[ConstantFolding] Generalize constant folding for vector_deinterleave2 to deinterleave3-8. #168640

topperc · 2025-11-19T00:39:14Z

No description provided.

…2 to deinterleave3-8.

llvmbot · 2025-11-19T00:39:47Z

@llvm/pr-subscribers-llvm-analysis

@llvm/pr-subscribers-llvm-transforms

Author: Craig Topper (topperc)

Changes

Patch is 29.22 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/168640.diff

2 Files Affected:

(modified) llvm/lib/Analysis/ConstantFolding.cpp (+31-14)
(modified) llvm/test/Transforms/InstSimplify/ConstProp/vector-calls.ll (+202-10)

diff --git a/llvm/lib/Analysis/ConstantFolding.cpp b/llvm/lib/Analysis/ConstantFolding.cpp
old mode 100755
new mode 100644
index a13df6c5bf552..c5eed99027cb0
--- a/llvm/lib/Analysis/ConstantFolding.cpp
+++ b/llvm/lib/Analysis/ConstantFolding.cpp
@@ -1666,6 +1666,12 @@ bool llvm::canConstantFoldCallTo(const CallBase *Call, const Function *F) {
   case Intrinsic::vector_interleave7:
   case Intrinsic::vector_interleave8:
   case Intrinsic::vector_deinterleave2:
+  case Intrinsic::vector_deinterleave3:
+  case Intrinsic::vector_deinterleave4:
+  case Intrinsic::vector_deinterleave5:
+  case Intrinsic::vector_deinterleave6:
+  case Intrinsic::vector_deinterleave7:
+  case Intrinsic::vector_deinterleave8:
   // Target intrinsics
   case Intrinsic::amdgcn_perm:
   case Intrinsic::amdgcn_wave_reduce_umin:
@@ -4425,31 +4431,42 @@ ConstantFoldStructCall(StringRef Name, Intrinsic::ID IntrinsicID,
       return nullptr;
     return ConstantStruct::get(StTy, SinResult, CosResult);
   }
-  case Intrinsic::vector_deinterleave2: {
+  case Intrinsic::vector_deinterleave2:
+  case Intrinsic::vector_deinterleave3:
+  case Intrinsic::vector_deinterleave4:
+  case Intrinsic::vector_deinterleave5:
+  case Intrinsic::vector_deinterleave6:
+  case Intrinsic::vector_deinterleave7:
+  case Intrinsic::vector_deinterleave8: {
+    unsigned NumResults = StTy->getNumElements();
     auto *Vec = Operands[0];
     auto *VecTy = cast<VectorType>(Vec->getType());
 
     if (auto *EltC = Vec->getSplatValue()) {
-      ElementCount HalfEC = VecTy->getElementCount().divideCoefficientBy(2);
+      ElementCount HalfEC =
+          VecTy->getElementCount().divideCoefficientBy(NumResults);
       auto *HalfVec = ConstantVector::getSplat(HalfEC, EltC);
-      return ConstantStruct::get(StTy, HalfVec, HalfVec);
+      SmallVector<Constant *, 8> Results(NumResults, HalfVec);
+      return ConstantStruct::get(StTy, Results);
     }
 
     if (!isa<FixedVectorType>(Vec->getType()))
       return nullptr;
 
-    unsigned NumElements = VecTy->getElementCount().getFixedValue() / 2;
-    SmallVector<Constant *, 4> Res0(NumElements), Res1(NumElements);
-    for (unsigned I = 0; I < NumElements; ++I) {
-      Constant *Elt0 = Vec->getAggregateElement(2 * I);
-      Constant *Elt1 = Vec->getAggregateElement(2 * I + 1);
-      if (!Elt0 || !Elt1)
-        return nullptr;
-      Res0[I] = Elt0;
-      Res1[I] = Elt1;
+    unsigned NumElements =
+        VecTy->getElementCount().getFixedValue() / NumResults;
+    SmallVector<Constant *, 8> Results(NumResults);
+    SmallVector<Constant *> Elements(NumElements);
+    for (unsigned I = 0; I != NumResults; ++I) {
+      for (unsigned J = 0; J != NumElements; ++J) {
+        Constant *Elt = Vec->getAggregateElement(J * NumResults + I);
+        if (!Elt)
+          return nullptr;
+        Elements[J] = Elt;
+      }
+      Results[I] = ConstantVector::get(Elements);
     }
-    return ConstantStruct::get(StTy, ConstantVector::get(Res0),
-                               ConstantVector::get(Res1));
+    return ConstantStruct::get(StTy, Results);
   }
   default:
     // TODO: Constant folding of vector intrinsics that fall through here does
diff --git a/llvm/test/Transforms/InstSimplify/ConstProp/vector-calls.ll b/llvm/test/Transforms/InstSimplify/ConstProp/vector-calls.ll
index 848f0d17ff373..4b6ee87c296f4 100644
--- a/llvm/test/Transforms/InstSimplify/ConstProp/vector-calls.ll
+++ b/llvm/test/Transforms/InstSimplify/ConstProp/vector-calls.ll
@@ -99,23 +99,23 @@ define <32 x i32> @fold_vector_interleave8() {
   ret <32 x i32> %1
 }
 
