Supports lowerInterleavedStore() in X86InterleavedAccess.

[Farhana]

Summary

This is a follow-up on D24681 which supports lowerInterleavedLoad() on X86.

This change-set supports lowerInterleavedStore(). It mainly provides the necessary infrastructure/utilities in order to have lowerInterleavedStore() in place. It does not try to support more patterns beyond what X86InterleaveAccess already supports (currently, X86InterleavedAccess supports interleaved access with 64 x 4 bits in transpose4_4()).

[RKSimon]

Performance numbers?

Also, please add context to the diff.

test/CodeGen/X86/x86-interleaved-access.ll

143

Please add these new tests to trunk now with the current codegen so your patch shows the diff.

test/Transforms/InterleavedAccess/X86/interleaved-accesses-64bits-avx.ll

105

Please add these new tests to trunk now with the current codegen so your patch shows the diff.

[RKSimon]

And don't forget to subscribe to llvm-commits to get visibility on the mailing list!

[RKSimon]

lib/Target/X86/X86InterleavedAccess.cpp

212

for (unsigned i = 0, e = Shuffles.size(); i < e; i++)

[Farhana]

In D32658#741498, @RKSimon wrote:

Performance numbers?

Also, please add context to the diff.

Hi Simon,

Thanks for your prompt review.

Regarding performance, I don't have real-world examples. So, I tested it on a kernel where a loop contains only four scatters. On broadwell, this support improves the performance of that loop by ~70%.

Farhana

In D32658#741501, @RKSimon wrote:

And don't forget to subscribe to llvm-commits to get visibility on the mailing list!

Thanks! I subscribed.

Also, I added context to the svn-diff.

[Farhana]

Ping.

[RKSimon]

Please can you add the new tests to trunk now with the current codegen so your patch shows the diff.

[DavidKreitzer]

lib/Target/X86/X86InterleavedAccess.cpp

115

This routine is making some assumptions about the nature of the re-interleave ShuffleVectorInst. The optimized code sequence that you are emitting is only correct for a shuffle mask that is exactly {0, 4, 8, 12, 1, 5, 9, 13, 2, 6, 10, 14, 3, 7, 11, 15}. That is the common case, but InterleavedAccessPass is more general than that.

Consider the following example:

define void @store_factorf64_4(<16 x double>* %ptr, <4 x double> %v0, <4 x double> %v1, <4 x double> %v2, <4 x double> %v3) {
  %s0 = shufflevector <4 x double> %v0, <4 x double> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  %s1 = shufflevector <4 x double> %v2, <4 x double> %v3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  %interleaved.vec = shufflevector <8 x double> %s0, <8 x double> %s1, <16 x i32> <i32 12, i32 8, i32 4, i32 0, i32 13, i32 9, i32 5, i32 1, i32 14, i32 10, i32 6, i32 2, i32 15, i32 11, i32 7, i32 3>
  store <16 x double> %interleaved.vec, <16 x double>* %ptr, align 16
  ret void
}

This is identical to your store_factorf64_4 test case except that I've changed the mask for %interleaved.vec. With the current implementation, this example gets optimized and is compiled to the exact same code as your original test case, which is obviously incorrect behavior.

137

++i is preferred

141

This call to createSequentialMask looks incorrect. I think you want something like this:

unsigned NumSubVectorElements = getVectorNumElements(VecTy) / NumSubVectors;
createSequentialMask(Builder, i * NumSubVectorElements, NumSubVectorElements, 0);

What you have written only works because Factor, NumSubVectors, and NumSubVectorElements are all the same (i.e. 4) in the cases you currently support.

197

I think it is incorrect to be using Factor here. SubVecTy is not <Factor x ShuffleEltTy> but rather <(Wide Vector # Elems / Factor) x ShuffleEltTy>. This only works because both computations evaluate to 4 in the cases you currently support.

To illustrate what I mean, suppose instead we were optimizing 4 interleaved i8 vectors with a wide vector type of <128 x i8>. In this case, the SubVecTy should be <32 x i8>, but your current code will compute it as <4 x i8>.

208

++i is preferred

214

Please add '.' at the end of each sentence.

274

This array of indices is meaningless for the store optimization, right? Why not simply pass an empty ArrayRef to the X86InterleavedAccessGroup constructor?

[Farhana]

Updated the changes with the feedback from Dave.

lib/Target/X86/X86InterleavedAccess.cpp

115

Thanks Dave for pointing this out.

197

These functions were making some assumptions. I agree that they should not have done so.

[DavidKreitzer]

lib/Target/X86/X86InterleavedAccess.cpp

115

I like the way you fixed this, thanks!

201

good catch!

217

Use ShuffleTy instead of Shuffles[0]->getType()

274

You didn't address this comment. Maybe the best change here would be to use "Indices.push_back(Mask[i]);" That would be consistent with the change you made at line 141:

DecomposedVectors.push_back(
    cast<ShuffleVectorInst>(Builder.CreateShuffleVector(
        Op0, Op1, createSequentialMask(Builder, Mask[i],
                                       SubVecTy->getVectorNumElements(), 0))));

test/CodeGen/X86/x86-interleaved-access.ll

143

This a good suggestion, Farhana. Can you please do this in a separate initial patch?

