_v_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfcvt_x_f_v_rm_tum:
@@ -5065,6 +5077,10 @@ bool Sema::CheckRISCVBuiltinFunctionCall(const TargetInfo &TI,
   case RISCVVector::BI__builtin_rvv_vfwmsac_vf_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfwnmsac_vv_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfwnmsac_vf_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfredosum_vs_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfredusum_vs_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwredosum_vs_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwredusum_vs_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfadd_vv_rm_tumu:
   case RISCVVector::BI__builtin_rvv_vfadd_vf_rm_tumu:
   case RISCVVector::BI__builtin_rvv_vfsub_vv_rm_tumu:
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfredosum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfredosum.c
index e419aa49ca5db..75f5db44d1715 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfredosum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfredosum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1
 // CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1(vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1(vfloat16mf4_t vector, vfloat16m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1
 // CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1(vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1(vfloat16mf2_t vector, vfloat16m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m1_f16m1(vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat16m1_t test_vfredosum_vs_f16m1_f16m1(vfloat16m1_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1
 // CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m2_f16m1(vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat16m1_t test_vfredosum_vs_f16m2_f16m1(vfloat16m2_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1
 // CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m4_f16m1(vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat16m1_t test_vfredosum_vs_f16m4_f16m1(vfloat16m4_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1
 // CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m8_f16m1(vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat16m1_t test_vfredosum_vs_f16m8_f16m1(vfloat16m8_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1(vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1(vfloat32mf2_t vector, vfloat32m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m1_f32m1(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat32m1_t test_vfredosum_vs_f32m1_f32m1(vfloat32m1_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m2_f32m1(vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat32m1_t test_vfredosum_vs_f32m2_f32m1(vfloat32m2_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m4_f32m1(vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat32m1_t test_vfredosum_vs_f32m4_f32m1(vfloat32m4_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m8_f32m1(vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat32m1_t test_vfredosum_vs_f32m8_f32m1(vfloat32m8_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m1_f64m1(vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat64m1_t test_vfredosum_vs_f64m1_f64m1(vfloat64m1_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m2_f64m1(vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat64m1_t test_vfredosum_vs_f64m2_f64m1(vfloat64m2_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m4_f64m1(vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat64m1_t test_vfredosum_vs_f64m4_f64m1(vfloat64m4_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m8_f64m1(vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat64m1_t test_vfredosum_vs_f64m8_f64m1(vfloat64m8_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_m(vbool64_t mask, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_m(vbool64_t mask, vfloat16mf4_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_m(vbool32_t mask, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_m(vbool32_t mask, vfloat16mf2_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_m(vbool16_t mask, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_m(vbool16_t mask, vfloat16m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_m(vbool8_t mask, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_m(vbool8_t mask, vfloat16m2_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_m(vbool4_t mask, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_m(vbool4_t mask, vfloat16m4_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_m(vbool2_t mask, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -220,7 +220,7 @@ vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_m(vbool2_t mask, vfloat16m8_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_m(vbool64_t mask, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -230,7 +230,7 @@ vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_m(vbool64_t mask, vfloat32mf2_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -240,7 +240,7 @@ vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_m(vbool32_t mask, vfloat32m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_m(vbool16_t mask, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -250,7 +250,7 @@ vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_m(vbool16_t mask, vfloat32m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_m(vbool8_t mask, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -260,7 +260,7 @@ vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_m(vbool8_t mask, vfloat32m4_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_m(vbool4_t mask, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -270,7 +270,7 @@ vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_m(vbool4_t mask, vfloat32m8_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_m(vbool64_t mask, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -280,7 +280,7 @@ vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_m(vbool64_t mask, vfloat64m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_m(vbool32_t mask, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -290,7 +290,7 @@ vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_m(vbool32_t mask, vfloat64m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_m(vbool16_t mask, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -300,10 +300,310 @@ vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_m(vbool16_t mask, vfloat64m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_m(vbool8_t mask, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfredosum_vs_f64m8_f64m1_m(mask, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_rm(vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16mf4_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_rm(vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16mf2_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_rm(vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m1_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_rm(vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m2_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_rm(vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m4_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_rm(vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m8_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_rm(vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32mf2_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_rm(vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m2_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_rm(vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m4_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_rm(vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m8_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_rm(vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m1_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_rm(vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m2_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_rm(vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m4_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_rm(vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m8_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_rm_m(vbool64_t mask, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16mf4_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_rm_m(vbool32_t mask, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16mf2_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m1_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_rm_m(vbool8_t mask, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m2_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_rm_m(vbool4_t mask, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m4_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_rm_m(vbool2_t mask, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m8_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_rm_m(vbool64_t mask, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32mf2_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_rm_m(vbool16_t mask, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m2_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_rm_m(vbool8_t mask, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m4_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_rm_m(vbool4_t mask, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m8_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m1_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_rm_m(vbool32_t mask, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m2_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_rm_m(vbool16_t mask, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m4_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_rm_m(vbool8_t mask, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m8_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfredusum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfredusum.c
index 99d01f2075f1a..33000f7c3ae6c 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfredusum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfredusum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1
 // CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1(vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1(vfloat16mf4_t vector, vfloat16m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1
 // CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1(vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1(vfloat16mf2_t vector, vfloat16m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m1_f16m1(vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat16m1_t test_vfredusum_vs_f16m1_f16m1(vfloat16m1_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1
 // CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m2_f16m1(vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat16m1_t test_vfredusum_vs_f16m2_f16m1(vfloat16m2_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1
 // CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m4_f16m1(vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat16m1_t test_vfredusum_vs_f16m4_f16m1(vfloat16m4_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1
 // CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m8_f16m1(vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat16m1_t test_vfredusum_vs_f16m8_f16m1(vfloat16m8_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1(vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1(vfloat32mf2_t vector, vfloat32m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m1_f32m1(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat32m1_t test_vfredusum_vs_f32m1_f32m1(vfloat32m1_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m2_f32m1(vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat32m1_t test_vfredusum_vs_f32m2_f32m1(vfloat32m2_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m4_f32m1(vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat32m1_t test_vfredusum_vs_f32m4_f32m1(vfloat32m4_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m8_f32m1(vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat32m1_t test_vfredusum_vs_f32m8_f32m1(vfloat32m8_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m1_f64m1(vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat64m1_t test_vfredusum_vs_f64m1_f64m1(vfloat64m1_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m2_f64m1(vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat64m1_t test_vfredusum_vs_f64m2_f64m1(vfloat64m2_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m4_f64m1(vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat64m1_t test_vfredusum_vs_f64m4_f64m1(vfloat64m4_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m8_f64m1(vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat64m1_t test_vfredusum_vs_f64m8_f64m1(vfloat64m8_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_m(vbool64_t mask, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_m(vbool64_t mask, vfloat16mf4_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_m(vbool32_t mask, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_m(vbool32_t mask, vfloat16mf2_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_m(vbool16_t mask, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_m(vbool16_t mask, vfloat16m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_m(vbool8_t mask, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_m(vbool8_t mask, vfloat16m2_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_m(vbool4_t mask, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_m(vbool4_t mask, vfloat16m4_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_m(vbool2_t mask, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -220,7 +220,7 @@ vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_m(vbool2_t mask, vfloat16m8_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_m(vbool64_t mask, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -230,7 +230,7 @@ vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_m(vbool64_t mask, vfloat32mf2_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -240,7 +240,7 @@ vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_m(vbool32_t mask, vfloat32m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_m(vbool16_t mask, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -250,7 +250,7 @@ vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_m(vbool16_t mask, vfloat32m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_m(vbool8_t mask, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -260,7 +260,7 @@ vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_m(vbool8_t mask, vfloat32m4_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_m(vbool4_t mask, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -270,7 +270,7 @@ vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_m(vbool4_t mask, vfloat32m8_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_m(vbool64_t mask, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -280,7 +280,7 @@ vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_m(vbool64_t mask, vfloat64m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_m(vbool32_t mask, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -290,7 +290,7 @@ vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_m(vbool32_t mask, vfloat64m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_m(vbool16_t mask, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -300,10 +300,310 @@ vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_m(vbool16_t mask, vfloat64m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_m(vbool8_t mask, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfredusum_vs_f64m8_f64m1_m(mask, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_rm(vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16mf4_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_rm(vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16mf2_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_rm(vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m1_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_rm(vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m2_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_rm(vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m4_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_rm(vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m8_f16m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_rm(vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32mf2_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m1_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_rm(vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m2_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_rm(vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m4_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_rm(vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m8_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_rm(vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m1_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_rm(vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m2_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_rm(vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m4_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_rm(vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m8_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_rm_m(vbool64_t mask, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16mf4_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_rm_m(vbool32_t mask, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16mf2_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m1_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_rm_m(vbool8_t mask, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m2_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_rm_m(vbool4_t mask, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m4_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_rm_m(vbool2_t mask, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m8_f16m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_rm_m(vbool64_t mask, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32mf2_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m1_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_rm_m(vbool16_t mask, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m2_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_rm_m(vbool8_t mask, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m4_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_rm_m(vbool4_t mask, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m8_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m1_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_rm_m(vbool32_t mask, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m2_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_rm_m(vbool16_t mask, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m4_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_rm_m(vbool8_t mask, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m8_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwredosum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwredosum.c
index fe60447089305..364319f44dafe 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwredosum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwredosum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1
 // CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1(vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1(vfloat16mf4_t vector, vfloat32m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1(vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1(vfloat16mf2_t vector, vfloat32m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1(vfloat16m1_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1
 // CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1(vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1(vfloat16m2_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1
 // CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1(vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1(vfloat16m4_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1
 // CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1(vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1(vfloat16m8_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1(vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1(vfloat32mf2_t vector, vfloat64m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1(vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1(vfloat32m1_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1(vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1(vfloat32m2_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1(vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1(vfloat32m4_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1(vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1(vfloat32m8_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_m(vbool64_t mask, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_m(vbool64_t mask, vfloat16mf4_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_m(vbool32_t mask, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_m(vbool32_t mask, vfloat16mf2_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_m(vbool16_t mask, vfloat16m1_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_m(vbool8_t mask, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_m(vbool8_t mask, vfloat16m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_m(vbool4_t mask, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_m(vbool4_t mask, vfloat16m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_m(vbool2_t mask, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_m(vbool2_t mask, vfloat16m8_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_m(vbool64_t mask, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_m(vbool64_t mask, vfloat32mf2_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_m(vbool32_t mask, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_m(vbool32_t mask, vfloat32m1_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_m(vbool16_t mask, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_m(vbool16_t mask, vfloat32m2_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_m(vbool8_t mask, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -220,10 +220,230 @@ vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_m(vbool8_t mask, vfloat32m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_m(vbool4_t mask, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfwredosum_vs_f32m8_f64m1_m(mask, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_rm(vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16mf4_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_rm(vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16mf2_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m1_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_rm(vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m2_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_rm(vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m4_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_rm(vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m8_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_rm(vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32mf2_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_rm(vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m1_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_rm(vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m2_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_rm(vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m4_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_rm(vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m8_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_rm_m(vbool64_t mask, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16mf4_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_rm_m(vbool32_t mask, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16mf2_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m1_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_rm_m(vbool8_t mask, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m2_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_rm_m(vbool4_t mask, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m4_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_rm_m(vbool2_t mask, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m8_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_rm_m(vbool64_t mask, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32mf2_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m1_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_rm_m(vbool16_t mask, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m2_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_rm_m(vbool8_t mask, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m4_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_rm_m(vbool4_t mask, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m8_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwredusum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwredusum.c
index e751e0ba0dca9..501411ef51fcd 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwredusum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwredusum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1
 // CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1(vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1(vfloat16mf4_t vector, vfloat32m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1(vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1(vfloat16mf2_t vector, vfloat32m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1(vfloat16m1_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1
 // CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1(vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1(vfloat16m2_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1
 // CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1(vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1(vfloat16m4_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1
 // CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1(vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1(vfloat16m8_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1(vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1(vfloat32mf2_t vector, vfloat64m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1(vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1(vfloat32m1_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1(vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1(vfloat32m2_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1(vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1(vfloat32m4_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1(vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1(vfloat32m8_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_m(vbool64_t mask, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_m(vbool64_t mask, vfloat16mf4_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_m(vbool32_t mask, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_m(vbool32_t mask, vfloat16mf2_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_m(vbool16_t mask, vfloat16m1_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_m(vbool8_t mask, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_m(vbool8_t mask, vfloat16m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_m(vbool4_t mask, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_m(vbool4_t mask, vfloat16m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_m(vbool2_t mask, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_m(vbool2_t mask, vfloat16m8_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_m(vbool64_t mask, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_m(vbool64_t mask, vfloat32mf2_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_m(vbool32_t mask, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_m(vbool32_t mask, vfloat32m1_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_m(vbool16_t mask, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_m(vbool16_t mask, vfloat32m2_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_m(vbool8_t mask, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -220,10 +220,230 @@ vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_m(vbool8_t mask, vfloat32m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_m(vbool4_t mask, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfwredusum_vs_f32m8_f64m1_m(mask, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_rm(vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16mf4_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_rm(vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16mf2_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m1_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_rm(vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m2_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_rm(vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m4_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_rm(vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m8_f32m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_rm(vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32mf2_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_rm(vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m1_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_rm(vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m2_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_rm(vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m4_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_rm(vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m8_f64m1_rm(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_rm_m(vbool64_t mask, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16mf4_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_rm_m(vbool32_t mask, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16mf2_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m1_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_rm_m(vbool8_t mask, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m2_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_rm_m(vbool4_t mask, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m4_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_rm_m(vbool2_t mask, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m8_f32m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_rm_m(vbool64_t mask, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32mf2_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m1_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_rm_m(vbool16_t mask, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m2_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_rm_m(vbool8_t mask, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m4_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_rm_m(vbool4_t mask, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m8_f64m1_rm_m(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfredosum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfredosum.c
index 4eb54775f2fb0..66888f1b0b849 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfredosum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfredosum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1
 // CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1(vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1(vfloat16mf4_t vector, vfloat16m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1
 // CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1(vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1(vfloat16mf2_t vector, vfloat16m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m1_f16m1(vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat16m1_t test_vfredosum_vs_f16m1_f16m1(vfloat16m1_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1
 // CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m2_f16m1(vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat16m1_t test_vfredosum_vs_f16m2_f16m1(vfloat16m2_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1
 // CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m4_f16m1(vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat16m1_t test_vfredosum_vs_f16m4_f16m1(vfloat16m4_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1
 // CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m8_f16m1(vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat16m1_t test_vfredosum_vs_f16m8_f16m1(vfloat16m8_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1(vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1(vfloat32mf2_t vector, vfloat32m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m1_f32m1(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat32m1_t test_vfredosum_vs_f32m1_f32m1(vfloat32m1_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m2_f32m1(vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat32m1_t test_vfredosum_vs_f32m2_f32m1(vfloat32m2_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m4_f32m1(vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat32m1_t test_vfredosum_vs_f32m4_f32m1(vfloat32m4_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m8_f32m1(vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat32m1_t test_vfredosum_vs_f32m8_f32m1(vfloat32m8_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m1_f64m1(vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat64m1_t test_vfredosum_vs_f64m1_f64m1(vfloat64m1_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m2_f64m1(vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat64m1_t test_vfredosum_vs_f64m2_f64m1(vfloat64m2_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m4_f64m1(vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat64m1_t test_vfredosum_vs_f64m4_f64m1(vfloat64m4_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m8_f64m1(vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat64m1_t test_vfredosum_vs_f64m8_f64m1(vfloat64m8_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_m(vbool64_t mask, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_m(vbool64_t mask, vfloat16mf4_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_m(vbool32_t mask, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_m(vbool32_t mask, vfloat16mf2_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_m(vbool16_t mask, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_m(vbool16_t mask, vfloat16m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_m(vbool8_t mask, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_m(vbool8_t mask, vfloat16m2_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_m(vbool4_t mask, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_m(vbool4_t mask, vfloat16m4_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_m(vbool2_t mask, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -220,7 +220,7 @@ vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_m(vbool2_t mask, vfloat16m8_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_m(vbool64_t mask, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -230,7 +230,7 @@ vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_m(vbool64_t mask, vfloat32mf2_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -240,7 +240,7 @@ vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_m(vbool32_t mask, vfloat32m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_m(vbool16_t mask, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -250,7 +250,7 @@ vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_m(vbool16_t mask, vfloat32m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_m(vbool8_t mask, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -260,7 +260,7 @@ vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_m(vbool8_t mask, vfloat32m4_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_m(vbool4_t mask, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -270,7 +270,7 @@ vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_m(vbool4_t mask, vfloat32m8_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_m(vbool64_t mask, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -280,7 +280,7 @@ vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_m(vbool64_t mask, vfloat64m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_m(vbool32_t mask, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -290,7 +290,7 @@ vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_m(vbool32_t mask, vfloat64m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_m(vbool16_t mask, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -300,10 +300,310 @@ vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_m(vbool16_t mask, vfloat64m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_m(vbool8_t mask, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfredosum(mask, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_rm(vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_rm(vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_rm(vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_rm(vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_rm(vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_rm(vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_rm(vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_rm(vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_rm(vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_rm(vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_rm(vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_rm(vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_rm(vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_rm(vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_rm_m(vbool64_t mask, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_rm_m(vbool32_t mask, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_rm_m(vbool8_t mask, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_rm_m(vbool4_t mask, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_rm_m(vbool2_t mask, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_rm_m(vbool64_t mask, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_rm_m(vbool16_t mask, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_rm_m(vbool8_t mask, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_rm_m(vbool4_t mask, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_rm_m(vbool32_t mask, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_rm_m(vbool16_t mask, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_rm_m(vbool8_t mask, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfredusum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfredusum.c
index 82ae943d90729..77ff64bbfe94b 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfredusum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfredusum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1
 // CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1(vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1(vfloat16mf4_t vector, vfloat16m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1
 // CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1(vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1(vfloat16mf2_t vector, vfloat16m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m1_f16m1(vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat16m1_t test_vfredusum_vs_f16m1_f16m1(vfloat16m1_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1
 // CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m2_f16m1(vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat16m1_t test_vfredusum_vs_f16m2_f16m1(vfloat16m2_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1
 // CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m4_f16m1(vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat16m1_t test_vfredusum_vs_f16m4_f16m1(vfloat16m4_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1
 // CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m8_f16m1(vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat16m1_t test_vfredusum_vs_f16m8_f16m1(vfloat16m8_t vector, vfloat16m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1(vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1(vfloat32mf2_t vector, vfloat32m1_t s
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m1_f32m1(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat32m1_t test_vfredusum_vs_f32m1_f32m1(vfloat32m1_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m2_f32m1(vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat32m1_t test_vfredusum_vs_f32m2_f32m1(vfloat32m2_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m4_f32m1(vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat32m1_t test_vfredusum_vs_f32m4_f32m1(vfloat32m4_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m8_f32m1(vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat32m1_t test_vfredusum_vs_f32m8_f32m1(vfloat32m8_t vector, vfloat32m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m1_f64m1(vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat64m1_t test_vfredusum_vs_f64m1_f64m1(vfloat64m1_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m2_f64m1(vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat64m1_t test_vfredusum_vs_f64m2_f64m1(vfloat64m2_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m4_f64m1(vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat64m1_t test_vfredusum_vs_f64m4_f64m1(vfloat64m4_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m8_f64m1(vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat64m1_t test_vfredusum_vs_f64m8_f64m1(vfloat64m8_t vector, vfloat64m1_t sca
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_m(vbool64_t mask, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_m(vbool64_t mask, vfloat16mf4_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_m(vbool32_t mask, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_m(vbool32_t mask, vfloat16mf2_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_m(vbool16_t mask, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_m(vbool16_t mask, vfloat16m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_m(vbool8_t mask, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_m(vbool8_t mask, vfloat16m2_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_m(vbool4_t mask, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_m(vbool4_t mask, vfloat16m4_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_m
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_m(vbool2_t mask, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -220,7 +220,7 @@ vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_m(vbool2_t mask, vfloat16m8_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_m(vbool64_t mask, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -230,7 +230,7 @@ vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_m(vbool64_t mask, vfloat32mf2_t vect
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -240,7 +240,7 @@ vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_m(vbool32_t mask, vfloat32m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_m(vbool16_t mask, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -250,7 +250,7 @@ vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_m(vbool16_t mask, vfloat32m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_m(vbool8_t mask, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -260,7 +260,7 @@ vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_m(vbool8_t mask, vfloat32m4_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_m(vbool4_t mask, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -270,7 +270,7 @@ vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_m(vbool4_t mask, vfloat32m8_t vector,
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_m(vbool64_t mask, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -280,7 +280,7 @@ vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_m(vbool64_t mask, vfloat64m1_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_m(vbool32_t mask, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -290,7 +290,7 @@ vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_m(vbool32_t mask, vfloat64m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_m(vbool16_t mask, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -300,10 +300,310 @@ vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_m(vbool16_t mask, vfloat64m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_m(vbool8_t mask, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfredusum(mask, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_rm(vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_rm(vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_rm(vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_rm(vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_rm(vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_rm
+// CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_rm(vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_rm(vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_rm(vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_rm(vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_rm(vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_rm(vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_rm(vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_rm(vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_rm(vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_rm_m(vbool64_t mask, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_rm_m(vbool32_t mask, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_rm_m(vbool8_t mask, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_rm_m(vbool4_t mask, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_rm_m(vbool2_t mask, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_rm_m(vbool64_t mask, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_rm_m(vbool16_t mask, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_rm_m(vbool8_t mask, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_rm_m(vbool4_t mask, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> poison, <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> poison, <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_rm_m(vbool32_t mask, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> poison, <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_rm_m(vbool16_t mask, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> poison, <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_rm_m(vbool8_t mask, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwredosum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwredosum.c
index 1a519d057fefb..055175633f0d9 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwredosum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwredosum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1
 // CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1(vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1(vfloat16mf4_t vector, vfloat32m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1(vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1(vfloat16mf2_t vector, vfloat32m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1(vfloat16m1_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1
 // CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1(vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1(vfloat16m2_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1
 // CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1(vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1(vfloat16m4_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1
 // CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1(vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1(vfloat16m8_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1(vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1(vfloat32mf2_t vector, vfloat64m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1(vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1(vfloat32m1_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1(vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1(vfloat32m2_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1(vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1(vfloat32m4_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1(vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1(vfloat32m8_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_m(vbool64_t mask, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_m(vbool64_t mask, vfloat16mf4_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_m(vbool32_t mask, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_m(vbool32_t mask, vfloat16mf2_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_m(vbool16_t mask, vfloat16m1_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_m(vbool8_t mask, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_m(vbool8_t mask, vfloat16m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_m(vbool4_t mask, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_m(vbool4_t mask, vfloat16m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_m(vbool2_t mask, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_m(vbool2_t mask, vfloat16m8_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_m(vbool64_t mask, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_m(vbool64_t mask, vfloat32mf2_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_m(vbool32_t mask, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_m(vbool32_t mask, vfloat32m1_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_m(vbool16_t mask, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_m(vbool16_t mask, vfloat32m2_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_m(vbool8_t mask, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -220,10 +220,230 @@ vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_m(vbool8_t mask, vfloat32m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_m(vbool4_t mask, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfwredosum(mask, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_rm(vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_rm(vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_rm(vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_rm(vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_rm(vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_rm(vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_rm(vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_rm(vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_rm(vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_rm(vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_rm_m(vbool64_t mask, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_rm_m(vbool32_t mask, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_rm_m(vbool8_t mask, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_rm_m(vbool4_t mask, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_rm_m(vbool2_t mask, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_rm_m(vbool64_t mask, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_rm_m(vbool16_t mask, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_rm_m(vbool8_t mask, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_rm_m(vbool4_t mask, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwredusum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwredusum.c
index e83b687a49047..fbd20e3d2bf8e 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwredusum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwredusum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1
 // CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1(vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1(vfloat16mf4_t vector, vfloat32m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1(vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1(vfloat16mf2_t vector, vfloat32m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1(vfloat16m1_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1
 // CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1(vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1(vfloat16m2_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1
 // CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1(vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1(vfloat16m4_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1
 // CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1(vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1(vfloat16m8_t vector, vfloat32m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1(vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1(vfloat32mf2_t vector, vfloat64m1_t
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1(vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1(vfloat32m1_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1(vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1(vfloat32m2_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1(vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1(vfloat32m4_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1(vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1(vfloat32m8_t vector, vfloat64m1_t sc
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_m(vbool64_t mask, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_m(vbool64_t mask, vfloat16mf4_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_m(vbool32_t mask, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_m(vbool32_t mask, vfloat16mf2_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_m(vbool16_t mask, vfloat16m1_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_m(vbool8_t mask, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_m(vbool8_t mask, vfloat16m2_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_m(vbool4_t mask, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_m(vbool4_t mask, vfloat16m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_m(vbool2_t mask, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_m(vbool2_t mask, vfloat16m8_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_m(vbool64_t mask, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_m(vbool64_t mask, vfloat32mf2_t vec
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_m(vbool32_t mask, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_m(vbool32_t mask, vfloat32m1_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_m(vbool16_t mask, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_m(vbool16_t mask, vfloat32m2_t vecto
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_m(vbool8_t mask, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -220,10 +220,230 @@ vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_m(vbool8_t mask, vfloat32m4_t vector
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_m(vbool4_t mask, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfwredusum(mask, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_rm(vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_rm(vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_rm(vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_rm(vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_rm(vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_rm(vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_rm(vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_rm(vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_rm(vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_rm(vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> poison, <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_rm_m(vbool64_t mask, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> poison, <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_rm_m(vbool32_t mask, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> poison, <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> poison, <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_rm_m(vbool8_t mask, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> poison, <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_rm_m(vbool4_t mask, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> poison, <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_rm_m(vbool2_t mask, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> poison, <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_rm_m(vbool64_t mask, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> poison, <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> poison, <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_rm_m(vbool16_t mask, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> poison, <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_rm_m(vbool8_t mask, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> poison, <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_rm_m(vbool4_t mask, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum(mask, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfredosum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfredosum.c
index 658af50af22f9..964c4f61011ea 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfredosum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfredosum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_tum(vbool64_t mask, vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_tum(vbool64_t mask, vfloat16m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_tum(vbool32_t mask, vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_tum(vbool32_t mask, vfloat16m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_tum(vbool16_t mask, vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_tum(vbool16_t mask, vfloat16m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_tum(vbool8_t mask, vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_tum(vbool8_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_tum(vbool4_t mask, vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_tum(vbool4_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_tum(vbool2_t mask, vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -220,7 +220,7 @@ vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_tum(vbool2_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -230,7 +230,7 @@ vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_tum(vbool64_t mask, vfloat32m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -240,7 +240,7 @@ vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_tum(vbool32_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -250,7 +250,7 @@ vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_tum(vbool16_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -260,7 +260,7 @@ vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_tum(vbool8_t mask, vfloat32m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -270,7 +270,7 @@ vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_tum(vbool4_t mask, vfloat32m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -280,7 +280,7 @@ vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_tum(vbool64_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -290,7 +290,7 @@ vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_tum(vbool32_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -300,10 +300,310 @@ vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_tum(vbool16_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfredosum_vs_f64m8_f64m1_tum(mask, maskedoff, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16mf4_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16mf2_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m1_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m2_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m4_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m8_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32mf2_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m2_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m4_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m8_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m1_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m2_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m4_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m8_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_rm_tum(vbool64_t mask, vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16mf4_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_rm_tum(vbool32_t mask, vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16mf2_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_rm_tum(vbool16_t mask, vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m1_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_rm_tum(vbool8_t mask, vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m2_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_rm_tum(vbool4_t mask, vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m4_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_rm_tum(vbool2_t mask, vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f16m8_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_rm_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32mf2_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m2_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_rm_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m4_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_rm_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f32m8_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m1_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_rm_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m2_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_rm_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m4_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_rm_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_vs_f64m8_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfredusum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfredusum.c
index e8755058d5209..b3c6b5e0902d1 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfredusum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfredusum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_tum(vbool64_t mask, vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_tum(vbool64_t mask, vfloat16m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_tum(vbool32_t mask, vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_tum(vbool32_t mask, vfloat16m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_tum(vbool16_t mask, vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_tum(vbool16_t mask, vfloat16m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_tum(vbool8_t mask, vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_tum(vbool8_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_tum(vbool4_t mask, vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_tum(vbool4_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_tum(vbool2_t mask, vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -220,7 +220,7 @@ vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_tum(vbool2_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -230,7 +230,7 @@ vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_tum(vbool64_t mask, vfloat32m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -240,7 +240,7 @@ vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_tum(vbool32_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -250,7 +250,7 @@ vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_tum(vbool16_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -260,7 +260,7 @@ vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_tum(vbool8_t mask, vfloat32m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -270,7 +270,7 @@ vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_tum(vbool4_t mask, vfloat32m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -280,7 +280,7 @@ vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_tum(vbool64_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -290,7 +290,7 @@ vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_tum(vbool32_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -300,10 +300,310 @@ vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_tum(vbool16_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfredusum_vs_f64m8_f64m1_tum(mask, maskedoff, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16mf4_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16mf2_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m1_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m2_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m4_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m8_f16m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32mf2_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m1_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m2_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m4_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m8_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m1_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m2_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m4_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m8_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_rm_tum(vbool64_t mask, vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16mf4_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_rm_tum(vbool32_t mask, vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16mf2_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_rm_tum(vbool16_t mask, vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m1_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_rm_tum(vbool8_t mask, vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m2_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_rm_tum(vbool4_t mask, vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m4_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_rm_tum(vbool2_t mask, vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f16m8_f16m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_rm_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32mf2_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m1_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m2_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_rm_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m4_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_rm_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f32m8_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m1_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_rm_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m2_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_rm_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m4_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_rm_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_vs_f64m8_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwredosum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwredosum.c