-define {<4 x i32>, <4 x i32>} @fold_vector_deinterleave2() {
+define { <4 x i32>, <4 x i32> } @fold_vector_deinterleave2() {
 ; CHECK-LABEL: define { <4 x i32>, <4 x i32> } @fold_vector_deinterleave2() {
 ; CHECK-NEXT:    ret { <4 x i32>, <4 x i32> } { <4 x i32> <i32 1, i32 2, i32 3, i32 4>, <4 x i32> <i32 5, i32 6, i32 7, i32 8> }
 ;
-  %1 = call {<4 x i32>, <4 x i32>} @llvm.vector.deinterleave2.v4i32.v8i32(<8 x i32> <i32 1, i32 5, i32 2, i32 6, i32 3, i32 7, i32 4, i32 8>)
-  ret {<4 x i32>, <4 x i32>} %1
+  %1 = call { <4 x i32>, <4 x i32> } @llvm.vector.deinterleave2.v4i32.v8i32(<8 x i32> <i32 1, i32 5, i32 2, i32 6, i32 3, i32 7, i32 4, i32 8>)
+  ret { <4 x i32>, <4 x i32> } %1
 }
 
-define {<vscale x 4 x i32>, <vscale x 4 x i32>} @fold_scalable_vector_deinterleave2() {
+define { <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave2() {
 ; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave2() {
 ; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } zeroinitializer
 ;
-  %1 = call {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.vector.deinterleave2.v4i32.v8i32(<vscale x 8 x i32> zeroinitializer)
-  ret {<vscale x 4 x i32>, <vscale x 4 x i32>} %1
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave2.v4i32.v8i32(<vscale x 8 x i32> zeroinitializer)
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32> } %1
 }
 
-define {<vscale x 4 x i32>, <vscale x 4 x i32>} @fold_scalable_vector_deinterleave2_splat() {
+define { <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave2_splat() {
 ; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave2_splat() {
 ; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32> } { <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1) }
 ;
@@ -123,10 +123,202 @@ define {<vscale x 4 x i32>, <vscale x 4 x i32>} @fold_scalable_vector_deinterlea
   ret {<vscale x 4 x i32>, <vscale x 4 x i32>} %1
 }
 