[Farhana]

Updated diff with the latest changes in the trunk.

This update shows the improvement due to this change-set in the following tests.

test/Transforms/InterleavedAccess/X86/interleaved-accesses-64bits-avx.ll
test/CodeGen/X86/x86-interleaved-access.ll

[Farhana]

ping.

[DavidKreitzer]

Thanks for the fixes, Farhana. Other than a couple typos, this looks good.

lib/Target/X86/X86InterleavedAccess.cpp

272

"index each" --> "index of each"
"interleave" --> "interleaved"

test/CodeGen/X86/x86-interleaved-access.ll

143

Thanks! These are nice improvements in the codegen.

[Farhana]

This change-set reflects the fixed typos.

[DavidKreitzer]

Thanks, Farhana. LGTM.

[Farhana]

Hi Simon,

Do you have further comments/concerns?

Farhana

test/CodeGen/X86/x86-interleaved-access.ll

143

A very good suggestion. I just did not understand the suggestion at the first place.

[RKSimon]

In D32658#788057, @Farhana wrote:

Hi Simon,

Do you have further comments/concerns?

Farhana

No more comments from me. LGTM

lib/Target/X86/X86ISelLowering.h

public:
 
     /// \brief Lower interleaved load(s) into target specific
     /// instructions/intrinsics.
     bool lowerInterleavedLoad(LoadInst *LI,
                               ArrayRef<ShuffleVectorInst *> Shuffles,
                               ArrayRef<unsigned> Indices,
                               unsigned Factor) const override;
 
+    /// \brief Lower interleaved store(s) into target specific
+    /// instructions/intrinsics.
+    bool lowerInterleavedStore(StoreInst *SI, ShuffleVectorInst *SVI,
+                               unsigned Factor) const override;
+
+
     void finalizeLowering(MachineFunction &MF) const override;
 
   protected:
     std::pair<const TargetRegisterClass *, uint8_t>
     findRepresentativeClass(const TargetRegisterInfo *TRI,
                             MVT VT) const override;
 
   private:

lib/Target/X86/X86InterleavedAccess.cpp

 /// This file contains the X86 implementation of the interleaved accesses
 /// optimization generating X86-specific instructions/intrinsics for
 /// interleaved access groups.
 ///
 //===--------------------------------------------------------------------===//
 
 #include "X86ISelLowering.h"
 #include "X86TargetMachine.h"
+#include "llvm/Analysis/VectorUtils.h"
 
 using namespace llvm;
 
 namespace {
 /// \brief This class holds necessary information to represent an interleaved
 /// access group and supports utilities to lower the group into
 /// X86-specific instructions/intrinsics.
 ///  E.g. A group of interleaving access loads (Factor = 2; accessing every
class X86InterleavedAccessGroup {
   /// \brief Reference to the underlying target.
   const X86Subtarget &Subtarget;
 
   const DataLayout &DL;
 
   IRBuilder<> &Builder;
 
   /// \brief Breaks down a vector \p 'Inst' of N elements into \p NumSubVectors
-  /// sub vectors of type \p T. Returns true and the sub-vectors in
-  /// \p DecomposedVectors if it decomposes the Inst, returns false otherwise.
-  bool decompose(Instruction *Inst, unsigned NumSubVectors, VectorType *T,
+  /// sub vectors of type \p T. Returns the sub-vectors in \p DecomposedVectors.
+  void decompose(Instruction *Inst, unsigned NumSubVectors, VectorType *T,
                  SmallVectorImpl<Instruction *> &DecomposedVectors);
 
   /// \brief Performs matrix transposition on a 4x4 matrix \p InputVectors and
   /// returns the transposed-vectors in \p TransposedVectors.
   /// E.g.
   /// InputVectors:
   ///   In-V0 = p1, p2, p3, p4
   ///   In-V1 = q1, q2, q3, q4
public:
   /// In order to form an interleaved access group X86InterleavedAccessGroup
   /// requires a wide-load instruction \p 'I', a group of interleaved-vectors
   /// \p Shuffs, reference to the first indices of each interleaved-vector
   /// \p 'Ind' and the interleaving stride factor \p F. In order to generate
   /// X86-specific instructions/intrinsics it also requires the underlying
   /// target information \p STarget.
   explicit X86InterleavedAccessGroup(Instruction *I,
                                      ArrayRef<ShuffleVectorInst *> Shuffs,
-                                     ArrayRef<unsigned> Ind,
-                                     const unsigned F,
+                                     ArrayRef<unsigned> Ind, const unsigned F,
                                      const X86Subtarget &STarget,
                                      IRBuilder<> &B)
       : Inst(I), Shuffles(Shuffs), Indices(Ind), Factor(F), Subtarget(STarget),
         DL(Inst->getModule()->getDataLayout()), Builder(B) {}
 
   /// \brief Returns true if this interleaved access group can be lowered into
   /// x86-specific instructions/intrinsics, false otherwise.
   bool isSupported() const;
 