index 4347cf070bd50..90c06b09b7c4f 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwredosum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwredosum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_tum(vbool64_t mask, vfloat32m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_tum(vbool32_t mask, vfloat32m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_tum(vbool16_t mask, vfloat32m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_tum(vbool8_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_tum(vbool4_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_tum(vbool2_t mask, vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_tum(vbool2_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_tum(vbool64_t mask, vfloat64m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_tum(vbool32_t mask, vfloat64m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_tum(vbool16_t mask, vfloat64m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -220,10 +220,230 @@ vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_tum(vbool8_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_tum(vbool4_t mask, vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfwredosum_vs_f32m8_f64m1_tum(mask, maskedoff, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16mf4_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16mf2_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m1_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m2_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m4_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m8_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32mf2_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m1_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m2_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m4_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m8_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_rm_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16mf4_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16mf2_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m1_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_rm_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m2_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_rm_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m4_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_rm_tum(vbool2_t mask, vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f16m8_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32mf2_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_rm_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m1_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_rm_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m2_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_rm_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m4_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_rm_tum(vbool4_t mask, vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_vs_f32m8_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwredusum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwredusum.c
index b6f5434b9ea63..920bf9558e87b 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwredusum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwredusum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_tum(vbool64_t mask, vfloat32m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_tum(vbool32_t mask, vfloat32m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_tum(vbool16_t mask, vfloat32m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_tum(vbool8_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_tum(vbool4_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_tum(vbool2_t mask, vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_tum(vbool2_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_tum(vbool64_t mask, vfloat64m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_tum(vbool32_t mask, vfloat64m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_tum(vbool16_t mask, vfloat64m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -220,10 +220,230 @@ vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_tum(vbool8_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_tum(vbool4_t mask, vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfwredusum_vs_f32m8_f64m1_tum(mask, maskedoff, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16mf4_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16mf2_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m1_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m2_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m4_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m8_f32m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32mf2_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m1_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m2_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m4_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m8_f64m1_rm_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_rm_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16mf4_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16mf2_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m1_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_rm_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m2_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_rm_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m4_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_rm_tum(vbool2_t mask, vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f16m8_f32m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32mf2_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_rm_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m1_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_rm_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m2_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_rm_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m4_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_rm_tum(vbool4_t mask, vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_vs_f32m8_f64m1_rm_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfredosum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfredosum.c
index c0d2a1448ab2a..d2ea8d1d5be6e 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfredosum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfredosum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_tum(vbool64_t mask, vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_tum(vbool64_t mask, vfloat16m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_tum(vbool32_t mask, vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_tum(vbool32_t mask, vfloat16m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_tum(vbool16_t mask, vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_tum(vbool16_t mask, vfloat16m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_tum(vbool8_t mask, vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_tum(vbool8_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_tum(vbool4_t mask, vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_tum(vbool4_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_tum(vbool2_t mask, vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -220,7 +220,7 @@ vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_tum(vbool2_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -230,7 +230,7 @@ vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_tum(vbool64_t mask, vfloat32m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -240,7 +240,7 @@ vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_tum(vbool32_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -250,7 +250,7 @@ vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_tum(vbool16_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -260,7 +260,7 @@ vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_tum(vbool8_t mask, vfloat32m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -270,7 +270,7 @@ vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_tum(vbool4_t mask, vfloat32m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -280,7 +280,7 @@ vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_tum(vbool64_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -290,7 +290,7 @@ vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_tum(vbool32_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -300,10 +300,310 @@ vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_tum(vbool16_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf4_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf4_f16m1_rm_tum(vbool64_t mask, vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16mf2_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16mf2_f16m1_rm_tum(vbool32_t mask, vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m1_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m1_f16m1_rm_tum(vbool16_t mask, vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m2_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m2_f16m1_rm_tum(vbool8_t mask, vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m4_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m4_f16m1_rm_tum(vbool4_t mask, vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredosum_vs_f16m8_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredosum_vs_f16m8_f16m1_rm_tum(vbool2_t mask, vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32mf2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32mf2_f32m1_rm_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m1_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m2_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m4_f32m1_rm_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredosum_vs_f32m8_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredosum_vs_f32m8_f32m1_rm_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m1_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m1_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m2_f64m1_rm_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m4_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m4_f64m1_rm_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredosum_vs_f64m8_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredosum_vs_f64m8_f64m1_rm_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfredusum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfredusum.c
index 7c1587f345722..ad13105909d1a 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfredusum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfredusum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_tu
 // CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_tu(vfloat16m1_t maskedoff, vfloat16m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat32m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m1
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m2
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m4
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat64m8
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_tum(vbool64_t mask, vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_tum(vbool64_t mask, vfloat16m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_tum(vbool32_t mask, vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_tum(vbool32_t mask, vfloat16m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_tum(vbool16_t mask, vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_tum(vbool16_t mask, vfloat16m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_tum(vbool8_t mask, vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_tum(vbool8_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_tum(vbool4_t mask, vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_tum(vbool4_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_tum
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
 vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_tum(vbool2_t mask, vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
@@ -220,7 +220,7 @@ vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_tum(vbool2_t mask, vfloat16m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -230,7 +230,7 @@ vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_tum(vbool64_t mask, vfloat32m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -240,7 +240,7 @@ vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_tum(vbool32_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -250,7 +250,7 @@ vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_tum(vbool16_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -260,7 +260,7 @@ vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_tum(vbool8_t mask, vfloat32m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -270,7 +270,7 @@ vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_tum(vbool4_t mask, vfloat32m1_t maske
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -280,7 +280,7 @@ vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_tum(vbool64_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -290,7 +290,7 @@ vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_tum(vbool32_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -300,10 +300,310 @@ vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_tum(vbool16_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_rm_tu(vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf4_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf4_f16m1_rm_tum(vbool64_t mask, vfloat16m1_t maskedoff, vfloat16mf4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16mf2_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16mf2_f16m1_rm_tum(vbool32_t mask, vfloat16m1_t maskedoff, vfloat16mf2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m1_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m1_f16m1_rm_tum(vbool16_t mask, vfloat16m1_t maskedoff, vfloat16m1_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m2_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m2_f16m1_rm_tum(vbool8_t mask, vfloat16m1_t maskedoff, vfloat16m2_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m4_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m4_f16m1_rm_tum(vbool4_t mask, vfloat16m1_t maskedoff, vfloat16m4_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x half> @test_vfredusum_vs_f16m8_f16m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 4 x half> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 4 x half> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.i64(<vscale x 4 x half> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 4 x half> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
+//
+vfloat16m1_t test_vfredusum_vs_f16m8_f16m1_rm_tum(vbool2_t mask, vfloat16m1_t maskedoff, vfloat16m8_t vector, vfloat16m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32mf2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32mf2_f32m1_rm_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat32mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m1_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m2_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat32m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m4_f32m1_rm_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat32m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfredusum_vs_f32m8_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfredusum_vs_f32m8_f32m1_rm_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat32m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m1_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m1_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat64m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m2_f64m1_rm_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat64m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m4_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m4_f64m1_rm_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat64m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfredusum_vs_f64m8_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x double> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x double> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfredusum_vs_f64m8_f64m1_rm_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat64m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwredosum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwredosum.c
index 85f787dd030d1..0edb20720400e 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwredosum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwredosum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_tum(vbool64_t mask, vfloat32m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_tum(vbool32_t mask, vfloat32m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_tum(vbool16_t mask, vfloat32m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_tum(vbool8_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_tum(vbool4_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_tum(vbool2_t mask, vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_tum(vbool2_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_tum(vbool64_t mask, vfloat64m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_tum(vbool32_t mask, vfloat64m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_tum(vbool16_t mask, vfloat64m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -220,10 +220,230 @@ vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_tum(vbool8_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_tum(vbool4_t mask, vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf4_f32m1_rm_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16mf2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16mf2_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m1_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m2_f32m1_rm_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m4_f32m1_rm_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredosum_vs_f16m8_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredosum_vs_f16m8_f32m1_rm_tum(vbool2_t mask, vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32mf2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32mf2_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m1_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m1_f64m1_rm_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m2_f64m1_rm_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m4_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m4_f64m1_rm_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredosum_vs_f32m8_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredosum_vs_f32m8_f64m1_rm_tum(vbool4_t mask, vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredosum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwredusum.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwredusum.c
index 6bd3ae1e7624a..b0a4cae7a6e84 100644
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwredusum.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwredusum.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -20,7 +20,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -30,7 +30,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -40,7 +40,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -50,7 +50,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -60,7 +60,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -70,7 +70,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_tu(vfloat32m1_t maskedoff, vfloat16m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -80,7 +80,7 @@ vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -90,7 +90,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -100,7 +100,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -110,7 +110,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -120,7 +120,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_tu(vfloat64m1_t maskedoff, vfloat32m
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -130,7 +130,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_tum(vbool64_t mask, vfloat32m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -140,7 +140,7 @@ vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_tum(vbool32_t mask, vfloat32m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -150,7 +150,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_tum(vbool16_t mask, vfloat32m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -160,7 +160,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_tum(vbool8_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -170,7 +170,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_tum(vbool4_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_tum(vbool2_t mask, vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
@@ -180,7 +180,7 @@ vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_tum(vbool2_t mask, vfloat32m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -190,7 +190,7 @@ vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_tum(vbool64_t mask, vfloat64m1_t ma
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -200,7 +200,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_tum(vbool32_t mask, vfloat64m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -210,7 +210,7 @@ vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_tum(vbool16_t mask, vfloat64m1_t mas
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
@@ -220,10 +220,230 @@ vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_tum(vbool8_t mask, vfloat64m1_t mask
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 [[VL]])
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_tum(vbool4_t mask, vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
   return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_rm_tu(vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tu(maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 1 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 1 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf4_f32m1_rm_tum(vbool64_t mask, vfloat32m1_t maskedoff, vfloat16mf4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16mf2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 2 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 2 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16mf2_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat16mf2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m1_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 4 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 4 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m2_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 8 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 8 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m2_f32m1_rm_tum(vbool8_t mask, vfloat32m1_t maskedoff, vfloat16m2_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m4_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 16 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 16 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m4_f32m1_rm_tum(vbool4_t mask, vfloat32m1_t maskedoff, vfloat16m4_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwredusum_vs_f16m8_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 32 x i1> [[MASK:%.*]], <vscale x 2 x float> [[MASKEDOFF:%.*]], <vscale x 32 x half> [[VECTOR:%.*]], <vscale x 2 x float> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16.i64(<vscale x 2 x float> [[MASKEDOFF]], <vscale x 32 x half> [[VECTOR]], <vscale x 2 x float> [[SCALAR]], <vscale x 32 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwredusum_vs_f16m8_f32m1_rm_tum(vbool2_t mask, vfloat32m1_t maskedoff, vfloat16m8_t vector, vfloat32m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32mf2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 1 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 1 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32mf2_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t maskedoff, vfloat32mf2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m1_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 2 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 2 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m1_f64m1_rm_tum(vbool32_t mask, vfloat64m1_t maskedoff, vfloat32m1_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m2_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 4 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 4 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m2_f64m1_rm_tum(vbool16_t mask, vfloat64m1_t maskedoff, vfloat32m2_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m4_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 8 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 8 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m4_f64m1_rm_tum(vbool8_t mask, vfloat64m1_t maskedoff, vfloat32m4_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwredusum_vs_f32m8_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 1 x double> [[MASKEDOFF:%.*]], <vscale x 16 x float> [[VECTOR:%.*]], <vscale x 1 x double> [[SCALAR:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.i64(<vscale x 1 x double> [[MASKEDOFF]], <vscale x 16 x float> [[VECTOR]], <vscale x 1 x double> [[SCALAR]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwredusum_vs_f32m8_f64m1_rm_tum(vbool4_t mask, vfloat64m1_t maskedoff, vfloat32m8_t vector, vfloat64m1_t scalar, size_t vl) {
+  return __riscv_vfwredusum_tum(mask, maskedoff, vector, scalar, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfredosum-out-of-range.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfredosum-out-of-range.c
new file mode 100644
index 0000000000000..2a99737e06570
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfredosum-out-of-range.c
@@ -0,0 +1,26 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
+// RUN:   -fsyntax-only -verify %s
+
+#include <riscv_vector.h>
+
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm(vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm_m(mask, vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm_tu(maskedoff, vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfredosum_vs_f32m1_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfredosum_vs_f32m1_f32m1_rm_tum(mask, maskedoff, vector, scalar, 5, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfredusum-out-of-range.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfredusum-out-of-range.c
new file mode 100644
index 0000000000000..36faa0f60c425
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfredusum-out-of-range.c
@@ -0,0 +1,26 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
+// RUN:   -fsyntax-only -verify %s
+
+#include <riscv_vector.h>
+
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm(vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfredusum_vs_f32m1_f32m1_rm(vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfredusum_vs_f32m1_f32m1_rm_m(mask, vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfredusum_vs_f32m1_f32m1_rm_tu(maskedoff, vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfredusum_vs_f32m1_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t maskedoff, vfloat32m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfredusum_vs_f32m1_f32m1_rm_tum(mask, maskedoff, vector, scalar, 5, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwredosum-out-of-range.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwredosum-out-of-range.c
new file mode 100644
index 0000000000000..df35fda7b537f
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwredosum-out-of-range.c
@@ -0,0 +1,26 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
+// RUN:   -fsyntax-only -verify %s
+
+#include <riscv_vector.h>
+
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwredosum_vs_f16m1_f32m1_rm(vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwredosum_vs_f16m1_f32m1_rm_m(mask, vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwredosum_vs_f16m1_f32m1_rm_tu(maskedoff, vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfwredosum_vs_f16m1_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwredosum_vs_f16m1_f32m1_rm_tum(mask, maskedoff, vector, scalar, 5, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwredusum-out-of-range.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwredusum-out-of-range.c
new file mode 100644
index 0000000000000..555fcab4d4d01
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwredusum-out-of-range.c
@@ -0,0 +1,26 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
+// RUN:   -fsyntax-only -verify %s
+
+#include <riscv_vector.h>
+
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm(vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwredusum_vs_f16m1_f32m1_rm(vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_m(vbool16_t mask, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwredusum_vs_f16m1_f32m1_rm_m(mask, vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_tu(vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwredusum_vs_f16m1_f32m1_rm_tu(maskedoff, vector, scalar, 5, vl);
+}
+
+vfloat32m1_t test_vfwredusum_vs_f16m1_f32m1_rm_tum(vbool16_t mask, vfloat32m1_t maskedoff, vfloat16m1_t vector, vfloat32m1_t scalar, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwredusum_vs_f16m1_f32m1_rm_tum(mask, maskedoff, vector, scalar, 5, vl);
+}
diff --git a/llvm/include/llvm/IR/IntrinsicsRISCV.td b/llvm/include/llvm/IR/IntrinsicsRISCV.td
index 64d2d61e53e5c..cfadbd6d2fa49 100644
--- a/llvm/include/llvm/IR/IntrinsicsRISCV.td
+++ b/llvm/include/llvm/IR/IntrinsicsRISCV.td
@@ -887,6 +887,28 @@ let TargetPrefix = "riscv" in {
                     [IntrNoMem]>, RISCVVIntrinsic {
     let VLOperand = 4;
   }
+  // For Reduction ternary operations.
+  // For destination vector type is the same as first and third source vector.
+  // Input: (vector_in, vector_in, vector_in, frm, vl)
+  class RISCVReductionUnMaskedRoundingMode
+        : DefaultAttrsIntrinsic<[llvm_anyvector_ty],
+                    [LLVMMatchType<0>, llvm_anyvector_ty, LLVMMatchType<0>,
+                     llvm_anyint_ty, LLVMMatchType<2>],
+                    [ImmArg<ArgIndex<3>>, IntrNoMem]>, RISCVVIntrinsic {
+    let VLOperand = 4;
+  }
+  // For Reduction ternary operations with mask.
+  // For destination vector type is the same as first and third source vector.
+  // The mask type come from second source vector.
+  // Input: (vector_in, vector_in, vector_in, mask, frm, vl)
+  class RISCVReductionMaskedRoundingMode
+        : DefaultAttrsIntrinsic<[llvm_anyvector_ty],
+                    [LLVMMatchType<0>, llvm_anyvector_ty, LLVMMatchType<0>,
+                     LLVMScalarOrSameVectorWidth<1, llvm_i1_ty>, llvm_anyint_ty,
+                     LLVMMatchType<2>],
+                    [ImmArg<ArgIndex<4>>, IntrNoMem]>, RISCVVIntrinsic {
+    let VLOperand = 5;
+  }
   // For unary operations with scalar type output without mask
   // Output: (scalar type)
   // Input: (vector_in, vl)
@@ -1312,6 +1334,10 @@ let TargetPrefix = "riscv" in {
     def "int_riscv_" # NAME : RISCVReductionUnMasked;
     def "int_riscv_" # NAME # "_mask" : RISCVReductionMasked;
   }
+  multiclass RISCVReductionRoundingMode {
+    def "int_riscv_" # NAME : RISCVReductionUnMaskedRoundingMode;
+    def "int_riscv_" # NAME # "_mask" : RISCVReductionMaskedRoundingMode;
+  }
   multiclass RISCVMaskedUnarySOut {
     def "int_riscv_" # NAME : RISCVMaskedUnarySOutUnMasked;
     def "int_riscv_" # NAME # "_mask" : RISCVMaskedUnarySOutMasked;
@@ -1592,13 +1618,13 @@ let TargetPrefix = "riscv" in {
   defm vwredsumu : RISCVReduction;
   defm vwredsum : RISCVReduction;
 