-define {<vscale x 4 x float>, <vscale x 4 x float>} @fold_scalable_vector_deinterleave2_splatfp() {
+define { <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave2_splatfp() {
 ; CHECK-LABEL: define { <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave2_splatfp() {
 ; CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float> } { <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00) }
 ;
-  %1 = call {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.vector.deinterleave2.v4f32.v8f32(<vscale x 8 x float> splat (float 1.0))
-  ret {<vscale x 4 x float>, <vscale x 4 x float>} %1
+  %1 = call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave2.v4f32.v8f32(<vscale x 8 x float> splat (float 1.0))
+  ret { <vscale x 4 x float>, <vscale x 4 x float> } %1
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave3() {
+; CHECK-LABEL: define { <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave3() {
+; CHECK-NEXT:    ret { <4 x i32>, <4 x i32>, <4 x i32> } { <4 x i32> <i32 1, i32 2, i32 3, i32 4>, <4 x i32> <i32 5, i32 6, i32 7, i32 8>, <4 x i32> <i32 9, i32 10, i32 11, i32 12> }
+;
+  %1 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.vector.deinterleave3.v4i32.v12i32(<12 x i32> <i32 1, i32 5, i32 9, i32 2, i32 6, i32 10, i32 3, i32 7, i32 11, i32 4, i32 8, i32 12>)
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %1
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave3() {
+; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave3() {
+; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } zeroinitializer
+;
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave3.v4i32.v12i32(<vscale x 12 x i32> zeroinitializer)
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %1
+}
+
+define {<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>} @fold_scalable_vector_deinterleave3_splat() {
+; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave3_splat() {
+; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } { <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1) }
+;
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave3.v4i32.v12i32(<vscale x 12 x i32> splat (i32 1))
+  ret {<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>} %1
+}
+
+define {<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>} @fold_scalable_vector_deinterleave3_splatfp() {
+; CHECK-LABEL: define { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave3_splatfp() {
+; CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } { <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00) }
+;
+  %1 = call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave3.v4f32.v12f32(<vscale x 12 x float> splat (float 1.0))
+  ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %1
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave4() {
+; CHECK-LABEL: define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave4() {
+; CHECK-NEXT:    ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } { <4 x i32> <i32 1, i32 2, i32 3, i32 4>, <4 x i32> <i32 5, i32 6, i32 7, i32 8>, <4 x i32> <i32 9, i32 10, i32 11, i32 12>, <4 x i32> <i32 13, i32 14, i32 15, i32 16> }
+;
+  %1 = call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.vector.deinterleave4.v4i32.v16i32(<16 x i32> <i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15, i32 4, i32 8, i32 12, i32 16>)
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %1
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave4() {
+; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave4() {
+; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } zeroinitializer
+;
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave4.v4i32.v16i32(<vscale x 16 x i32> zeroinitializer)
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %1
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave4_splat() {
+; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave4_splat() {
+; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } { <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1) }
+;
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave4.v4i32.v16i32(<vscale x 16 x i32> splat (i32 1))
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %1
+}
+
+define { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave4_splatfp() {
+; CHECK-LABEL: define { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave4_splatfp() {
+; CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } { <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00) }
+;
+  %1 = call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave4.v4f32.v16f32(<vscale x 16 x float> splat (float 1.0))
+  ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %1
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave5() {
+; CHECK-LABEL: define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave5() {
+; CHECK-NEXT:    ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } { <4 x i32> <i32 1, i32 2, i32 3, i32 4>, <4 x i32> <i32 5, i32 6, i32 7, i32 8>, <4 x i32> <i32 9, i32 10, i32 11, i32 12>, <4 x i32> <i32 13, i32 14, i32 15, i32 16>, <4 x i32> <i32 17, i32 18, i32 19, i32 20> }
+;
+  %1 = call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.vector.deinterleave5.v4i32.v20i32(<20 x i32> <i32 1, i32 5, i32 9, i32 13, i32 17, i32 2, i32 6, i32 10, i32 14, i32 18, i32 3, i32 7, i32 11, i32 15, i32 19, i32 4, i32 8, i32 12, i32 16, i32 20>)
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %1
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave5() {
+; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave5() {
+; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } zeroinitializer
+;
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave5.v4i32.v20i32(<vscale x 20 x i32> zeroinitializer)
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %1
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave5_splat() {
+; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave5_splat() {
+; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } { <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1) }
+;
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave5.v4i32.v20i32(<vscale x 20 x i32> splat (i32 1))
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %1
+}
+
+define { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave5_splatfp() {
+; CHECK-LABEL: define { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave5_splatfp() {
+; CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } { <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00) }
+;
+  %1 = call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave5.v4f32.v20f32(<vscale x 20 x float> splat (float 1.0))
+  ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %1
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave6() {
+; CHECK-LABEL: define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave6() {
+; CHECK-NEXT:    ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } { <4 x i32> <i32 1, i32 2, i32 3, i32 4>, <4 x i32> <i32 5, i32 6, i32 7, i32 8>, <4 x i32> <i32 9, i32 10, i32 11, i32 12>, <4 x i32> <i32 13, i32 14, i32 15, i32 16>, <4 x i32> <i32 17, i32 18, i32 19, i32 20>, <4 x i32> <i32 21, i32 22, i32 23, i32 24> }
+;
+  %1 = call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.vector.deinterleave6.v4i32.v24i32(<24 x i32> <i32 1, i32 5, i32 9, i32 13, i32 17, i32 21, i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 3, i32 7, i32 11, i32 15, i32 19, i32 23, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24>)
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %1
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave6() {
+; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave6() {
+; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } zeroinitializer
+;
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave6.v4i32.v24i32(<vscale x 24 x i32> zeroinitializer)
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %1
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave6_splat() {
+; CHECK-LABEL: define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @fold_scalable_vector_deinterleave6_splat() {
+; CHECK-NEXT:    ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } { <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1), <vscale x 4 x i32> splat (i32 1) }
+;
+  %1 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave6.v4i32.v24i32(<vscale x 24 x i32> splat (i32 1))
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %1
+}
+
+define { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave6_splatfp() {
+; CHECK-LABEL: define { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @fold_scalable_vector_deinterleave6_splatfp() {
+; CHECK-NEXT:    ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } { <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00), <vscale x 4 x float> splat (float 1.000000e+00) }
+;
+  %1 = call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave6.v4f32.v24f32(<vscale x 24 x float> splat (float 1.0))
+  ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %1
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave7() {
+; CHECK-LABEL: define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @fold_vector_deinterleave7() {
+; CHECK-NEXT:    ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } { <4 x i32> <i32 1, i32 2, i32 3, i32 4>, <4 x i32> <i32 5, i32 6, i32 7, i32 8>, <4 x i32> <i32 9, i32...
[truncated]

