diff --git a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
index ac95ef5f30888..0c503bbc2f8c8 100644
--- a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
+++ b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
@@ -327,7 +327,7 @@ LoongArchTargetLowering::LoongArchTargetLowering(const TargetMachine &TM,
       setOperationAction(ISD::BUILD_VECTOR, VT, Custom);
 
       setOperationAction(ISD::SETCC, VT, Legal);
-      setOperationAction(ISD::VSELECT, VT, Legal);
+      setOperationAction(ISD::VSELECT, VT, Custom);
       setOperationAction(ISD::VECTOR_SHUFFLE, VT, Custom);
       setOperationAction(ISD::EXTRACT_SUBVECTOR, VT, Legal);
     }
@@ -415,7 +415,7 @@ LoongArchTargetLowering::LoongArchTargetLowering(const TargetMachine &TM,
       setOperationAction(ISD::INSERT_SUBVECTOR, VT, Legal);
 
       setOperationAction(ISD::SETCC, VT, Legal);
-      setOperationAction(ISD::VSELECT, VT, Legal);
+      setOperationAction(ISD::VSELECT, VT, Custom);
       setOperationAction(ISD::VECTOR_SHUFFLE, VT, Custom);
     }
     for (MVT VT : {MVT::v4i64, MVT::v8i32, MVT::v16i16, MVT::v32i8}) {
@@ -611,10 +611,57 @@ SDValue LoongArchTargetLowering::LowerOperation(SDValue Op,
     return lowerVECREDUCE(Op, DAG);
   case ISD::ConstantFP:
     return lowerConstantFP(Op, DAG);
+  case ISD::VSELECT:
+    return lowerVSELECT(Op, DAG);
   }
   return SDValue();
 }
 
+SDValue LoongArchTargetLowering::lowerVSELECT(SDValue Op,
+                                              SelectionDAG &DAG) const {
+  SDValue Cond = Op.getOperand(0);
+  SDValue LHS = Op.getOperand(1);
+  SDValue RHS = Op.getOperand(2);
+  MVT VT = Op.getSimpleValueType();
+  SDLoc DL(Op);
+
+  // Try to lower vselect to vector_shuffle. All cases with constant
+  // build_vector condition will be handled. NOTE: On 32-bit platform, vselect
+  // with v2i64/v4i64 type condition will never enter this because of the extra
+  // bitcast. It can be considered separately later.
+  if (ISD::isBuildVectorOfConstantSDNodes(Cond.getNode())) {
+    BuildVectorSDNode *BVN = cast<BuildVectorSDNode>(Cond.getNode());
+    APInt SplatValue, SplatUndef;
+    unsigned SplatBitSize;
+    bool HasAnyUndefs;
+
+    // If the Cond is a BUILD_VECTOR with splat constants, using a vldi or
+    // constant broadcast instruction and a vbitsel is better.
+    if (BVN->isConstantSplat(SplatValue, SplatUndef, SplatBitSize, HasAnyUndefs,
+                             /*MinSplatBits=*/8) &&
+        (SplatBitSize == 8 || SplatBitSize == 16 || SplatBitSize == 32 ||
+         SplatBitSize == 64))
+      return Op;
+
+    SmallVector<int, 32> Mask;
+    EVT CondVT = Cond.getValueType();
+    int NumElts = CondVT.getVectorNumElements();
+    Mask.resize(NumElts, -1);
+    for (int i = 0; i < NumElts; ++i) {
+      auto *CondElt = dyn_cast<ConstantSDNode>(Cond.getOperand(i));
+      if (CondElt->getZExtValue() == 0)
+        Mask[i] = i + NumElts;
+      else
+        Mask[i] = i;
+    }
+
+    return DAG.getVectorShuffle(VT, DL, LHS, RHS, Mask);
+  }
+
+  // Fallback to match patterns in tablegen.
+  return Op;
+}
+
 SDValue LoongArchTargetLowering::lowerConstantFP(SDValue Op,
                                                  SelectionDAG &DAG) const {
   EVT VT = Op.getValueType();
diff --git a/llvm/lib/Target/LoongArch/LoongArchISelLowering.h b/llvm/lib/Target/LoongArch/LoongArchISelLowering.h
index 232ac6092149d..75b2db87e0cfa 100644
--- a/llvm/lib/Target/LoongArch/LoongArchISelLowering.h
+++ b/llvm/lib/Target/LoongArch/LoongArchISelLowering.h
@@ -239,6 +239,7 @@ class LoongArchTargetLowering : public TargetLowering {
   SDValue lowerVECREDUCE_ADD(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerVECREDUCE(SDValue Op, SelectionDAG &DAG) const;
   SDValue lowerConstantFP(SDValue Op, SelectionDAG &DAG) const;
+  SDValue lowerVSELECT(SDValue Op, SelectionDAG &DAG) const;
 