-  defm vfredosum : RISCVReduction;
-  defm vfredusum : RISCVReduction;
+  defm vfredosum : RISCVReductionRoundingMode;
+  defm vfredusum : RISCVReductionRoundingMode;
   defm vfredmin : RISCVReduction;
   defm vfredmax : RISCVReduction;
 
-  defm vfwredusum : RISCVReduction;
-  defm vfwredosum : RISCVReduction;
+  defm vfwredusum : RISCVReductionRoundingMode;
+  defm vfwredosum : RISCVReductionRoundingMode;
 
   def int_riscv_vmand: RISCVBinaryAAAUnMasked;
   def int_riscv_vmnand: RISCVBinaryAAAUnMasked;
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
index 8951d734cf86d..f03d266e46320 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
@@ -1290,6 +1290,28 @@ class VPseudoTernaryMaskPolicy<VReg RetClass,
   let HasVecPolicyOp = 1;
 }
 
+class VPseudoTernaryMaskPolicyRoundingMode<VReg RetClass,
+                                           RegisterClass Op1Class,
+                                           DAGOperand Op2Class,
+                                           string Constraint> :
+        Pseudo<(outs GetVRegNoV0<RetClass>.R:$rd),
+                (ins GetVRegNoV0<RetClass>.R:$merge,
+                     Op1Class:$rs2, Op2Class:$rs1,
+                     VMaskOp:$vm,
+                     ixlenimm:$rm,
+                     AVL:$vl, ixlenimm:$sew, ixlenimm:$policy), []>,
+        RISCVVPseudo {
+  let mayLoad = 0;
+  let mayStore = 0;
+  let hasSideEffects = 0;
+  let Constraints = !interleave([Constraint, "$rd = $merge"], ",");
+  let HasVLOp = 1;
+  let HasSEWOp = 1;
+  let HasVecPolicyOp = 1;
+  let HasRoundModeOp = 1;
+  let UsesVXRM = 0;
+}
+
 // Like VPseudoBinaryNoMask, but output can be V0.
 class VPseudoBinaryMOutNoMask<VReg RetClass,
                               VReg Op1Class,
@@ -3230,6 +3252,25 @@ multiclass VPseudoTernaryWithTailPolicy<VReg RetClass,
   }
 }
 
+multiclass VPseudoTernaryWithTailPolicyRoundingMode<VReg RetClass,
+                                          RegisterClass Op1Class,
+                                          DAGOperand Op2Class,
+                                          LMULInfo MInfo,
+                                          int sew,
+                                          string Constraint = "",
+                                          bit Commutable = 0> {
+  let VLMul = MInfo.value in {
+    defvar mx = MInfo.MX;
+    let isCommutable = Commutable in
+    def "_" # mx # "_E" # sew
+        : VPseudoTernaryNoMaskWithPolicyRoundingMode<RetClass, Op1Class, 
+                                                     Op2Class, Constraint>;
+    def "_" # mx # "_E" # sew # "_MASK"
+        : VPseudoTernaryMaskPolicyRoundingMode<RetClass, Op1Class,
+                                               Op2Class, Constraint>;
+  }
+}
+
 multiclass VPseudoTernaryWithPolicy<VReg RetClass,
                                     RegisterClass Op1Class,
                                     DAGOperand Op2Class,
@@ -3595,12 +3636,14 @@ multiclass VPseudoVWRED_VS {
   }
 }
 
-multiclass VPseudoVFRED_VS {
+multiclass VPseudoVFRED_VS_RM {
   foreach m = MxListF in {
     defvar mx = m.MX;
     foreach e = SchedSEWSet<mx, /*isF*/ 1>.val in {
       defvar WriteVFRedV_From_MX_E = !cast<SchedWrite>("WriteVFRedV_From_" # mx # "_E" # e);
-      defm _VS : VPseudoTernaryWithTailPolicy<V_M1.vrclass, m.vrclass, V_M1.vrclass, m, e>,
+      defm _VS
+          : VPseudoTernaryWithTailPolicyRoundingMode<V_M1.vrclass, m.vrclass, 
+                                                     V_M1.vrclass, m, e>,
                  Sched<[WriteVFRedV_From_MX_E, ReadVFRedV, ReadVFRedV, ReadVFRedV,
                         ReadVMask]>;
     }
@@ -3619,24 +3662,27 @@ multiclass VPseudoVFREDMINMAX_VS {
   }
 }
 
-multiclass VPseudoVFREDO_VS {
+multiclass VPseudoVFREDO_VS_RM {
   foreach m = MxListF in {
     defvar mx = m.MX;
     foreach e = SchedSEWSet<mx, /*isF*/ 1>.val in {
       defvar WriteVFRedOV_From_MX_E = !cast<SchedWrite>("WriteVFRedOV_From_" # mx # "_E" # e);
-      defm _VS : VPseudoTernaryWithTailPolicy<V_M1.vrclass, m.vrclass, V_M1.vrclass, m, e>,
+      defm _VS : VPseudoTernaryWithTailPolicyRoundingMode<V_M1.vrclass, m.vrclass,
+                                                          V_M1.vrclass, m, e>,
                  Sched<[WriteVFRedOV_From_MX_E, ReadVFRedOV, ReadVFRedOV,
                         ReadVFRedOV, ReadVMask]>;
     }
   }
 }
 
-multiclass VPseudoVFWRED_VS {
+multiclass VPseudoVFWRED_VS_RM {
   foreach m = MxListFWRed in {
     defvar mx = m.MX;
     foreach e = SchedSEWSet<mx, /*isF*/ 1, /*isWidening*/ 1>.val in {
       defvar WriteVFWRedV_From_MX_E = !cast<SchedWrite>("WriteVFWRedV_From_" # mx # "_E" # e);
-      defm _VS : VPseudoTernaryWithTailPolicy<V_M1.vrclass, m.vrclass, V_M1.vrclass, m, e>,
+      defm _VS
+          : VPseudoTernaryWithTailPolicyRoundingMode<V_M1.vrclass, m.vrclass,
+                                                     V_M1.vrclass, m, e>,
                  Sched<[WriteVFWRedV_From_MX_E, ReadVFWRedV, ReadVFWRedV,
                         ReadVFWRedV, ReadVMask]>;
     }
@@ -4553,6 +4599,30 @@ class VPatTernaryNoMaskTA<string intrinsic,
                     op2_kind:$rs2,
                     GPR:$vl, log2sew, TAIL_AGNOSTIC)>;
 
+class VPatTernaryNoMaskTARoundingMode<string intrinsic,
+                          string inst,
+                          string kind,
+                          ValueType result_type,
+                          ValueType op1_type,
+                          ValueType op2_type,
+                          int log2sew,
+                          LMULInfo vlmul,
+                          VReg result_reg_class,
+                          RegisterClass op1_reg_class,
+                          DAGOperand op2_kind> :
+  Pat<(result_type (!cast<Intrinsic>(intrinsic)
+                    (result_type result_reg_class:$rs3),
+                    (op1_type op1_reg_class:$rs1),
+                    (op2_type op2_kind:$rs2),
+                    (XLenVT timm:$round),
+                    VLOpFrag)),
+                   (!cast<Instruction>(inst#"_"#kind#"_"#vlmul.MX#"_E"#!shl(1, log2sew))
+                    result_reg_class:$rs3,
+                    (op1_type op1_reg_class:$rs1),
+                    op2_kind:$rs2,
+                    (XLenVT timm:$round),
+                    GPR:$vl, log2sew, TAIL_AGNOSTIC)>;
+
 class VPatTernaryNoMaskWithPolicy<string intrinsic,
                                   string inst,
                                   string kind,
@@ -4701,6 +4771,33 @@ class VPatTernaryMaskTA<string intrinsic,
                     (mask_type V0),
                     GPR:$vl, log2sew, TAIL_AGNOSTIC)>;
 