   /// \brief Lowers this interleaved access group into X86-specific
   /// instructions/intrinsics.
   bool lowerIntoOptimizedSequence();
 };
 } // end anonymous namespace
 
 bool X86InterleavedAccessGroup::isSupported() const {
   VectorType *ShuffleVecTy = Shuffles[0]->getType();
   uint64_t ShuffleVecSize = DL.getTypeSizeInBits(ShuffleVecTy);
   Type *ShuffleEltTy = ShuffleVecTy->getVectorElementType();
 
-  if (DL.getTypeSizeInBits(Inst->getType()) < Factor * ShuffleVecSize)
-    return false;
+  // Currently, lowering is supported for 4-element vectors of 64 bits on AVX.
+  uint64_t ExpectedShuffleVecSize;
+  if (isa<LoadInst>(Inst))
+    ExpectedShuffleVecSize = 256;
+  else
+    ExpectedShuffleVecSize = 1024;
 
-  // Currently, lowering is supported for 64 bits on AVX.
-  if (!Subtarget.hasAVX() || ShuffleVecSize != 256 ||
+  if (!Subtarget.hasAVX() || ShuffleVecSize != ExpectedShuffleVecSize ||
       DL.getTypeSizeInBits(ShuffleEltTy) != 64 || Factor != 4)
     return false;
 
   return true;
 }
 
-bool X86InterleavedAccessGroup::decompose(
+void X86InterleavedAccessGroup::decompose(
     Instruction *VecInst, unsigned NumSubVectors, VectorType *SubVecTy,
     SmallVectorImpl<Instruction *> &DecomposedVectors) {
+
+  assert((isa<LoadInst>(VecInst) || isa<ShuffleVectorInst>(VecInst)) &&
+         "Expected Load or Shuffle");
+
   Type *VecTy = VecInst->getType();
   (void)VecTy;
   assert(VecTy->isVectorTy() &&
          DL.getTypeSizeInBits(VecTy) >=
              DL.getTypeSizeInBits(SubVecTy) * NumSubVectors &&
          "Invalid Inst-size!!!");
-  assert(VecTy->getVectorElementType() == SubVecTy->getVectorElementType() &&
-         "Element type mismatched!!!");
 
-  if (!isa<LoadInst>(VecInst))
-    return false;
+  if (auto *SVI = dyn_cast<ShuffleVectorInst>(VecInst)) {
+    Value *Op0 = SVI->getOperand(0);
+    Value *Op1 = SVI->getOperand(1);
+
+    // Generate N(= NumSubVectors) shuffles of T(= SubVecTy) type.
+    for (unsigned i = 0; i < NumSubVectors; ++i)
+      DecomposedVectors.push_back(
+          cast<ShuffleVectorInst>(Builder.CreateShuffleVector(
+              Op0, Op1, createSequentialMask(Builder, Indices[i],
+                                             SubVecTy->getVectorNumElements(), 0))));
+    return;
+  }
 
+  // Decompose the load instruction.
   LoadInst *LI = cast<LoadInst>(VecInst);
   Type *VecBasePtrTy = SubVecTy->getPointerTo(LI->getPointerAddressSpace());
-
   Value *VecBasePtr =
       Builder.CreateBitCast(LI->getPointerOperand(), VecBasePtrTy);
 
-  // Generate N loads of T type
+  // Generate N loads of T type.
   for (unsigned i = 0; i < NumSubVectors; i++) {
-    // TODO: Support inbounds GEP
+    // TODO: Support inbounds GEP.
     Value *NewBasePtr = Builder.CreateGEP(VecBasePtr, Builder.getInt32(i));
     Instruction *NewLoad =
         Builder.CreateAlignedLoad(NewBasePtr, LI->getAlignment());
     DecomposedVectors.push_back(NewLoad);
   }
-
-  return true;
 }
 
 void X86InterleavedAccessGroup::transpose_4x4(
     ArrayRef<Instruction *> Matrix,
     SmallVectorImpl<Value *> &TransposedMatrix) {
   assert(Matrix.size() == 4 && "Invalid matrix size");
   TransposedMatrix.resize(4);
 
void X86InterleavedAccessGroup::transpose_4x4(
   TransposedMatrix[1] = Builder.CreateShuffleVector(IntrVec1, IntrVec2, Mask);
   TransposedMatrix[3] = Builder.CreateShuffleVector(IntrVec3, IntrVec4, Mask);
 }
 
 // Lowers this interleaved access group into X86-specific
 // instructions/intrinsics.
 bool X86InterleavedAccessGroup::lowerIntoOptimizedSequence() {
   SmallVector<Instruction *, 4> DecomposedVectors;
-  VectorType *VecTy = Shuffles[0]->getType();
+  SmallVector<Value *, 4> TransposedVectors;
+  VectorType *ShuffleTy = Shuffles[0]->getType();
+
+  if (isa<LoadInst>(Inst)) {
     // Try to generate target-sized register(/instruction).
-  if (!decompose(Inst, Factor, VecTy, DecomposedVectors))
-    return false;
+    decompose(Inst, Factor, ShuffleTy, DecomposedVectors);
 