   bool isFPImmLegal(const APFloat &Imm, EVT VT,
                     bool ForCodeSize) const override;
diff --git a/llvm/test/CodeGen/LoongArch/lasx/vselect.ll b/llvm/test/CodeGen/LoongArch/lasx/vselect.ll
index 559cc53062566..ccf4f57c24841 100644
--- a/llvm/test/CodeGen/LoongArch/lasx/vselect.ll
+++ b/llvm/test/CodeGen/LoongArch/lasx/vselect.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc --mtriple=loongarch32 --mattr=+32s,+lasx < %s | FileCheck %s
-; RUN: llc --mtriple=loongarch64 --mattr=+lasx < %s | FileCheck %s
+; RUN: llc --mtriple=loongarch32 --mattr=+32s,+lasx < %s | FileCheck %s --check-prefixes=CHECK,LA32
+; RUN: llc --mtriple=loongarch64 --mattr=+lasx < %s | FileCheck %s --check-prefixes=CHECK,LA64
 
 define void @select_v32i8_imm(ptr %res, ptr %a0) nounwind {
 ; CHECK-LABEL: select_v32i8_imm:
@@ -37,10 +37,8 @@ define void @select_v32i8_1(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    xvld $xr0, $a1, 0
 ; CHECK-NEXT:    xvld $xr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI2_0)
-; CHECK-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI2_0)
-; CHECK-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
-; CHECK-NEXT:    xvst $xr0, $a0, 0
+; CHECK-NEXT:    xvinsve0.w $xr1, $xr0, 0
+; CHECK-NEXT:    xvst $xr1, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <32 x i8>, ptr %a0
   %v1 = load <32 x i8>, ptr %a1
@@ -56,7 +54,7 @@ define void @select_v32i8_2(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK-NEXT:    xvld $xr1, $a2, 0
 ; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI3_0)
 ; CHECK-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI3_0)
-; CHECK-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
+; CHECK-NEXT:    xvshuf.b $xr0, $xr1, $xr0, $xr2
 ; CHECK-NEXT:    xvst $xr0, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <32 x i8>, ptr %a0
@@ -88,9 +86,7 @@ define void @select_v16i16_1(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    xvld $xr0, $a1, 0
 ; CHECK-NEXT:    xvld $xr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI5_0)
-; CHECK-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI5_0)
-; CHECK-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
+; CHECK-NEXT:    xvinsve0.w $xr0, $xr1, 0
 ; CHECK-NEXT:    xvst $xr0, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <16 x i16>, ptr %a0
@@ -107,8 +103,8 @@ define void @select_v16i16_2(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK-NEXT:    xvld $xr1, $a2, 0
 ; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI6_0)
 ; CHECK-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI6_0)
-; CHECK-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
-; CHECK-NEXT:    xvst $xr0, $a0, 0
+; CHECK-NEXT:    xvshuf.h $xr2, $xr1, $xr0
+; CHECK-NEXT:    xvst $xr2, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <16 x i16>, ptr %a0
   %v1 = load <16 x i16>, ptr %a1
@@ -138,10 +134,8 @@ define void @select_v8i32_1(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    xvld $xr0, $a1, 0
 ; CHECK-NEXT:    xvld $xr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI8_0)
-; CHECK-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI8_0)
-; CHECK-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
-; CHECK-NEXT:    xvst $xr0, $a0, 0
+; CHECK-NEXT:    xvinsve0.d $xr1, $xr0, 0
+; CHECK-NEXT:    xvst $xr1, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <8 x i32>, ptr %a0
   %v1 = load <8 x i32>, ptr %a1
@@ -157,8 +151,8 @@ define void @select_v8f32(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK-NEXT:    xvld $xr1, $a2, 0
 ; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI9_0)
 ; CHECK-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI9_0)
-; CHECK-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
-; CHECK-NEXT:    xvst $xr0, $a0, 0
+; CHECK-NEXT:    xvshuf.w $xr2, $xr1, $xr0
+; CHECK-NEXT:    xvst $xr2, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <8 x float>, ptr %a0
   %v1 = load <8 x float>, ptr %a1
@@ -168,15 +162,23 @@ define void @select_v8f32(ptr %res, ptr %a0, ptr %a1) nounwind {
 }
 