+class VPatTernaryMaskTARoundingMode<string intrinsic,
+                                    string inst,
+                                    string kind,
+                                    ValueType result_type,
+                                    ValueType op1_type,
+                                    ValueType op2_type,
+                                    ValueType mask_type,
+                                    int log2sew,
+                                    LMULInfo vlmul,
+                                    VReg result_reg_class,
+                                    RegisterClass op1_reg_class,
+                                    DAGOperand op2_kind> :
+  Pat<(result_type (!cast<Intrinsic>(intrinsic#"_mask")
+                    (result_type result_reg_class:$rs3),
+                    (op1_type op1_reg_class:$rs1),
+                    (op2_type op2_kind:$rs2),
+                    (mask_type V0),
+                    (XLenVT timm:$round),
+                    VLOpFrag)),
+                   (!cast<Instruction>(inst#"_"#kind#"_"#vlmul.MX#"_E"#!shl(1, log2sew)# "_MASK")
+                    result_reg_class:$rs3,
+                    (op1_type op1_reg_class:$rs1),
+                    op2_kind:$rs2,
+                    (mask_type V0),
+                    (XLenVT timm:$round),
+                    GPR:$vl, log2sew, TAIL_AGNOSTIC)>;
+
 multiclass VPatUnaryS_M<string intrinsic_name,
                              string inst> {
   foreach mti = AllMasks in {
@@ -5674,6 +5771,26 @@ multiclass VPatTernaryTA<string intrinsic,
                           result_reg_class, op1_reg_class, op2_kind>;
 }
 
+multiclass VPatTernaryTARoundingMode<string intrinsic,
+                                     string inst,
+                                     string kind,
+                                     ValueType result_type,
+                                     ValueType op1_type,
+                                     ValueType op2_type,
+                                     ValueType mask_type,
+                                     int log2sew,
+                                     LMULInfo vlmul,
+                                     VReg result_reg_class,
+                                     RegisterClass op1_reg_class,
+                                     DAGOperand op2_kind> {
+  def : VPatTernaryNoMaskTARoundingMode<intrinsic, inst, kind, result_type, op1_type,
+                            op2_type, log2sew, vlmul, result_reg_class,
+                            op1_reg_class, op2_kind>;
+  def : VPatTernaryMaskTARoundingMode<intrinsic, inst, kind, result_type, op1_type,
+                          op2_type, mask_type, log2sew, vlmul,
+                          result_reg_class, op1_reg_class, op2_kind>;
+}
+
 multiclass VPatTernaryV_VV_AAXA<string intrinsic, string instruction,
                                 list<VTypeInfo> vtilist> {
   foreach vti = vtilist in
@@ -5863,6 +5980,26 @@ multiclass VPatReductionV_VS<string intrinsic, string instruction, bit IsFloat =
   }
 }
 
+multiclass VPatReductionV_VS_RM<string intrinsic, string instruction, bit IsFloat = 0> {
+  foreach vti = !if(IsFloat, NoGroupFloatVectors, NoGroupIntegerVectors) in {
+    defvar vectorM1 = !cast<VTypeInfo>(!if(IsFloat, "VF", "VI") # vti.SEW # "M1");
+    let Predicates = GetVTypePredicates<vti>.Predicates in
+    defm : VPatTernaryTARoundingMode<intrinsic, instruction, "VS",
+                                     vectorM1.Vector, vti.Vector,
+                                     vectorM1.Vector, vti.Mask,
+                                     vti.Log2SEW, vti.LMul,
+                                     VR, vti.RegClass, VR>;
+  }
+  foreach gvti = !if(IsFloat, GroupFloatVectors, GroupIntegerVectors) in {
+    let Predicates = GetVTypePredicates<gvti>.Predicates in
+    defm : VPatTernaryTARoundingMode<intrinsic, instruction, "VS",
+                                     gvti.VectorM1, gvti.Vector,
+                                     gvti.VectorM1, gvti.Mask,
+                                     gvti.Log2SEW, gvti.LMul,
+                                     VR, gvti.RegClass, VR>;
+  }
+}
+
 multiclass VPatReductionW_VS<string intrinsic, string instruction, bit IsFloat = 0> {
   foreach vti = !if(IsFloat, AllFloatVectors, AllIntegerVectors) in {
     defvar wtiSEW = !mul(vti.SEW, 2);
@@ -5879,6 +6016,22 @@ multiclass VPatReductionW_VS<string intrinsic, string instruction, bit IsFloat =
   }
 }
 
+multiclass VPatReductionW_VS_RM<string intrinsic, string instruction, bit IsFloat = 0> {
+  foreach vti = !if(IsFloat, AllFloatVectors, AllIntegerVectors) in {
+    defvar wtiSEW = !mul(vti.SEW, 2);
+    if !le(wtiSEW, 64) then {
+      defvar wtiM1 = !cast<VTypeInfo>(!if(IsFloat, "VF", "VI") # wtiSEW # "M1");
+      let Predicates = GetVTypePredicates<vti>.Predicates in
+      defm : VPatTernaryTARoundingMode<intrinsic, instruction, "VS",
+                                       wtiM1.Vector, vti.Vector,
+                                       wtiM1.Vector, vti.Mask,
+                                       vti.Log2SEW, vti.LMul,
+                                       wtiM1.RegClass, vti.RegClass,
+                                       wtiM1.RegClass>;
+    }
+  }
+}
+
 multiclass VPatConversionVI_VF<string intrinsic,
                                string instruction> {
   foreach fvti = AllFloatVectors in {
@@ -6637,9 +6790,10 @@ let Predicates = [HasVInstructionsAnyF] in {
 //===----------------------------------------------------------------------===//
 // 14.3. Vector Single-Width Floating-Point Reduction Instructions
 //===----------------------------------------------------------------------===//
-let Uses = [FRM], mayRaiseFPException = true in {
-defm PseudoVFREDOSUM : VPseudoVFREDO_VS;
-defm PseudoVFREDUSUM : VPseudoVFRED_VS;
+let mayRaiseFPException = true,
+    hasSideEffects = 0 in {
+defm PseudoVFREDOSUM : VPseudoVFREDO_VS_RM;
+defm PseudoVFREDUSUM : VPseudoVFRED_VS_RM;
 }
 let mayRaiseFPException = true in {
 defm PseudoVFREDMIN  : VPseudoVFREDMINMAX_VS;
@@ -6650,10 +6804,10 @@ defm PseudoVFREDMAX  : VPseudoVFREDMINMAX_VS;
 // 14.4. Vector Widening Floating-Point Reduction Instructions
 //===----------------------------------------------------------------------===//
 let IsRVVWideningReduction = 1,
-    Uses = [FRM],
+    hasSideEffects = 0,
     mayRaiseFPException = true in {
-defm PseudoVFWREDUSUM  : VPseudoVFWRED_VS;
-defm PseudoVFWREDOSUM  : VPseudoVFWRED_VS;
+defm PseudoVFWREDUSUM  : VPseudoVFWRED_VS_RM;
+defm PseudoVFWREDOSUM  : VPseudoVFWRED_VS_RM;
 }
 
 } // Predicates = [HasVInstructionsAnyF]
@@ -7278,16 +7432,16 @@ defm : VPatReductionW_VS<"int_riscv_vwredsum", "PseudoVWREDSUM">;
 //===----------------------------------------------------------------------===//
 // 14.3. Vector Single-Width Floating-Point Reduction Instructions
 //===----------------------------------------------------------------------===//
-defm : VPatReductionV_VS<"int_riscv_vfredosum", "PseudoVFREDOSUM", /*IsFloat=*/1>;
-defm : VPatReductionV_VS<"int_riscv_vfredusum", "PseudoVFREDUSUM", /*IsFloat=*/1>;
+defm : VPatReductionV_VS_RM<"int_riscv_vfredosum", "PseudoVFREDOSUM", /*IsFloat=*/1>;
+defm : VPatReductionV_VS_RM<"int_riscv_vfredusum", "PseudoVFREDUSUM", /*IsFloat=*/1>;
 defm : VPatReductionV_VS<"int_riscv_vfredmin", "PseudoVFREDMIN", /*IsFloat=*/1>;
 defm : VPatReductionV_VS<"int_riscv_vfredmax", "PseudoVFREDMAX", /*IsFloat=*/1>;
 
 //===----------------------------------------------------------------------===//
 // 14.4. Vector Widening Floating-Point Reduction Instructions
 //===----------------------------------------------------------------------===//
-defm : VPatReductionW_VS<"int_riscv_vfwredusum", "PseudoVFWREDUSUM", /*IsFloat=*/1>;
-defm : VPatReductionW_VS<"int_riscv_vfwredosum", "PseudoVFWREDOSUM", /*IsFloat=*/1>;
+defm : VPatReductionW_VS_RM<"int_riscv_vfwredusum", "PseudoVFWREDUSUM", /*IsFloat=*/1>;
+defm : VPatReductionW_VS_RM<"int_riscv_vfwredosum", "PseudoVFWREDOSUM", /*IsFloat=*/1>;
 
 //===----------------------------------------------------------------------===//
 // 15. Vector Mask Instructions
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
index 06925ad9f3fa8..ab38b1b65f6f8 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
@@ -1381,6 +1381,39 @@ multiclass VPatReductionVL<SDNode vop, string instruction_name, bit is_float> {
   }
 }
 
+multiclass VPatReductionVL_RM<SDNode vop, string instruction_name, bit is_float> {
+  foreach vti = !if(is_float, AllFloatVectors, AllIntegerVectors) in {
+    defvar vti_m1 = !cast<VTypeInfo>(!if(is_float, "VF", "VI") # vti.SEW # "M1");
+    let Predicates = GetVTypePredicates<vti>.Predicates in {
+      def: Pat<(vti_m1.Vector (vop (vti_m1.Vector VR:$merge),
+                                   (vti.Vector vti.RegClass:$rs1), VR:$rs2,
+                                   (vti.Mask true_mask), VLOpFrag,
+                                   (XLenVT timm:$policy))),
+          (!cast<Instruction>(instruction_name#"_VS_"#vti.LMul.MX#"_E"#vti.SEW)
+              (vti_m1.Vector VR:$merge),
+              (vti.Vector vti.RegClass:$rs1),
+              (vti_m1.Vector VR:$rs2),
+              // Value to indicate no rounding mode change in
+              // RISCVInsertReadWriteCSR
+              FRM_DYN,
+              GPR:$vl, vti.Log2SEW, (XLenVT timm:$policy))>;
+
+      def: Pat<(vti_m1.Vector (vop (vti_m1.Vector VR:$merge),
+                                   (vti.Vector vti.RegClass:$rs1), VR:$rs2,
+                                   (vti.Mask V0), VLOpFrag,
+                                   (XLenVT timm:$policy))),
+          (!cast<Instruction>(instruction_name#"_VS_"#vti.LMul.MX#"_E"#vti.SEW#"_MASK")
+              (vti_m1.Vector VR:$merge),
+              (vti.Vector vti.RegClass:$rs1),
+              (vti_m1.Vector VR:$rs2),
+              (vti.Mask V0),
+              // Value to indicate no rounding mode change in
+              // RISCVInsertReadWriteCSR
+              FRM_DYN,
+              GPR:$vl, vti.Log2SEW, (XLenVT timm:$policy))>;
+    }
+  }
+}
 
 multiclass VPatBinaryExtVL_WV_WX<SDNode op, PatFrags extop, string instruction_name> {
   foreach vtiToWti = AllWidenableIntVectors in {
@@ -1459,6 +1492,41 @@ multiclass VPatWidenReductionVL<SDNode vop, PatFrags extop, string instruction_n
   }
 }
 
+multiclass VPatWidenReductionVL_RM<SDNode vop, PatFrags extop, string instruction_name, bit is_float> {
+  foreach vtiToWti = !if(is_float, AllWidenableFloatVectors, AllWidenableIntVectors) in {
+    defvar vti = vtiToWti.Vti;
+    defvar wti = vtiToWti.Wti;
+    defvar wti_m1 = !cast<VTypeInfo>(!if(is_float, "VF", "VI") # wti.SEW # "M1");
+    let Predicates = !listconcat(GetVTypePredicates<vti>.Predicates,
+                                 GetVTypePredicates<wti>.Predicates) in {
+      def: Pat<(wti_m1.Vector (vop (wti_m1.Vector VR:$merge),
+                                   (wti.Vector (extop (vti.Vector vti.RegClass:$rs1))),
+                                   VR:$rs2, (vti.Mask true_mask), VLOpFrag,
+                                   (XLenVT timm:$policy))),
+               (!cast<Instruction>(instruction_name#"_VS_"#vti.LMul.MX#"_E"#vti.SEW)
+                  (wti_m1.Vector VR:$merge), (vti.Vector vti.RegClass:$rs1),
+                  (wti_m1.Vector VR:$rs2),
+                  // Value to indicate no rounding mode change in
+                  // RISCVInsertReadWriteCSR
+                  FRM_DYN,
+                  GPR:$vl, vti.Log2SEW,
+                  (XLenVT timm:$policy))>;
+      def: Pat<(wti_m1.Vector (vop (wti_m1.Vector VR:$merge),
+                                   (wti.Vector (extop (vti.Vector vti.RegClass:$rs1))),
+                                   VR:$rs2, (vti.Mask V0), VLOpFrag,
+                                   (XLenVT timm:$policy))),
+               (!cast<Instruction>(instruction_name#"_VS_"#vti.LMul.MX#"_E"#vti.SEW#"_MASK")
+                  (wti_m1.Vector VR:$merge), (vti.Vector vti.RegClass:$rs1),
+                  (wti_m1.Vector VR:$rs2), (vti.Mask V0),
+                  // Value to indicate no rounding mode change in
+                  // RISCVInsertReadWriteCSR
+                  FRM_DYN,
+                  GPR:$vl, vti.Log2SEW,
+                  (XLenVT timm:$policy))>;
+    }
+  }
+}
+
 multiclass VPatWidenReductionVL_Ext_VL<SDNode vop, PatFrags extop, string instruction_name, bit is_float> {
   foreach vtiToWti = !if(is_float, AllWidenableFloatVectors, AllWidenableIntVectors) in {
     defvar vti = vtiToWti.Vti;
@@ -1486,6 +1554,41 @@ multiclass VPatWidenReductionVL_Ext_VL<SDNode vop, PatFrags extop, string instru
   }
 }
 