npanchen

overall lgtm

llvm/lib/Analysis/ConstantFolding.cpp

github-actions · 2025-11-19T01:23:10Z

🐧 Linux x64 Test Results

186402 tests passed
4858 tests skipped

preames

LGTM once stray whitespace is removed.

preames · 2025-11-19T15:00:55Z

llvm/test/Transforms/InstSimplify/ConstProp/vector-calls.ll

 }

-define {<4 x i32>, <4 x i32>} @fold_vector_deinterleave2() {
+define { <4 x i32>, <4 x i32> } @fold_vector_deinterleave2() {


Stray white-space change

I did it on purpose so it was in the format the IR printer uses so it matches the CHECK lines.

I'd prefer a precommit in that case, but don't need to block the review. Land it.

topperc added 2 commits November 18, 2025 15:23

Pre-commit tests

b99a2aa

[ConstantFolding] Generalize constant folding for vector_deinterleave…

72eac44

…2 to deinterleave3-8.

topperc requested review from mshockwave, nikic, npanchen and preames November 19, 2025 00:39

llvmbot added llvm:instcombine Covers the InstCombine, InstSimplify and AggressiveInstCombine passes llvm:analysis Includes value tracking, cost tables and constant folding llvm:transforms labels Nov 19, 2025

npanchen approved these changes Nov 19, 2025

View reviewed changes

llvm/lib/Analysis/ConstantFolding.cpp Outdated Show resolved Hide resolved

llvm/lib/Analysis/ConstantFolding.cpp Outdated Show resolved Hide resolved

fixup! address review comment

664b3ad

preames approved these changes Nov 19, 2025

View reviewed changes

topperc added 2 commits November 19, 2025 11:06

Merge remote-tracking branch 'origin/main' into pr/deinterleave

3e41d89

fixup! formatting

4b05507

topperc force-pushed the pr/deinterleave branch from 4149280 to 4b05507 Compare November 19, 2025 19:17

mshockwave approved these changes Nov 19, 2025

View reviewed changes

topperc merged commit 90ea49a into llvm:main Nov 19, 2025
8 of 9 checks passed

topperc deleted the pr/deinterleave branch November 19, 2025 22:27

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[ConstantFolding] Generalize constant folding for vector_deinterleave2 to deinterleave3-8. #168640

[ConstantFolding] Generalize constant folding for vector_deinterleave2 to deinterleave3-8. #168640

Uh oh!

topperc commented Nov 19, 2025

Uh oh!

llvmbot commented Nov 19, 2025 •

edited

Loading

Uh oh!

npanchen left a comment

Uh oh!

Uh oh!

Uh oh!

github-actions bot commented Nov 19, 2025 •

edited

Loading

Uh oh!

preames left a comment

Uh oh!

preames Nov 19, 2025

Uh oh!

topperc Nov 19, 2025 •

edited

Loading

Uh oh!

preames Nov 19, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

[ConstantFolding] Generalize constant folding for vector_deinterleave2 to deinterleave3-8. #168640

[ConstantFolding] Generalize constant folding for vector_deinterleave2 to deinterleave3-8. #168640

Uh oh!

Conversation

topperc commented Nov 19, 2025

Uh oh!

llvmbot commented Nov 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

npanchen left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

github-actions bot commented Nov 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🐧 Linux x64 Test Results

Uh oh!

preames left a comment

Choose a reason for hiding this comment

Uh oh!

preames Nov 19, 2025

Choose a reason for hiding this comment

Uh oh!

topperc Nov 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

preames Nov 19, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

llvmbot commented Nov 19, 2025 •

edited

Loading

github-actions bot commented Nov 19, 2025 •

edited

Loading

topperc Nov 19, 2025 •

edited

Loading