 define void @select_v4i64(ptr %res, ptr %a0, ptr %a1) nounwind {
-; CHECK-LABEL: select_v4i64:
-; CHECK:       # %bb.0:
-; CHECK-NEXT:    xvld $xr0, $a1, 0
-; CHECK-NEXT:    xvld $xr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI10_0)
-; CHECK-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI10_0)
-; CHECK-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
-; CHECK-NEXT:    xvst $xr0, $a0, 0
-; CHECK-NEXT:    ret
+; LA32-LABEL: select_v4i64:
+; LA32:       # %bb.0:
+; LA32-NEXT:    xvld $xr0, $a1, 0
+; LA32-NEXT:    xvld $xr1, $a2, 0
+; LA32-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI10_0)
+; LA32-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI10_0)
+; LA32-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
+; LA32-NEXT:    xvst $xr0, $a0, 0
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: select_v4i64:
+; LA64:       # %bb.0:
+; LA64-NEXT:    xvld $xr0, $a1, 0
+; LA64-NEXT:    xvld $xr1, $a2, 0
+; LA64-NEXT:    xvinsve0.d $xr1, $xr0, 0
+; LA64-NEXT:    xvst $xr1, $a0, 0
+; LA64-NEXT:    ret
   %v0 = load <4 x i64>, ptr %a0
   %v1 = load <4 x i64>, ptr %a1
   %sel = select <4 x i1> <i1 true, i1 false, i1 false, i1 false>, <4 x i64> %v0, <4 x i64> %v1
@@ -185,15 +187,25 @@ define void @select_v4i64(ptr %res, ptr %a0, ptr %a1) nounwind {
 }
 