+multiclass VPatWidenReductionVL_Ext_VL_RM<SDNode vop, PatFrags extop, string instruction_name, bit is_float> {
+  foreach vtiToWti = !if(is_float, AllWidenableFloatVectors, AllWidenableIntVectors) in {
+    defvar vti = vtiToWti.Vti;
+    defvar wti = vtiToWti.Wti;
+    defvar wti_m1 = !cast<VTypeInfo>(!if(is_float, "VF", "VI") # wti.SEW # "M1");
+    let Predicates = !listconcat(GetVTypePredicates<vti>.Predicates,
+                                 GetVTypePredicates<wti>.Predicates) in {
+      def: Pat<(wti_m1.Vector (vop (wti_m1.Vector VR:$merge),
+                                   (wti.Vector (extop (vti.Vector vti.RegClass:$rs1), (vti.Mask true_mask), VLOpFrag)),
+                                   VR:$rs2, (vti.Mask true_mask), VLOpFrag,
+                                   (XLenVT timm:$policy))),
+               (!cast<Instruction>(instruction_name#"_VS_"#vti.LMul.MX#"_E"#vti.SEW)
+                  (wti_m1.Vector VR:$merge), (vti.Vector vti.RegClass:$rs1),
+                  (wti_m1.Vector VR:$rs2),
+                  // Value to indicate no rounding mode change in
+                  // RISCVInsertReadWriteCSR
+                  FRM_DYN,
+                  GPR:$vl, vti.Log2SEW,
+                  (XLenVT timm:$policy))>;
+      def: Pat<(wti_m1.Vector (vop (wti_m1.Vector VR:$merge),
+                                   (wti.Vector (extop (vti.Vector vti.RegClass:$rs1), (vti.Mask true_mask), VLOpFrag)),
+                                   VR:$rs2, (vti.Mask V0), VLOpFrag,
+                                   (XLenVT timm:$policy))),
+               (!cast<Instruction>(instruction_name#"_VS_"#vti.LMul.MX#"_E"#vti.SEW#"_MASK")
+                  (wti_m1.Vector VR:$merge), (vti.Vector vti.RegClass:$rs1),
+                  (wti_m1.Vector VR:$rs2), (vti.Mask V0),
+                  // Value to indicate no rounding mode change in
+                  // RISCVInsertReadWriteCSR
+                  FRM_DYN,
+                  GPR:$vl, vti.Log2SEW,
+                  (XLenVT timm:$policy))>;
+    }
+  }
+}
+
 multiclass VPatBinaryFPWVL_VV_VF<SDNode vop, string instruction_name> {
   foreach fvtiToFWti = AllWidenableFloatVectors in {
     defvar vti = fvtiToFWti.Vti;
@@ -2518,16 +2621,22 @@ defm : VPatWidenReductionVL<rvv_vecreduce_ADD_vl, sext_oneuse, "PseudoVWREDSUM",
 defm : VPatWidenReductionVL_Ext_VL<rvv_vecreduce_ADD_vl, riscv_sext_vl_oneuse, "PseudoVWREDSUM", /*is_float*/0>;
 
 // 14.3. Vector Single-Width Floating-Point Reduction Instructions
-defm : VPatReductionVL<rvv_vecreduce_SEQ_FADD_vl, "PseudoVFREDOSUM", /*is_float*/1>;
-defm : VPatReductionVL<rvv_vecreduce_FADD_vl,     "PseudoVFREDUSUM", /*is_float*/1>;
+defm : VPatReductionVL_RM<rvv_vecreduce_SEQ_FADD_vl, "PseudoVFREDOSUM", /*is_float*/1>;
+defm : VPatReductionVL_RM<rvv_vecreduce_FADD_vl,     "PseudoVFREDUSUM", /*is_float*/1>;
 defm : VPatReductionVL<rvv_vecreduce_FMIN_vl,     "PseudoVFREDMIN", /*is_float*/1>;
 defm : VPatReductionVL<rvv_vecreduce_FMAX_vl,     "PseudoVFREDMAX", /*is_float*/1>;
 
 // 14.4. Vector Widening Floating-Point Reduction Instructions
-defm : VPatWidenReductionVL<rvv_vecreduce_SEQ_FADD_vl, fpext_oneuse, "PseudoVFWREDOSUM", /*is_float*/1>;
-defm : VPatWidenReductionVL_Ext_VL<rvv_vecreduce_SEQ_FADD_vl, riscv_fpextend_vl_oneuse, "PseudoVFWREDOSUM", /*is_float*/1>;
-defm : VPatWidenReductionVL<rvv_vecreduce_FADD_vl, fpext_oneuse, "PseudoVFWREDUSUM", /*is_float*/1>;
-defm : VPatWidenReductionVL_Ext_VL<rvv_vecreduce_FADD_vl, riscv_fpextend_vl_oneuse, "PseudoVFWREDUSUM", /*is_float*/1>;
+defm : VPatWidenReductionVL_RM<rvv_vecreduce_SEQ_FADD_vl, fpext_oneuse,
+                               "PseudoVFWREDOSUM", /*is_float*/1>;
+defm : VPatWidenReductionVL_Ext_VL_RM<rvv_vecreduce_SEQ_FADD_vl,
+                                      riscv_fpextend_vl_oneuse,
+                                      "PseudoVFWREDOSUM", /*is_float*/1>;
+defm : VPatWidenReductionVL_RM<rvv_vecreduce_FADD_vl, fpext_oneuse,
+                               "PseudoVFWREDUSUM", /*is_float*/1>;
+defm : VPatWidenReductionVL_Ext_VL_RM<rvv_vecreduce_FADD_vl,
+                                      riscv_fpextend_vl_oneuse,
+                                      "PseudoVFWREDUSUM", /*is_float*/1>;
 
 // 15. Vector Mask Instructions
 
diff --git a/llvm/test/CodeGen/RISCV/rvv/unmasked-ta.ll b/llvm/test/CodeGen/RISCV/rvv/unmasked-ta.ll
index 75ee82a2731b8..941653848183b 100644
--- a/llvm/test/CodeGen/RISCV/rvv/unmasked-ta.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/unmasked-ta.ll
@@ -724,7 +724,7 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16(
   <vscale x 4 x half>,
   <vscale x 1 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_vs_nxv4f16_nxv1f16_nxv4f16(<vscale x 1 x half> %0, <vscale x 4 x half> %1, iXLen %2) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv4f16_nxv1f16_nxv4f16:
@@ -737,7 +737,7 @@ entry:
     <vscale x 4 x half> undef,
     <vscale x 1 x half> %0,
     <vscale x 4 x half> %1,
-    iXLen %2)
+    iXLen 7, iXLen %2)
 
   ret <vscale x 4 x half> %a
 }
@@ -746,7 +746,7 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16(
   <vscale x 4 x half>,
   <vscale x 1 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_vs_nxv4f16_nxv1f16_nxv4f16(<vscale x 1 x half> %0, <vscale x 4 x half> %1, iXLen %2) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv4f16_nxv1f16_nxv4f16:
@@ -759,7 +759,7 @@ entry:
     <vscale x 4 x half> undef,
     <vscale x 1 x half> %0,
     <vscale x 4 x half> %1,
-    iXLen %2)
+    iXLen 7, iXLen %2)
 
   ret <vscale x 4 x half> %a
 }
@@ -812,7 +812,7 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16(
   <vscale x 2 x float>,
   <vscale x 1 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_vs_nxv2f32_nxv1f16_nxv2f32(<vscale x 1 x half> %0, <vscale x 2 x float> %1, iXLen %2) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv2f32_nxv1f16_nxv2f32:
@@ -825,7 +825,7 @@ entry:
     <vscale x 2 x float> undef,
     <vscale x 1 x half> %0,
     <vscale x 2 x float> %1,
-    iXLen %2)
+    iXLen 7, iXLen %2)
 
   ret <vscale x 2 x float> %a
 }
@@ -833,7 +833,7 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16(
   <vscale x 2 x float>,
   <vscale x 1 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_vs_nxv2f32_nxv1f16_nxv2f32(<vscale x 1 x half> %0, <vscale x 2 x float> %1, iXLen %2) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv2f32_nxv1f16_nxv2f32:
@@ -846,7 +846,7 @@ entry:
     <vscale x 2 x float> undef,
     <vscale x 1 x half> %0,
     <vscale x 2 x float> %1,
-    iXLen %2)
+    iXLen 7, iXLen %2)
 