-  SmallVector<Value *, 4> TransposedVectors;
     // Perform matrix-transposition in order to compute interleaved
     // results by generating some sort of (optimized) target-specific
     // instructions.
     transpose_4x4(DecomposedVectors, TransposedVectors);
 
     // Now replace the unoptimized-interleaved-vectors with the
     // transposed-interleaved vectors.
-  for (unsigned i = 0; i < Shuffles.size(); i++)
+    for (unsigned i = 0, e = Shuffles.size(); i < e; ++i)
       Shuffles[i]->replaceAllUsesWith(TransposedVectors[Indices[i]]);
 
     return true;
   }
 
+  Type *ShuffleEltTy = ShuffleTy->getVectorElementType();
+  unsigned NumSubVecElems = ShuffleTy->getVectorNumElements() / Factor;
+
+  // Lower the interleaved stores:
+  //   1. Decompose the interleaved wide shuffle into individual shuffle
+  //   vectors.
+  decompose(Shuffles[0], Factor,
+            VectorType::get(ShuffleEltTy, NumSubVecElems), DecomposedVectors);
+
+  //   2. Transpose the interleaved-vectors into vectors of contiguous
+  //      elements.
+  transpose_4x4(DecomposedVectors, TransposedVectors);
+
+  //   3. Concatenate the contiguous-vectors back into a wide vector.
+  Value *WideVec = concatenateVectors(Builder, TransposedVectors);
+
+  //   4. Generate a store instruction for wide-vec.
+  StoreInst *SI = cast<StoreInst>(Inst);
+  Builder.CreateAlignedStore(WideVec, SI->getPointerOperand(),
+                             SI->getAlignment());
+
+  return true;
+}
+
 // Lower interleaved load(s) into target specific instructions/
 // intrinsics. Lowering sequence varies depending on the vector-types, factor,
 // number of shuffles and ISA.
 // Currently, lowering is supported for 4x64 bits with Factor = 4 on AVX.
 bool X86TargetLowering::lowerInterleavedLoad(
     LoadInst *LI, ArrayRef<ShuffleVectorInst *> Shuffles,
     ArrayRef<unsigned> Indices, unsigned Factor) const {
   assert(Factor >= 2 && Factor <= getMaxSupportedInterleaveFactor() &&
          "Invalid interleave factor");
   assert(!Shuffles.empty() && "Empty shufflevector input");
   assert(Shuffles.size() == Indices.size() &&
          "Unmatched number of shufflevectors and indices");
 
   // Create an interleaved access group.
   IRBuilder<> Builder(LI);
   X86InterleavedAccessGroup Grp(LI, Shuffles, Indices, Factor, Subtarget,
                                 Builder);
 
   return Grp.isSupported() && Grp.lowerIntoOptimizedSequence();
 }
+
+bool X86TargetLowering::lowerInterleavedStore(StoreInst *SI,
+                                              ShuffleVectorInst *SVI,
+                                              unsigned Factor) const {
+  assert(Factor >= 2 && Factor <= getMaxSupportedInterleaveFactor() &&
+         "Invalid interleave factor");
+
+  VectorType *VecTy = SVI->getType();
+  assert(VecTy->getVectorNumElements() % Factor == 0 &&
+         "Invalid interleaved store");
+
+  // Holds the indices of SVI that correspond to the starting index each
+  // interleave shuffle.
+  SmallVector<unsigned, 4> Indices;
+  auto Mask = SVI->getShuffleMask();
+  for (unsigned i = 0; i < Factor; i++)
+    Indices.push_back(Mask[i]);
+
+  ArrayRef<ShuffleVectorInst *> Shuffles = makeArrayRef(SVI);
+
+  // Create an interleaved access group.
+  IRBuilder<> Builder(SI);
+  X86InterleavedAccessGroup Grp(SI, Shuffles, Indices, Factor, Subtarget,
+                                Builder);
+
+  return Grp.isSupported() && Grp.lowerIntoOptimizedSequence();
+}

test/CodeGen/X86/x86-interleaved-access.ll

; AVX2-NEXT:    retq
   %add2 = add <4 x i64> %add1, %strided.v2
   %add3 = add <4 x i64> %add2, %strided.v3
   ret <4 x i64> %add3
 }
 