 define void @select_v4f64(ptr %res, ptr %a0, ptr %a1) nounwind {
-; CHECK-LABEL: select_v4f64:
-; CHECK:       # %bb.0:
-; CHECK-NEXT:    xvld $xr0, $a1, 0
-; CHECK-NEXT:    xvld $xr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI11_0)
-; CHECK-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI11_0)
-; CHECK-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
-; CHECK-NEXT:    xvst $xr0, $a0, 0
-; CHECK-NEXT:    ret
+; LA32-LABEL: select_v4f64:
+; LA32:       # %bb.0:
+; LA32-NEXT:    xvld $xr0, $a1, 0
+; LA32-NEXT:    xvld $xr1, $a2, 0
+; LA32-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI11_0)
+; LA32-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI11_0)
+; LA32-NEXT:    xvbitsel.v $xr0, $xr1, $xr0, $xr2
+; LA32-NEXT:    xvst $xr0, $a0, 0
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: select_v4f64:
+; LA64:       # %bb.0:
+; LA64-NEXT:    xvld $xr0, $a1, 0
+; LA64-NEXT:    xvld $xr1, $a2, 0
+; LA64-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI11_0)
+; LA64-NEXT:    xvld $xr2, $a1, %pc_lo12(.LCPI11_0)
+; LA64-NEXT:    xvshuf.d $xr2, $xr1, $xr0
+; LA64-NEXT:    xvst $xr2, $a0, 0
+; LA64-NEXT:    ret
   %v0 = load <4 x double>, ptr %a0
   %v1 = load <4 x double>, ptr %a1
   %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x double> %v0, <4 x double> %v1
diff --git a/llvm/test/CodeGen/LoongArch/lsx/vselect.ll b/llvm/test/CodeGen/LoongArch/lsx/vselect.ll
index 25c4f099099db..e944c1d7fc418 100644
--- a/llvm/test/CodeGen/LoongArch/lsx/vselect.ll
+++ b/llvm/test/CodeGen/LoongArch/lsx/vselect.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc --mtriple=loongarch32 --mattr=+32s,+lsx < %s | FileCheck %s
-; RUN: llc --mtriple=loongarch64 --mattr=+lsx < %s | FileCheck %s
+; RUN: llc --mtriple=loongarch32 --mattr=+32s,+lsx < %s | FileCheck %s --check-prefixes=CHECK,LA32
+; RUN: llc --mtriple=loongarch64 --mattr=+lsx < %s | FileCheck %s --check-prefixes=CHECK,LA64
 
 define void @select_v16i8_imm(ptr %res, ptr %a0) nounwind {
 ; CHECK-LABEL: select_v16i8_imm:
@@ -51,9 +51,7 @@ define void @select_v16i8_1(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vld $vr0, $a1, 0
 ; CHECK-NEXT:    vld $vr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI3_0)
-; CHECK-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI3_0)
-; CHECK-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
+; CHECK-NEXT:    vshuf4i.d $vr0, $vr1, 6
 ; CHECK-NEXT:    vst $vr0, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <16 x i8>, ptr %a0
@@ -70,7 +68,7 @@ define void @select_v16i8_2(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK-NEXT:    vld $vr1, $a2, 0
 ; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI4_0)
 ; CHECK-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI4_0)
-; CHECK-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
+; CHECK-NEXT:    vshuf.b $vr0, $vr1, $vr0, $vr2
 ; CHECK-NEXT:    vst $vr0, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <16 x i8>, ptr %a0
@@ -102,9 +100,7 @@ define void @select_v8i16_1(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vld $vr0, $a1, 0
 ; CHECK-NEXT:    vld $vr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI6_0)
-; CHECK-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI6_0)
-; CHECK-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
+; CHECK-NEXT:    vshuf4i.d $vr0, $vr1, 12
 ; CHECK-NEXT:    vst $vr0, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <8 x i16>, ptr %a0
@@ -121,8 +117,8 @@ define void @select_v8i16_2(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK-NEXT:    vld $vr1, $a2, 0
 ; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI7_0)
 ; CHECK-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI7_0)
-; CHECK-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
-; CHECK-NEXT:    vst $vr0, $a0, 0
+; CHECK-NEXT:    vshuf.w $vr2, $vr1, $vr0
+; CHECK-NEXT:    vst $vr2, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <8 x i16>, ptr %a0
   %v1 = load <8 x i16>, ptr %a1
@@ -152,9 +148,7 @@ define void @select_v4i32_1(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vld $vr0, $a1, 0
 ; CHECK-NEXT:    vld $vr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI9_0)
-; CHECK-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI9_0)
-; CHECK-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
+; CHECK-NEXT:    vshuf4i.d $vr0, $vr1, 12
 ; CHECK-NEXT:    vst $vr0, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <4 x i32>, ptr %a0
@@ -171,8 +165,8 @@ define void @select_v4f32(ptr %res, ptr %a0, ptr %a1) nounwind {
 ; CHECK-NEXT:    vld $vr1, $a2, 0
 ; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI10_0)
 ; CHECK-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI10_0)
-; CHECK-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
-; CHECK-NEXT:    vst $vr0, $a0, 0
+; CHECK-NEXT:    vshuf.w $vr2, $vr1, $vr0
+; CHECK-NEXT:    vst $vr2, $a0, 0
 ; CHECK-NEXT:    ret
   %v0 = load <4 x float>, ptr %a0
   %v1 = load <4 x float>, ptr %a1
@@ -182,15 +176,23 @@ define void @select_v4f32(ptr %res, ptr %a0, ptr %a1) nounwind {
 }
 