   ret <vscale x 2 x float> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfredosum.ll b/llvm/test/CodeGen/RISCV/rvv/vfredosum.ll
index 6d49023f6d525..c8cb1d592f0b3 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfredosum.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfredosum.ll
@@ -7,20 +7,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16(
   <vscale x 4 x half>,
   <vscale x 1 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_vs_nxv4f16_nxv1f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 1 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv4f16_nxv1f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv1f16(
     <vscale x 4 x half> %0,
     <vscale x 1 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -30,13 +32,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.nxv1i1(
   <vscale x 1 x half>,
   <vscale x 4 x half>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_mask_vs_nxv4f16_nxv1f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 1 x half> %1, <vscale x 4 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv4f16_nxv1f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv1f16.nxv1i1(
@@ -44,7 +48,7 @@ entry:
     <vscale x 1 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -53,20 +57,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16(
   <vscale x 4 x half>,
   <vscale x 2 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_vs_nxv4f16_nxv2f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 2 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv4f16_nxv2f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv2f16(
     <vscale x 4 x half> %0,
     <vscale x 2 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -76,13 +82,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.nxv2i1(
   <vscale x 2 x half>,
   <vscale x 4 x half>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_mask_vs_nxv4f16_nxv2f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 2 x half> %1, <vscale x 4 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv4f16_nxv2f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv2f16.nxv2i1(
@@ -90,7 +98,7 @@ entry:
     <vscale x 2 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -99,20 +107,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16(
   <vscale x 4 x half>,
   <vscale x 4 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_vs_nxv4f16_nxv4f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv4f16_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv4f16(
     <vscale x 4 x half> %0,
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -122,13 +132,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.nxv4i1(
   <vscale x 4 x half>,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_mask_vs_nxv4f16_nxv4f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv4f16_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv4f16.nxv4i1(
@@ -136,7 +148,7 @@ entry:
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -145,20 +157,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16(
   <vscale x 4 x half>,
   <vscale x 8 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_vs_nxv4f16_nxv8f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 8 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv4f16_nxv8f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv8f16(
     <vscale x 4 x half> %0,
     <vscale x 8 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -168,13 +182,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.nxv8i1(
   <vscale x 8 x half>,
   <vscale x 4 x half>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_mask_vs_nxv4f16_nxv8f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 8 x half> %1, <vscale x 4 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv4f16_nxv8f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv8f16.nxv8i1(
@@ -182,7 +198,7 @@ entry:
     <vscale x 8 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -191,20 +207,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16(
   <vscale x 4 x half>,
   <vscale x 16 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_vs_nxv4f16_nxv16f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 16 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv4f16_nxv16f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv16f16(
     <vscale x 4 x half> %0,
     <vscale x 16 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -214,13 +232,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.nxv16i1(
   <vscale x 16 x half>,
   <vscale x 4 x half>,
   <vscale x 16 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_mask_vs_nxv4f16_nxv16f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 16 x half> %1, <vscale x 4 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv4f16_nxv16f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv16f16.nxv16i1(
@@ -228,7 +248,7 @@ entry:
     <vscale x 16 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -237,20 +257,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16(
   <vscale x 4 x half>,
   <vscale x 32 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_vs_nxv4f16_nxv32f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 32 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv4f16_nxv32f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.nxv4f16.nxv32f16(
     <vscale x 4 x half> %0,
     <vscale x 32 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -260,13 +282,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.nxv32i1(
   <vscale x 32 x half>,
   <vscale x 4 x half>,
   <vscale x 32 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredosum_mask_vs_nxv4f16_nxv32f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 32 x half> %1, <vscale x 4 x half> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv4f16_nxv32f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredosum.mask.nxv4f16.nxv32f16.nxv32i1(
@@ -274,7 +298,7 @@ entry:
     <vscale x 32 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 32 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -283,20 +307,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32(
   <vscale x 2 x float>,
   <vscale x 1 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_vs_nxv2f32_nxv1f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 1 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv2f32_nxv1f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv1f32(
     <vscale x 2 x float> %0,
     <vscale x 1 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -306,13 +332,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.nxv1i1(
   <vscale x 1 x float>,
   <vscale x 2 x float>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_mask_vs_nxv2f32_nxv1f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 1 x float> %1, <vscale x 2 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv2f32_nxv1f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv1f32.nxv1i1(
@@ -320,7 +348,7 @@ entry:
     <vscale x 1 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -329,20 +357,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32(
   <vscale x 2 x float>,
   <vscale x 2 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_vs_nxv2f32_nxv2f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv2f32_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv2f32(
     <vscale x 2 x float> %0,
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -352,13 +382,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.nxv2i1(
   <vscale x 2 x float>,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_mask_vs_nxv2f32_nxv2f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv2f32_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv2f32.nxv2i1(
@@ -366,7 +398,7 @@ entry:
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -375,20 +407,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32(
   <vscale x 2 x float>,
   <vscale x 4 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_vs_nxv2f32_nxv4f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 4 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv2f32_nxv4f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv4f32(
     <vscale x 2 x float> %0,
     <vscale x 4 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -398,13 +432,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.nxv4i1(
   <vscale x 4 x float>,
   <vscale x 2 x float>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_mask_vs_nxv2f32_nxv4f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 4 x float> %1, <vscale x 2 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv2f32_nxv4f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv4f32.nxv4i1(
@@ -412,7 +448,7 @@ entry:
     <vscale x 4 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -421,20 +457,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32(
   <vscale x 2 x float>,
   <vscale x 8 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_vs_nxv2f32_nxv8f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 8 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv2f32_nxv8f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv8f32(
     <vscale x 2 x float> %0,
     <vscale x 8 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -444,13 +482,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.nxv8i1(
   <vscale x 8 x float>,
   <vscale x 2 x float>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_mask_vs_nxv2f32_nxv8f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 8 x float> %1, <vscale x 2 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv2f32_nxv8f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv8f32.nxv8i1(
@@ -458,7 +498,7 @@ entry:
     <vscale x 8 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -467,20 +507,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32(
   <vscale x 2 x float>,
   <vscale x 16 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_vs_nxv2f32_nxv16f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 16 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv2f32_nxv16f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.nxv2f32.nxv16f32(
     <vscale x 2 x float> %0,
     <vscale x 16 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -490,13 +532,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.nxv16i1
   <vscale x 16 x float>,
   <vscale x 2 x float>,
   <vscale x 16 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredosum_mask_vs_nxv2f32_nxv16f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 16 x float> %1, <vscale x 2 x float> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv2f32_nxv16f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredosum.mask.nxv2f32.nxv16f32.nxv16i1(
@@ -504,7 +548,7 @@ entry:
     <vscale x 16 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -513,20 +557,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64(
   <vscale x 1 x double>,
   <vscale x 1 x double>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredosum_vs_nxv1f64_nxv1f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 1 x double> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv1f64_nxv1f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv1f64(
     <vscale x 1 x double> %0,
     <vscale x 1 x double> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -536,13 +582,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.nxv1i1(
   <vscale x 1 x double>,
   <vscale x 1 x double>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredosum_mask_vs_nxv1f64_nxv1f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 1 x double> %1, <vscale x 1 x double> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv1f64_nxv1f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv1f64.nxv1i1(
@@ -550,7 +598,7 @@ entry:
     <vscale x 1 x double> %1,
     <vscale x 1 x double> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -559,20 +607,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64(
   <vscale x 1 x double>,
   <vscale x 2 x double>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredosum_vs_nxv1f64_nxv2f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 2 x double> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv1f64_nxv2f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv2f64(
     <vscale x 1 x double> %0,
     <vscale x 2 x double> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -582,13 +632,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.nxv2i1(
   <vscale x 2 x double>,
   <vscale x 1 x double>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredosum_mask_vs_nxv1f64_nxv2f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 2 x double> %1, <vscale x 1 x double> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv1f64_nxv2f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv2f64.nxv2i1(
@@ -596,7 +648,7 @@ entry:
     <vscale x 2 x double> %1,
     <vscale x 1 x double> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -605,20 +657,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64(
   <vscale x 1 x double>,
   <vscale x 4 x double>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredosum_vs_nxv1f64_nxv4f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 4 x double> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv1f64_nxv4f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv4f64(
     <vscale x 1 x double> %0,
     <vscale x 4 x double> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -628,13 +682,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.nxv4i1(
   <vscale x 4 x double>,
   <vscale x 1 x double>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredosum_mask_vs_nxv1f64_nxv4f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 4 x double> %1, <vscale x 1 x double> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv1f64_nxv4f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv4f64.nxv4i1(
@@ -642,7 +698,7 @@ entry:
     <vscale x 4 x double> %1,
     <vscale x 1 x double> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -651,20 +707,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64(
   <vscale x 1 x double>,
   <vscale x 8 x double>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredosum_vs_nxv1f64_nxv8f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 8 x double> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_vs_nxv1f64_nxv8f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredosum.nxv1f64.nxv8f64(
     <vscale x 1 x double> %0,
     <vscale x 8 x double> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -674,13 +732,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.nxv8i1(
   <vscale x 8 x double>,
   <vscale x 1 x double>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredosum_mask_vs_nxv1f64_nxv8f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 8 x double> %1, <vscale x 1 x double> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredosum_mask_vs_nxv1f64_nxv8f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredosum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredosum.mask.nxv1f64.nxv8f64.nxv8i1(
@@ -688,7 +748,7 @@ entry:
     <vscale x 8 x double> %1,
     <vscale x 1 x double> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfredusum.ll b/llvm/test/CodeGen/RISCV/rvv/vfredusum.ll
index f72aed34226ee..c0514e2664049 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfredusum.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfredusum.ll
@@ -7,20 +7,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16(
   <vscale x 4 x half>,
   <vscale x 1 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_vs_nxv4f16_nxv1f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 1 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv4f16_nxv1f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv1f16(
     <vscale x 4 x half> %0,
     <vscale x 1 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -30,13 +32,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.nxv1i1(
   <vscale x 1 x half>,
   <vscale x 4 x half>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_mask_vs_nxv4f16_nxv1f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 1 x half> %1, <vscale x 4 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv4f16_nxv1f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv1f16.nxv1i1(
@@ -44,7 +48,7 @@ entry:
     <vscale x 1 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -53,20 +57,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16(
   <vscale x 4 x half>,
   <vscale x 2 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_vs_nxv4f16_nxv2f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 2 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv4f16_nxv2f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv2f16(
     <vscale x 4 x half> %0,
     <vscale x 2 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -76,13 +82,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.nxv2i1(
   <vscale x 2 x half>,
   <vscale x 4 x half>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_mask_vs_nxv4f16_nxv2f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 2 x half> %1, <vscale x 4 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv4f16_nxv2f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv2f16.nxv2i1(
@@ -90,7 +98,7 @@ entry:
     <vscale x 2 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -99,20 +107,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16(
   <vscale x 4 x half>,
   <vscale x 4 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_vs_nxv4f16_nxv4f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv4f16_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv4f16(
     <vscale x 4 x half> %0,
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -122,13 +132,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.nxv4i1(
   <vscale x 4 x half>,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_mask_vs_nxv4f16_nxv4f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv4f16_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv4f16.nxv4i1(
@@ -136,7 +148,7 @@ entry:
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -145,20 +157,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16(
   <vscale x 4 x half>,
   <vscale x 8 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_vs_nxv4f16_nxv8f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 8 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv4f16_nxv8f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv8f16(
     <vscale x 4 x half> %0,
     <vscale x 8 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -168,13 +182,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.nxv8i1(
   <vscale x 8 x half>,
   <vscale x 4 x half>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_mask_vs_nxv4f16_nxv8f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 8 x half> %1, <vscale x 4 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv4f16_nxv8f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv8f16.nxv8i1(
@@ -182,7 +198,7 @@ entry:
     <vscale x 8 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -191,20 +207,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16(
   <vscale x 4 x half>,
   <vscale x 16 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_vs_nxv4f16_nxv16f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 16 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv4f16_nxv16f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv16f16(
     <vscale x 4 x half> %0,
     <vscale x 16 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -214,13 +232,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.nxv16i1(
   <vscale x 16 x half>,
   <vscale x 4 x half>,
   <vscale x 16 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_mask_vs_nxv4f16_nxv16f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 16 x half> %1, <vscale x 4 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv4f16_nxv16f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv16f16.nxv16i1(
@@ -228,7 +248,7 @@ entry:
     <vscale x 16 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -237,20 +257,22 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16(
   <vscale x 4 x half>,
   <vscale x 32 x half>,
   <vscale x 4 x half>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_vs_nxv4f16_nxv32f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 32 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv4f16_nxv32f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.nxv4f16.nxv32f16(
     <vscale x 4 x half> %0,
     <vscale x 32 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 4 x half> %a
 }
@@ -260,13 +282,15 @@ declare <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.nxv32i1(
   <vscale x 32 x half>,
   <vscale x 4 x half>,
   <vscale x 32 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 4 x half> @intrinsic_vfredusum_mask_vs_nxv4f16_nxv32f16_nxv4f16(<vscale x 4 x half> %0, <vscale x 32 x half> %1, <vscale x 4 x half> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv4f16_nxv32f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x half> @llvm.riscv.vfredusum.mask.nxv4f16.nxv32f16.nxv32i1(
@@ -274,7 +298,7 @@ entry:
     <vscale x 32 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 32 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 4 x half> %a
 }
@@ -283,20 +307,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32(
   <vscale x 2 x float>,
   <vscale x 1 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_vs_nxv2f32_nxv1f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 1 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv2f32_nxv1f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv1f32(
     <vscale x 2 x float> %0,
     <vscale x 1 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -306,13 +332,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.nxv1i1(
   <vscale x 1 x float>,
   <vscale x 2 x float>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_mask_vs_nxv2f32_nxv1f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 1 x float> %1, <vscale x 2 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv2f32_nxv1f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv1f32.nxv1i1(
@@ -320,7 +348,7 @@ entry:
     <vscale x 1 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -329,20 +357,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32(
   <vscale x 2 x float>,
   <vscale x 2 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_vs_nxv2f32_nxv2f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv2f32_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv2f32(
     <vscale x 2 x float> %0,
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -352,13 +382,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.nxv2i1(
   <vscale x 2 x float>,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_mask_vs_nxv2f32_nxv2f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv2f32_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv2f32.nxv2i1(
@@ -366,7 +398,7 @@ entry:
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -375,20 +407,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32(
   <vscale x 2 x float>,
   <vscale x 4 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_vs_nxv2f32_nxv4f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 4 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv2f32_nxv4f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv4f32(
     <vscale x 2 x float> %0,
     <vscale x 4 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -398,13 +432,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.nxv4i1(
   <vscale x 4 x float>,
   <vscale x 2 x float>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_mask_vs_nxv2f32_nxv4f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 4 x float> %1, <vscale x 2 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv2f32_nxv4f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv4f32.nxv4i1(
@@ -412,7 +448,7 @@ entry:
     <vscale x 4 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -421,20 +457,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32(
   <vscale x 2 x float>,
   <vscale x 8 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_vs_nxv2f32_nxv8f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 8 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv2f32_nxv8f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32(
     <vscale x 2 x float> %0,
     <vscale x 8 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -444,13 +482,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.nxv8i1(
   <vscale x 8 x float>,
   <vscale x 2 x float>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_mask_vs_nxv2f32_nxv8f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 8 x float> %1, <vscale x 2 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv2f32_nxv8f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv8f32.nxv8i1(
@@ -458,7 +498,7 @@ entry:
     <vscale x 8 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -467,20 +507,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32(
   <vscale x 2 x float>,
   <vscale x 16 x float>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_vs_nxv2f32_nxv16f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 16 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv2f32_nxv16f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv16f32(
     <vscale x 2 x float> %0,
     <vscale x 16 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -490,13 +532,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.nxv16i1
   <vscale x 16 x float>,
   <vscale x 2 x float>,
   <vscale x 16 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfredusum_mask_vs_nxv2f32_nxv16f32_nxv2f32(<vscale x 2 x float> %0, <vscale x 16 x float> %1, <vscale x 2 x float> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv2f32_nxv16f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfredusum.mask.nxv2f32.nxv16f32.nxv16i1(
@@ -504,7 +548,7 @@ entry:
     <vscale x 16 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -513,20 +557,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64(
   <vscale x 1 x double>,
   <vscale x 1 x double>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredusum_vs_nxv1f64_nxv1f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 1 x double> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv1f64_nxv1f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv1f64(
     <vscale x 1 x double> %0,
     <vscale x 1 x double> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -536,13 +582,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.nxv1i1(
   <vscale x 1 x double>,