 define void @store_factorf64_4(<16 x double>* %ptr, <4 x double> %v0, <4 x double> %v1, <4 x double> %v2, <4 x double> %v3) {
 ; AVX-LABEL: store_factorf64_4:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vunpcklpd {{.*#+}} xmm4 = xmm2[0],xmm3[0]
-; AVX-NEXT:    vinsertf128 $1, %xmm4, %ymm0, %ymm4
-; AVX-NEXT:    vunpcklpd {{.*#+}} xmm5 = xmm0[0],xmm1[0]
-; AVX-NEXT:    vblendpd {{.*#+}} ymm4 = ymm5[0,1],ymm4[2,3]
-; AVX-NEXT:    vunpckhpd {{.*#+}} xmm5 = xmm2[1],xmm3[1]
-; AVX-NEXT:    vinsertf128 $1, %xmm5, %ymm0, %ymm5
-; AVX-NEXT:    vunpckhpd {{.*#+}} xmm6 = xmm0[1],xmm1[1]
-; AVX-NEXT:    vblendpd {{.*#+}} ymm5 = ymm6[0,1],ymm5[2,3]
-; AVX-NEXT:    vunpcklpd {{.*#+}} ymm6 = ymm2[0],ymm3[0],ymm2[2],ymm3[2]
-; AVX-NEXT:    vunpcklpd {{.*#+}} ymm7 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
-; AVX-NEXT:    vextractf128 $1, %ymm7, %xmm7
-; AVX-NEXT:    vblendpd {{.*#+}} ymm6 = ymm7[0,1],ymm6[2,3]
-; AVX-NEXT:    vunpckhpd {{.*#+}} ymm2 = ymm2[1],ymm3[1],ymm2[3],ymm3[3]
+; AVX-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm4
+; AVX-NEXT:    vinsertf128 $1, %xmm3, %ymm1, %ymm5
+; AVX-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm2[2,3]
+; AVX-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm1[2,3],ymm3[2,3]
+; AVX-NEXT:    vunpcklpd {{.*#+}} ymm2 = ymm4[0],ymm5[0],ymm4[2],ymm5[2]
+; AVX-NEXT:    vunpcklpd {{.*#+}} ymm3 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
+; AVX-NEXT:    vunpckhpd {{.*#+}} ymm4 = ymm4[1],ymm5[1],ymm4[3],ymm5[3]
 ; AVX-NEXT:    vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
-; AVX-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm2[2,3]
 ; AVX-NEXT:    vmovupd %ymm0, 96(%rdi)
-; AVX-NEXT:    vmovupd %ymm6, 64(%rdi)
-; AVX-NEXT:    vmovupd %ymm5, 32(%rdi)
-; AVX-NEXT:    vmovupd %ymm4, (%rdi)
+; AVX-NEXT:    vmovupd %ymm3, 64(%rdi)
+; AVX-NEXT:    vmovupd %ymm4, 32(%rdi)
+; AVX-NEXT:    vmovupd %ymm2, (%rdi)
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
   %s0 = shufflevector <4 x double> %v0, <4 x double> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %s1 = shufflevector <4 x double> %v2, <4 x double> %v3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %interleaved.vec = shufflevector <8 x double> %s0, <8 x double> %s1, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15>
   store <16 x double> %interleaved.vec, <16 x double>* %ptr, align 16
   ret void
 }
 