 define void @select_v2i64(ptr %res, ptr %a0, ptr %a1) nounwind {
-; CHECK-LABEL: select_v2i64:
-; CHECK:       # %bb.0:
-; CHECK-NEXT:    vld $vr0, $a1, 0
-; CHECK-NEXT:    vld $vr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI11_0)
-; CHECK-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI11_0)
-; CHECK-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
-; CHECK-NEXT:    vst $vr0, $a0, 0
-; CHECK-NEXT:    ret
+; LA32-LABEL: select_v2i64:
+; LA32:       # %bb.0:
+; LA32-NEXT:    vld $vr0, $a1, 0
+; LA32-NEXT:    vld $vr1, $a2, 0
+; LA32-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI11_0)
+; LA32-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI11_0)
+; LA32-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
+; LA32-NEXT:    vst $vr0, $a0, 0
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: select_v2i64:
+; LA64:       # %bb.0:
+; LA64-NEXT:    vld $vr0, $a1, 0
+; LA64-NEXT:    vld $vr1, $a2, 0
+; LA64-NEXT:    vshuf4i.d $vr0, $vr1, 12
+; LA64-NEXT:    vst $vr0, $a0, 0
+; LA64-NEXT:    ret
   %v0 = load <2 x i64>, ptr %a0
   %v1 = load <2 x i64>, ptr %a1
   %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %v0, <2 x i64> %v1
@@ -199,15 +201,23 @@ define void @select_v2i64(ptr %res, ptr %a0, ptr %a1) nounwind {
 }
 
 define void @select_v2f64(ptr %res, ptr %a0, ptr %a1) nounwind {
-; CHECK-LABEL: select_v2f64:
-; CHECK:       # %bb.0:
-; CHECK-NEXT:    vld $vr0, $a1, 0
-; CHECK-NEXT:    vld $vr1, $a2, 0
-; CHECK-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI12_0)
-; CHECK-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI12_0)
-; CHECK-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
-; CHECK-NEXT:    vst $vr0, $a0, 0
-; CHECK-NEXT:    ret
+; LA32-LABEL: select_v2f64:
+; LA32:       # %bb.0:
+; LA32-NEXT:    vld $vr0, $a1, 0
+; LA32-NEXT:    vld $vr1, $a2, 0
+; LA32-NEXT:    pcalau12i $a1, %pc_hi20(.LCPI12_0)
+; LA32-NEXT:    vld $vr2, $a1, %pc_lo12(.LCPI12_0)
+; LA32-NEXT:    vbitsel.v $vr0, $vr1, $vr0, $vr2
+; LA32-NEXT:    vst $vr0, $a0, 0
+; LA32-NEXT:    ret
+;
+; LA64-LABEL: select_v2f64:
+; LA64:       # %bb.0:
+; LA64-NEXT:    vld $vr0, $a1, 0
+; LA64-NEXT:    vld $vr1, $a2, 0
+; LA64-NEXT:    vshuf4i.d $vr0, $vr1, 6
+; LA64-NEXT:    vst $vr0, $a0, 0
+; LA64-NEXT:    ret
   %v0 = load <2 x double>, ptr %a0
   %v1 = load <2 x double>, ptr %a1
   %sel = select <2 x i1> <i1 false, i1 true>, <2 x double> %v0, <2 x double> %v1