   <vscale x 1 x double>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredusum_mask_vs_nxv1f64_nxv1f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 1 x double> %1, <vscale x 1 x double> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv1f64_nxv1f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv1f64.nxv1i1(
@@ -550,7 +598,7 @@ entry:
     <vscale x 1 x double> %1,
     <vscale x 1 x double> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -559,20 +607,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64(
   <vscale x 1 x double>,
   <vscale x 2 x double>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredusum_vs_nxv1f64_nxv2f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 2 x double> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv1f64_nxv2f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv2f64(
     <vscale x 1 x double> %0,
     <vscale x 2 x double> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -582,13 +632,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.nxv2i1(
   <vscale x 2 x double>,
   <vscale x 1 x double>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredusum_mask_vs_nxv1f64_nxv2f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 2 x double> %1, <vscale x 1 x double> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv1f64_nxv2f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv2f64.nxv2i1(
@@ -596,7 +648,7 @@ entry:
     <vscale x 2 x double> %1,
     <vscale x 1 x double> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -605,20 +657,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64(
   <vscale x 1 x double>,
   <vscale x 4 x double>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredusum_vs_nxv1f64_nxv4f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 4 x double> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv1f64_nxv4f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv4f64(
     <vscale x 1 x double> %0,
     <vscale x 4 x double> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -628,13 +682,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.nxv4i1(
   <vscale x 4 x double>,
   <vscale x 1 x double>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredusum_mask_vs_nxv1f64_nxv4f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 4 x double> %1, <vscale x 1 x double> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv1f64_nxv4f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv4f64.nxv4i1(
@@ -642,7 +698,7 @@ entry:
     <vscale x 4 x double> %1,
     <vscale x 1 x double> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -651,20 +707,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64(
   <vscale x 1 x double>,
   <vscale x 8 x double>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredusum_vs_nxv1f64_nxv8f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 8 x double> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_vs_nxv1f64_nxv8f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredusum.nxv1f64.nxv8f64(
     <vscale x 1 x double> %0,
     <vscale x 8 x double> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -674,13 +732,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.nxv8i1(
   <vscale x 8 x double>,
   <vscale x 1 x double>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfredusum_mask_vs_nxv1f64_nxv8f64_nxv1f64(<vscale x 1 x double> %0, <vscale x 8 x double> %1, <vscale x 1 x double> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfredusum_mask_vs_nxv1f64_nxv8f64_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfredusum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfredusum.mask.nxv1f64.nxv8f64.nxv8i1(
@@ -688,7 +748,7 @@ entry:
     <vscale x 8 x double> %1,
     <vscale x 1 x double> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwredosum.ll b/llvm/test/CodeGen/RISCV/rvv/vfwredosum.ll
index c850c303592cd..d2396b20e10a9 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfwredosum.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwredosum.ll
@@ -7,20 +7,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16(
   <vscale x 2 x float>,
   <vscale x 1 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_vs_nxv2f32_nxv1f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 1 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv2f32_nxv1f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv1f16(
     <vscale x 2 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -30,13 +32,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.nxv2f32
   <vscale x 1 x half>,
   <vscale x 2 x float>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_mask_vs_nxv2f32_nxv1f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 1 x half> %1, <vscale x 2 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv2f32_nxv1f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv1f16.nxv2f32(
@@ -44,7 +48,7 @@ entry:
     <vscale x 1 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -53,20 +57,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16(
   <vscale x 2 x float>,
   <vscale x 2 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_vs_nxv2f32_nxv2f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv2f32_nxv2f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv2f16(
     <vscale x 2 x float> %0,
     <vscale x 2 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -76,13 +82,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.nxv2f32
   <vscale x 2 x half>,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_mask_vs_nxv2f32_nxv2f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv2f32_nxv2f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv2f16.nxv2f32(
@@ -90,7 +98,7 @@ entry:
     <vscale x 2 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -99,20 +107,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16(
   <vscale x 2 x float>,
   <vscale x 4 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_vs_nxv2f32_nxv4f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 4 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv2f32_nxv4f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv4f16(
     <vscale x 2 x float> %0,
     <vscale x 4 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -122,13 +132,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.nxv2f32
   <vscale x 4 x half>,
   <vscale x 2 x float>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_mask_vs_nxv2f32_nxv4f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 4 x half> %1, <vscale x 2 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv2f32_nxv4f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv4f16.nxv2f32(
@@ -136,7 +148,7 @@ entry:
     <vscale x 4 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -145,20 +157,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16(
   <vscale x 2 x float>,
   <vscale x 8 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_vs_nxv2f32_nxv8f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 8 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv2f32_nxv8f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv8f16(
     <vscale x 2 x float> %0,
     <vscale x 8 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -168,13 +182,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.nxv2f32
   <vscale x 8 x half>,
   <vscale x 2 x float>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_mask_vs_nxv2f32_nxv8f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 8 x half> %1, <vscale x 2 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv2f32_nxv8f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv8f16.nxv2f32(
@@ -182,7 +198,7 @@ entry:
     <vscale x 8 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -191,20 +207,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16(
   <vscale x 2 x float>,
   <vscale x 16 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_vs_nxv2f32_nxv16f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 16 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv2f32_nxv16f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv16f16(
     <vscale x 2 x float> %0,
     <vscale x 16 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -214,13 +232,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.nxv2f3
   <vscale x 16 x half>,
   <vscale x 2 x float>,
   <vscale x 16 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_mask_vs_nxv2f32_nxv16f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 16 x half> %1, <vscale x 2 x float> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv2f32_nxv16f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv16f16.nxv2f32(
@@ -228,7 +248,7 @@ entry:
     <vscale x 16 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -237,20 +257,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16(
   <vscale x 2 x float>,
   <vscale x 32 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_vs_nxv2f32_nxv32f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 32 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv2f32_nxv32f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.nxv2f32.nxv32f16(
     <vscale x 2 x float> %0,
     <vscale x 32 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -260,13 +282,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16(
   <vscale x 32 x half>,
   <vscale x 2 x float>,
   <vscale x 32 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredosum_mask_vs_nxv2f32_nxv32f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 32 x half> %1, <vscale x 2 x float> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv2f32_nxv32f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredosum.mask.nxv2f32.nxv32f16(
@@ -274,7 +298,7 @@ entry:
     <vscale x 32 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 32 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -283,20 +307,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32(
   <vscale x 1 x double>,
   <vscale x 1 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_vs_nxv1f64_nxv1f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv1f64_nxv1f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv1f32(
     <vscale x 1 x double> %0,
     <vscale x 1 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -306,13 +332,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.nxv1f6
   <vscale x 1 x float>,
   <vscale x 1 x double>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_mask_vs_nxv1f64_nxv1f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x double> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv1f64_nxv1f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv1f32.nxv1f64(
@@ -320,7 +348,7 @@ entry:
     <vscale x 1 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -329,20 +357,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32(
   <vscale x 1 x double>,
   <vscale x 2 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_vs_nxv1f64_nxv2f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 2 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv1f64_nxv2f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv2f32(
     <vscale x 1 x double> %0,
     <vscale x 2 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -352,13 +382,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.nxv1f6
   <vscale x 2 x float>,
   <vscale x 1 x double>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_mask_vs_nxv1f64_nxv2f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 2 x float> %1, <vscale x 1 x double> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv1f64_nxv2f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv2f32.nxv1f64(
@@ -366,7 +398,7 @@ entry:
     <vscale x 2 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -375,20 +407,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32(
   <vscale x 1 x double>,
   <vscale x 4 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_vs_nxv1f64_nxv4f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 4 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv1f64_nxv4f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv4f32(
     <vscale x 1 x double> %0,
     <vscale x 4 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -398,13 +432,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.nxv1f6
   <vscale x 4 x float>,
   <vscale x 1 x double>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_mask_vs_nxv1f64_nxv4f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 4 x float> %1, <vscale x 1 x double> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv1f64_nxv4f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv4f32.nxv1f64(
@@ -412,7 +448,7 @@ entry:
     <vscale x 4 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -421,20 +457,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32(
   <vscale x 1 x double>,
   <vscale x 8 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_vs_nxv1f64_nxv8f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 8 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv1f64_nxv8f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv8f32(
     <vscale x 1 x double> %0,
     <vscale x 8 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -444,13 +482,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.nxv1f6
   <vscale x 8 x float>,
   <vscale x 1 x double>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_mask_vs_nxv1f64_nxv8f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 8 x float> %1, <vscale x 1 x double> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv1f64_nxv8f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv8f32.nxv1f64(
@@ -458,7 +498,7 @@ entry:
     <vscale x 8 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -467,20 +507,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32(
   <vscale x 1 x double>,
   <vscale x 16 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_vs_nxv1f64_nxv16f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 16 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_vs_nxv1f64_nxv16f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.nxv1f64.nxv16f32(
     <vscale x 1 x double> %0,
     <vscale x 16 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -490,13 +532,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.nxv1f
   <vscale x 16 x float>,
   <vscale x 1 x double>,
   <vscale x 16 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredosum_mask_vs_nxv1f64_nxv16f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 16 x float> %1, <vscale x 1 x double> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredosum_mask_vs_nxv1f64_nxv16f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredosum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredosum.mask.nxv1f64.nxv16f32.nxv1f64(
@@ -504,7 +548,7 @@ entry:
     <vscale x 16 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwredusum.ll b/llvm/test/CodeGen/RISCV/rvv/vfwredusum.ll
index 77df494ccd402..53a39d441c806 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vfwredusum.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwredusum.ll
@@ -7,20 +7,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16(
   <vscale x 2 x float>,
   <vscale x 1 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_vs_nxv2f32_nxv1f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 1 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv2f32_nxv1f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv1f16(
     <vscale x 2 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -30,13 +32,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.nxv2f32
   <vscale x 1 x half>,
   <vscale x 2 x float>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_mask_vs_nxv2f32_nxv1f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 1 x half> %1, <vscale x 2 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv2f32_nxv1f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv1f16.nxv2f32(
@@ -44,7 +48,7 @@ entry:
     <vscale x 1 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -53,20 +57,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16(
   <vscale x 2 x float>,
   <vscale x 2 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_vs_nxv2f32_nxv2f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv2f32_nxv2f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv2f16(
     <vscale x 2 x float> %0,
     <vscale x 2 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -76,13 +82,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.nxv2f32
   <vscale x 2 x half>,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_mask_vs_nxv2f32_nxv2f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv2f32_nxv2f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv2f16.nxv2f32(
@@ -90,7 +98,7 @@ entry:
     <vscale x 2 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -99,20 +107,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16(
   <vscale x 2 x float>,
   <vscale x 4 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_vs_nxv2f32_nxv4f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 4 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv2f32_nxv4f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv4f16(
     <vscale x 2 x float> %0,
     <vscale x 4 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -122,13 +132,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.nxv2f32
   <vscale x 4 x half>,
   <vscale x 2 x float>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_mask_vs_nxv2f32_nxv4f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 4 x half> %1, <vscale x 2 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv2f32_nxv4f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv4f16.nxv2f32(
@@ -136,7 +148,7 @@ entry:
     <vscale x 4 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -145,20 +157,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16(
   <vscale x 2 x float>,
   <vscale x 8 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_vs_nxv2f32_nxv8f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 8 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv2f32_nxv8f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv8f16(
     <vscale x 2 x float> %0,
     <vscale x 8 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -168,13 +182,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.nxv2f32
   <vscale x 8 x half>,
   <vscale x 2 x float>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_mask_vs_nxv2f32_nxv8f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 8 x half> %1, <vscale x 2 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv2f32_nxv8f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv8f16.nxv2f32(
@@ -182,7 +198,7 @@ entry:
     <vscale x 8 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -191,20 +207,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16(
   <vscale x 2 x float>,
   <vscale x 16 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_vs_nxv2f32_nxv16f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 16 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv2f32_nxv16f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv16f16(
     <vscale x 2 x float> %0,
     <vscale x 16 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -214,13 +232,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.nxv2f3
   <vscale x 16 x half>,
   <vscale x 2 x float>,
   <vscale x 16 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_mask_vs_nxv2f32_nxv16f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 16 x half> %1, <vscale x 2 x float> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv2f32_nxv16f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv16f16.nxv2f32(
@@ -228,7 +248,7 @@ entry:
     <vscale x 16 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -237,20 +257,22 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16(
   <vscale x 2 x float>,
   <vscale x 32 x half>,
   <vscale x 2 x float>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_vs_nxv2f32_nxv32f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 32 x half> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv2f32_nxv32f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.nxv2f32.nxv32f16(
     <vscale x 2 x float> %0,
     <vscale x 32 x half> %1,
     <vscale x 2 x float> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 2 x float> %a
 }
@@ -260,13 +282,15 @@ declare <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16(
   <vscale x 32 x half>,
   <vscale x 2 x float>,
   <vscale x 32 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwredusum_mask_vs_nxv2f32_nxv32f16_nxv2f32(<vscale x 2 x float> %0, <vscale x 32 x half> %1, <vscale x 2 x float> %2, <vscale x 32 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv2f32_nxv32f16_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwredusum.mask.nxv2f32.nxv32f16(
@@ -274,7 +298,7 @@ entry:
     <vscale x 32 x half> %1,
     <vscale x 2 x float> %2,
     <vscale x 32 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 2 x float> %a
 }
@@ -283,20 +307,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32(
   <vscale x 1 x double>,
   <vscale x 1 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_vs_nxv1f64_nxv1f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv1f64_nxv1f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv1f32(
     <vscale x 1 x double> %0,
     <vscale x 1 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -306,13 +332,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.nxv1f6
   <vscale x 1 x float>,
   <vscale x 1 x double>,
   <vscale x 1 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_mask_vs_nxv1f64_nxv1f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x double> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv1f64_nxv1f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv1f32.nxv1f64(
@@ -320,7 +348,7 @@ entry:
     <vscale x 1 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -329,20 +357,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32(
   <vscale x 1 x double>,
   <vscale x 2 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_vs_nxv1f64_nxv2f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 2 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv1f64_nxv2f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv2f32(
     <vscale x 1 x double> %0,
     <vscale x 2 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -352,13 +382,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.nxv1f6
   <vscale x 2 x float>,
   <vscale x 1 x double>,
   <vscale x 2 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_mask_vs_nxv1f64_nxv2f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 2 x float> %1, <vscale x 1 x double> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv1f64_nxv2f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv2f32.nxv1f64(
@@ -366,7 +398,7 @@ entry:
     <vscale x 2 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -375,20 +407,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32(
   <vscale x 1 x double>,
   <vscale x 4 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_vs_nxv1f64_nxv4f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 4 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv1f64_nxv4f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v10, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv4f32(
     <vscale x 1 x double> %0,
     <vscale x 4 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -398,13 +432,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.nxv1f6
   <vscale x 4 x float>,
   <vscale x 1 x double>,
   <vscale x 4 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_mask_vs_nxv1f64_nxv4f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 4 x float> %1, <vscale x 1 x double> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv1f64_nxv4f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v10, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv4f32.nxv1f64(
@@ -412,7 +448,7 @@ entry:
     <vscale x 4 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -421,20 +457,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32(
   <vscale x 1 x double>,
   <vscale x 8 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_vs_nxv1f64_nxv8f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 8 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv1f64_nxv8f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v12, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv8f32(
     <vscale x 1 x double> %0,
     <vscale x 8 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -444,13 +482,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.nxv1f6
   <vscale x 8 x float>,
   <vscale x 1 x double>,
   <vscale x 8 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_mask_vs_nxv1f64_nxv8f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 8 x float> %1, <vscale x 1 x double> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv1f64_nxv8f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v12, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv8f32.nxv1f64(
@@ -458,7 +498,7 @@ entry:
     <vscale x 8 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }
@@ -467,20 +507,22 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32(
   <vscale x 1 x double>,
   <vscale x 16 x float>,
   <vscale x 1 x double>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_vs_nxv1f64_nxv16f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 16 x float> %1, <vscale x 1 x double> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_vs_nxv1f64_nxv16f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v16, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.nxv1f64.nxv16f32(
     <vscale x 1 x double> %0,
     <vscale x 16 x float> %1,
     <vscale x 1 x double> %2,
-    iXLen %3)
+    iXLen 0, iXLen %3)
 
   ret <vscale x 1 x double> %a
 }
@@ -490,13 +532,15 @@ declare <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.nxv1f
   <vscale x 16 x float>,
   <vscale x 1 x double>,
   <vscale x 16 x i1>,
-  iXLen);
+  iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwredusum_mask_vs_nxv1f64_nxv16f32_nxv1f64(<vscale x 1 x double> %0, <vscale x 16 x float> %1, <vscale x 1 x double> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwredusum_mask_vs_nxv1f64_nxv16f32_nxv1f64:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwredusum.vs v8, v16, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwredusum.mask.nxv1f64.nxv16f32.nxv1f64(
@@ -504,7 +548,7 @@ entry:
     <vscale x 16 x float> %1,
     <vscale x 1 x double> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4)
+    iXLen 0, iXLen %4)
 
   ret <vscale x 1 x double> %a
 }