 define void @store_factori64_4(<16 x i64>* %ptr, <4 x i64> %v0, <4 x i64> %v1, <4 x i64> %v2, <4 x i64> %v3) {
 ; AVX1-LABEL: store_factori64_4:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm4 = xmm2[0],xmm3[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm4, %ymm0, %ymm4
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm5 = xmm0[0],xmm1[0]
-; AVX1-NEXT:    vblendpd {{.*#+}} ymm4 = ymm5[0,1],ymm4[2,3]
-; AVX1-NEXT:    vpunpckhqdq {{.*#+}} xmm5 = xmm2[1],xmm3[1]
-; AVX1-NEXT:    vinsertf128 $1, %xmm5, %ymm0, %ymm5
-; AVX1-NEXT:    vpunpckhqdq {{.*#+}} xmm6 = xmm0[1],xmm1[1]
-; AVX1-NEXT:    vblendpd {{.*#+}} ymm5 = ymm6[0,1],ymm5[2,3]
-; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm6 = ymm2[0],ymm3[0],ymm2[2],ymm3[2]
-; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm7 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
-; AVX1-NEXT:    vextractf128 $1, %ymm7, %xmm7
-; AVX1-NEXT:    vblendpd {{.*#+}} ymm6 = ymm7[0,1],ymm6[2,3]
-; AVX1-NEXT:    vunpckhpd {{.*#+}} ymm2 = ymm2[1],ymm3[1],ymm2[3],ymm3[3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm4
+; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm1, %ymm5
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm2[2,3]
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm1[2,3],ymm3[2,3]
+; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm2 = ymm4[0],ymm5[0],ymm4[2],ymm5[2]
+; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm3 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
+; AVX1-NEXT:    vunpckhpd {{.*#+}} ymm4 = ymm4[1],ymm5[1],ymm4[3],ymm5[3]
 ; AVX1-NEXT:    vunpckhpd {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm2[2,3]
 ; AVX1-NEXT:    vmovupd %ymm0, 96(%rdi)
-; AVX1-NEXT:    vmovupd %ymm6, 64(%rdi)
-; AVX1-NEXT:    vmovupd %ymm5, 32(%rdi)
-; AVX1-NEXT:    vmovupd %ymm4, (%rdi)
+; AVX1-NEXT:    vmovupd %ymm3, 64(%rdi)
+; AVX1-NEXT:    vmovupd %ymm4, 32(%rdi)
+; AVX1-NEXT:    vmovupd %ymm2, (%rdi)
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: store_factori64_4:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} ymm4 = ymm2[0],ymm3[0],ymm2[2],ymm3[2]
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm5
-; AVX2-NEXT:    vpermq {{.*#+}} ymm6 = ymm1[0,2,2,3]
-; AVX2-NEXT:    vpblendd {{.*#+}} xmm5 = xmm5[0,1],xmm6[2,3]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm4 = ymm5[0,1,2,3],ymm4[4,5,6,7]
-; AVX2-NEXT:    vpunpckhqdq {{.*#+}} ymm5 = ymm2[1],ymm3[1],ymm2[3],ymm3[3]
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm6
-; AVX2-NEXT:    vpermq {{.*#+}} ymm7 = ymm0[3,1,2,3]
-; AVX2-NEXT:    vpblendd {{.*#+}} xmm6 = xmm7[0,1],xmm6[2,3]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm5 = ymm6[0,1,2,3],ymm5[4,5,6,7]
-; AVX2-NEXT:    vinserti128 $1, %xmm2, %ymm0, %ymm6
-; AVX2-NEXT:    vpbroadcastq %xmm3, %ymm7
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm6 = ymm6[0,1,2,3,4,5],ymm7[6,7]
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm7 = xmm0[0],xmm1[0]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm6 = ymm7[0,1,2,3],ymm6[4,5,6,7]
-; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm0, %ymm3
-; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,1,1,3]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5],ymm3[6,7]
-; AVX2-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm2[4,5,6,7]
-; AVX2-NEXT:    vmovdqu %ymm0, 32(%rdi)
-; AVX2-NEXT:    vmovdqu %ymm6, (%rdi)
-; AVX2-NEXT:    vmovdqu %ymm5, 96(%rdi)
-; AVX2-NEXT:    vmovdqu %ymm4, 64(%rdi)
+; AVX2-NEXT:    vinserti128 $1, %xmm2, %ymm0, %ymm4
+; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm1, %ymm5
+; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm2[2,3]
+; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm1[2,3],ymm3[2,3]
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} ymm2 = ymm4[0],ymm5[0],ymm4[2],ymm5[2]
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} ymm3 = ymm0[0],ymm1[0],ymm0[2],ymm1[2]
+; AVX2-NEXT:    vpunpckhqdq {{.*#+}} ymm4 = ymm4[1],ymm5[1],ymm4[3],ymm5[3]
+; AVX2-NEXT:    vpunpckhqdq {{.*#+}} ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3]
+; AVX2-NEXT:    vmovdqu %ymm0, 96(%rdi)
+; AVX2-NEXT:    vmovdqu %ymm3, 64(%rdi)
+; AVX2-NEXT:    vmovdqu %ymm4, 32(%rdi)
+; AVX2-NEXT:    vmovdqu %ymm2, (%rdi)
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
   %s0 = shufflevector <4 x i64> %v0, <4 x i64> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %s1 = shufflevector <4 x i64> %v2, <4 x i64> %v3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %interleaved.vec = shufflevector <8 x i64> %s0, <8 x i64> %s1, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15>
   store <16 x i64> %interleaved.vec, <16 x i64>* %ptr, align 16
   ret void
 }

test/Transforms/InterleavedAccess/X86/interleaved-accesses-64bits-avx.ll

 ;
   %wide.vec = load <16 x double>, <16 x double>* %ptr, align 16
   %strided.v0 = shufflevector <16 x double> %wide.vec, <16 x double> undef, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
   %strided.v3 = shufflevector <16 x double> %wide.vec, <16 x double> undef, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
   %mul = fmul <4 x double> %strided.v0, %strided.v3
   ret <4 x double> %mul
 }
 
 define void @store_factorf64_4(<16 x double>* %ptr, <4 x double> %v0, <4 x double> %v1, <4 x double> %v2, <4 x double> %v3) {
 ; CHECK-LABEL: @store_factorf64_4(
 ; CHECK-NEXT:    [[S0:%.*]] = shufflevector <4 x double> [[V0:%.*]], <4 x double> [[V1:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
 ; CHECK-NEXT:    [[S1:%.*]] = shufflevector <4 x double> [[V2:%.*]], <4 x double> [[V3:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEXT:    [[INTERLEAVED_VEC:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15>
-; CHECK-NEXT:    store <16 x double> [[INTERLEAVED_VEC]], <16 x double>* [[PTR:%.*]], align 16
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <4 x double> [[TMP1]], <4 x double> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> [[TMP4]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+; CHECK-NEXT:    [[TMP7:%.*]] = shufflevector <4 x double> [[TMP1]], <4 x double> [[TMP3]], <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP8:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> [[TMP4]], <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP9:%.*]] = shufflevector <4 x double> [[TMP5]], <4 x double> [[TMP6]], <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:    [[TMP10:%.*]] = shufflevector <4 x double> [[TMP7]], <4 x double> [[TMP8]], <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:    [[TMP11:%.*]] = shufflevector <4 x double> [[TMP5]], <4 x double> [[TMP6]], <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:    [[TMP12:%.*]] = shufflevector <4 x double> [[TMP7]], <4 x double> [[TMP8]], <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:    [[TMP13:%.*]] = shufflevector <4 x double> [[TMP9]], <4 x double> [[TMP11]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP14:%.*]] = shufflevector <4 x double> [[TMP10]], <4 x double> [[TMP12]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP15:%.*]] = shufflevector <8 x double> [[TMP13]], <8 x double> [[TMP14]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    store <16 x double> [[TMP15]], <16 x double>* [[PTR:%.*]], align 16
 ; CHECK-NEXT:    ret void
 ;
   %s0 = shufflevector <4 x double> %v0, <4 x double> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %s1 = shufflevector <4 x double> %v2, <4 x double> %v3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %interleaved.vec = shufflevector <8 x double> %s0, <8 x double> %s1, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15>
   store <16 x double> %interleaved.vec, <16 x double>* %ptr, align 16
   ret void
 }
 
 define void @store_factori64_4(<16 x i64>* %ptr, <4 x i64> %v0, <4 x i64> %v1, <4 x i64> %v2, <4 x i64> %v3) {
 ; CHECK-LABEL: @store_factori64_4(
 ; CHECK-NEXT:    [[S0:%.*]] = shufflevector <4 x i64> [[V0:%.*]], <4 x i64> [[V1:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
 ; CHECK-NEXT:    [[S1:%.*]] = shufflevector <4 x i64> [[V2:%.*]], <4 x i64> [[V3:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEXT:    [[INTERLEAVED_VEC:%.*]] = shufflevector <8 x i64> [[S0]], <8 x i64> [[S1]], <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15>
-; CHECK-NEXT:    store <16 x i64> [[INTERLEAVED_VEC]], <16 x i64>* [[PTR:%.*]], align 16
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <8 x i64> [[S0]], <8 x i64> [[S1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <8 x i64> [[S0]], <8 x i64> [[S1]], <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <8 x i64> [[S0]], <8 x i64> [[S1]], <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <8 x i64> [[S0]], <8 x i64> [[S1]], <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <4 x i64> [[TMP1]], <4 x i64> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <4 x i64> [[TMP2]], <4 x i64> [[TMP4]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+; CHECK-NEXT:    [[TMP7:%.*]] = shufflevector <4 x i64> [[TMP1]], <4 x i64> [[TMP3]], <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP8:%.*]] = shufflevector <4 x i64> [[TMP2]], <4 x i64> [[TMP4]], <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP9:%.*]] = shufflevector <4 x i64> [[TMP5]], <4 x i64> [[TMP6]], <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:    [[TMP10:%.*]] = shufflevector <4 x i64> [[TMP7]], <4 x i64> [[TMP8]], <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:    [[TMP11:%.*]] = shufflevector <4 x i64> [[TMP5]], <4 x i64> [[TMP6]], <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:    [[TMP12:%.*]] = shufflevector <4 x i64> [[TMP7]], <4 x i64> [[TMP8]], <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:    [[TMP13:%.*]] = shufflevector <4 x i64> [[TMP9]], <4 x i64> [[TMP11]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP14:%.*]] = shufflevector <4 x i64> [[TMP10]], <4 x i64> [[TMP12]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP15:%.*]] = shufflevector <8 x i64> [[TMP13]], <8 x i64> [[TMP14]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    store <16 x i64> [[TMP15]], <16 x i64>* [[PTR:%.*]], align 16
 ; CHECK-NEXT:    ret void
 ;
   %s0 = shufflevector <4 x i64> %v0, <4 x i64> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %s1 = shufflevector <4 x i64> %v2, <4 x i64> %v3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %interleaved.vec = shufflevector <8 x i64> %s0, <8 x i64> %s1, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15>
   store <16 x i64> %interleaved.vec, <16 x i64>* %ptr, align 16
   ret void
 }
 
 define void @store_factorf64_4_revMask(<16 x double>* %ptr, <4 x double> %v0, <4 x double> %v1, <4 x double> %v2, <4 x double> %v3) {
 ; CHECK-LABEL: @store_factorf64_4_revMask(
 ; CHECK-NEXT:    [[S0:%.*]] = shufflevector <4 x double> [[V0:%.*]], <4 x double> [[V1:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
 ; CHECK-NEXT:    [[S1:%.*]] = shufflevector <4 x double> [[V2:%.*]], <4 x double> [[V3:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEXT:    [[INTERLEAVED_VEC:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <16 x i32> <i32 12, i32 8, i32 4, i32 0, i32 13, i32 9, i32 5, i32 1, i32 14, i32 10, i32 6, i32 2, i32 15, i32 11, i32 7, i32 3>
-; CHECK-NEXT:    store <16 x double> [[INTERLEAVED_VEC]], <16 x double>* [[PTR:%.*]], align 16
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <8 x double> [[S0]], <8 x double> [[S1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <4 x double> [[TMP1]], <4 x double> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> [[TMP4]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+; CHECK-NEXT:    [[TMP7:%.*]] = shufflevector <4 x double> [[TMP1]], <4 x double> [[TMP3]], <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP8:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> [[TMP4]], <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP9:%.*]] = shufflevector <4 x double> [[TMP5]], <4 x double> [[TMP6]], <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:    [[TMP10:%.*]] = shufflevector <4 x double> [[TMP7]], <4 x double> [[TMP8]], <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:    [[TMP11:%.*]] = shufflevector <4 x double> [[TMP5]], <4 x double> [[TMP6]], <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:    [[TMP12:%.*]] = shufflevector <4 x double> [[TMP7]], <4 x double> [[TMP8]], <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:    [[TMP13:%.*]] = shufflevector <4 x double> [[TMP9]], <4 x double> [[TMP11]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP14:%.*]] = shufflevector <4 x double> [[TMP10]], <4 x double> [[TMP12]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP15:%.*]] = shufflevector <8 x double> [[TMP13]], <8 x double> [[TMP14]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    store <16 x double> [[TMP15]], <16 x double>* [[PTR:%.*]], align 16
 ; CHECK-NEXT:    ret void
 ;
   %s0 = shufflevector <4 x double> %v0, <4 x double> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %s1 = shufflevector <4 x double> %v2, <4 x double> %v3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %interleaved.vec = shufflevector <8 x double> %s0, <8 x double> %s1, <16 x i32> <i32 12, i32 8, i32 4, i32 0, i32 13, i32 9, i32 5, i32 1, i32 14, i32 10, i32 6, i32 2, i32 15, i32 11, i32 7, i32 3>
   store <16 x double> %interleaved.vec, <16 x double>* %ptr, align 16
   ret void
 }
 
 define void @store_factorf64_4_arbitraryMask(<16 x double>* %ptr, <16 x double> %v0, <16 x double> %v1, <16 x double> %v2, <16 x double> %v3) {
 ; CHECK-LABEL: @store_factorf64_4_arbitraryMask(
 ; CHECK-NEXT:    [[S0:%.*]] = shufflevector <16 x double> [[V0:%.*]], <16 x double> [[V1:%.*]], <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
 ; CHECK-NEXT:    [[S1:%.*]] = shufflevector <16 x double> [[V2:%.*]], <16 x double> [[V3:%.*]], <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
-; CHECK-NEXT:    [[INTERLEAVED_VEC:%.*]] = shufflevector <32 x double> [[S0]], <32 x double> [[S1]], <16 x i32> <i32 4, i32 32, i32 16, i32 8, i32 5, i32 33, i32 17, i32 9, i32 6, i32 34, i32 18, i32 10, i32 7, i32 35, i32 19, i32 11>
-; CHECK-NEXT:    store <16 x double> [[INTERLEAVED_VEC]], <16 x double>* [[PTR:%.*]], align 16
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <32 x double> [[S0]], <32 x double> [[S1]], <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <32 x double> [[S0]], <32 x double> [[S1]], <4 x i32> <i32 32, i32 33, i32 34, i32 35>
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <32 x double> [[S0]], <32 x double> [[S1]], <4 x i32> <i32 16, i32 17, i32 18, i32 19>
+; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <32 x double> [[S0]], <32 x double> [[S1]], <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <4 x double> [[TMP1]], <4 x double> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> [[TMP4]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+; CHECK-NEXT:    [[TMP7:%.*]] = shufflevector <4 x double> [[TMP1]], <4 x double> [[TMP3]], <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP8:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> [[TMP4]], <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP9:%.*]] = shufflevector <4 x double> [[TMP5]], <4 x double> [[TMP6]], <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:    [[TMP10:%.*]] = shufflevector <4 x double> [[TMP7]], <4 x double> [[TMP8]], <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:    [[TMP11:%.*]] = shufflevector <4 x double> [[TMP5]], <4 x double> [[TMP6]], <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:    [[TMP12:%.*]] = shufflevector <4 x double> [[TMP7]], <4 x double> [[TMP8]], <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:    [[TMP13:%.*]] = shufflevector <4 x double> [[TMP9]], <4 x double> [[TMP11]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP14:%.*]] = shufflevector <4 x double> [[TMP10]], <4 x double> [[TMP12]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT:    [[TMP15:%.*]] = shufflevector <8 x double> [[TMP13]], <8 x double> [[TMP14]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    store <16 x double> [[TMP15]], <16 x double>* [[PTR:%.*]], align 16
 ; CHECK-NEXT:    ret void
 ;
   %s0 = shufflevector <16 x double> %v0, <16 x double> %v1, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
   %s1 = shufflevector <16 x double> %v2, <16 x double> %v3, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
   %interleaved.vec = shufflevector <32 x double> %s0, <32 x double> %s1, <16 x i32> <i32 4, i32 32, i32 16, i32 8, i32 5, i32 33, i32 17, i32 9, i32 6, i32 34, i32 18, i32 10, i32 7, i32 35, i32 19, i32 11>
   store <16 x double> %interleaved.vec, <16 x double>* %ptr, align 16
   ret void
 }
-