[llvm] b215282 - Revert "[RISCV][TTI] Add shuffle costing for masked slide lowering (#128537)"

Philip Reames via llvm-commits llvm-commits at lists.llvm.org
Thu Feb 27 22:03:32 PST 2025


Author: Philip Reames
Date: 2025-02-27T22:03:18-08:00
New Revision: b2152823e003bb29c9161a55fabe76a3a3cb8b0a

URL: https://github.com/llvm/llvm-project/commit/b2152823e003bb29c9161a55fabe76a3a3cb8b0a
DIFF: https://github.com/llvm/llvm-project/commit/b2152823e003bb29c9161a55fabe76a3a3cb8b0a.diff

LOG: Revert "[RISCV][TTI] Add shuffle costing for masked slide lowering (#128537)"

This reverts commit 4904728cab8596320a77a895cb712fba07ea7bb1.  Downstream
test failed, reverting during investigation.

Added: 
    

Modified: 
    llvm/include/llvm/Analysis/VectorUtils.h
    llvm/lib/Analysis/VectorUtils.cpp
    llvm/lib/Target/RISCV/RISCVISelLowering.cpp
    llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
    llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
    llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll
    llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
    llvm/test/Analysis/CostModel/RISCV/shuffle-transpose.ll
    llvm/test/Transforms/SLPVectorizer/RISCV/complex-loads.ll
    llvm/test/Transforms/SLPVectorizer/RISCV/reductions.ll

Removed: 
    


################################################################################
diff  --git a/llvm/include/llvm/Analysis/VectorUtils.h b/llvm/include/llvm/Analysis/VectorUtils.h
index 4390b45f1f730..f21594c557e0e 100644
--- a/llvm/include/llvm/Analysis/VectorUtils.h
+++ b/llvm/include/llvm/Analysis/VectorUtils.h
@@ -203,15 +203,6 @@ bool getShuffleDemandedElts(int SrcWidth, ArrayRef<int> Mask,
                             const APInt &DemandedElts, APInt &DemandedLHS,
                             APInt &DemandedRHS, bool AllowUndefElts = false);
 
-/// Does this shuffle mask represent either one slide shuffle or a pair of
-/// two slide shuffles, combined with a select on some constant vector mask?
-/// A slide is a shuffle mask which shifts some set of elements up or down
-/// the vector, with all other elements being undefined.  An identity shuffle
-/// will be matched a slide by 0.  The output parameter provides the source
-/// (-1 means no source), and slide direction for each slide.
-bool isMaskedSlidePair(ArrayRef<int> Mask, int NumElts,
-                       std::array<std::pair<int, int>, 2> &SrcInfo);
-
 /// Replace each shuffle mask index with the scaled sequential indices for an
 /// equivalent mask of narrowed elements. Mask elements that are less than 0
 /// (sentinel values) are repeated in the output mask.

diff  --git a/llvm/lib/Analysis/VectorUtils.cpp b/llvm/lib/Analysis/VectorUtils.cpp
index 160913841a6e2..dcfd3d5a8bd6e 100644
--- a/llvm/lib/Analysis/VectorUtils.cpp
+++ b/llvm/lib/Analysis/VectorUtils.cpp
@@ -415,36 +415,6 @@ bool llvm::getShuffleDemandedElts(int SrcWidth, ArrayRef<int> Mask,
   return true;
 }
 
-bool llvm::isMaskedSlidePair(ArrayRef<int> Mask, int NumElts,
-                             std::array<std::pair<int, int>, 2> &SrcInfo) {
-  const int SignalValue = NumElts * 2;
-  SrcInfo[0] = {-1, SignalValue};
-  SrcInfo[1] = {-1, SignalValue};
-  for (auto [i, M] : enumerate(Mask)) {
-    if (M < 0)
-      continue;
-    int Src = M >= (int)NumElts;
-    int Diff = (int)i - (M % NumElts);
-    bool Match = false;
-    for (int j = 0; j < 2; j++) {
-      auto &[SrcE, DiffE] = SrcInfo[j];
-      if (SrcE == -1) {
-        assert(DiffE == SignalValue);
-        SrcE = Src;
-        DiffE = Diff;
-      }
-      if (SrcE == Src && DiffE == Diff) {
-        Match = true;
-        break;
-      }
-    }
-    if (!Match)
-      return false;
-  }
-  assert(SrcInfo[0].first != -1 && "Must find one slide");
-  return true;
-}
-
 void llvm::narrowShuffleMaskElts(int Scale, ArrayRef<int> Mask,
                                  SmallVectorImpl<int> &ScaledMask) {
   assert(Scale > 0 && "Unexpected scaling factor");

diff  --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 71fd3ab28b273..6076fe56416ad 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -4562,9 +4562,32 @@ static bool isInterleaveShuffle(ArrayRef<int> Mask, MVT VT, int &EvenSrc,
 
 /// Is this mask representing a masked combination of two slides?
 static bool isMaskedSlidePair(ArrayRef<int> Mask,
-                              std::array<std::pair<int, int>, 2> &SrcInfo) {
-  if (!llvm::isMaskedSlidePair(Mask, Mask.size(), SrcInfo))
-    return false;
+                              std::pair<int, int> SrcInfo[2]) {
+  int NumElts = Mask.size();
+  int SignalValue = NumElts * 2;
+  SrcInfo[0] = {-1, SignalValue};
+  SrcInfo[1] = {-1, SignalValue};
+  for (unsigned i = 0; i != Mask.size(); ++i) {
+    int M = Mask[i];
+    if (M < 0)
+      continue;
+    int Src = M >= (int)NumElts;
+    int Diff = (int)i - (M % NumElts);
+    bool Match = false;
+    for (int j = 0; j < 2; j++) {
+      if (SrcInfo[j].first == -1) {
+        assert(SrcInfo[j].second == SignalValue);
+        SrcInfo[j].first = Src;
+        SrcInfo[j].second = Diff;
+      }
+      if (SrcInfo[j].first == Src && SrcInfo[j].second == Diff) {
+        Match = true;
+        break;
+      }
+    }
+    if (!Match)
+      return false;
+  }
 
   // Avoid matching vselect idioms
   if (SrcInfo[0].second == 0 && SrcInfo[1].second == 0)
@@ -4580,8 +4603,7 @@ static bool isMaskedSlidePair(ArrayRef<int> Mask,
 
 // Exactly matches the semantics of a previously existing custom matcher
 // to allow migration to new matcher without changing output.
-static bool isElementRotate(std::array<std::pair<int, int>, 2> &SrcInfo,
-                            unsigned NumElts) {
+static bool isElementRotate(std::pair<int, int> SrcInfo[2], unsigned NumElts) {
   if (SrcInfo[1].first == -1)
     return true;
   return SrcInfo[0].second < 0 && SrcInfo[1].second > 0 &&
@@ -5582,10 +5604,10 @@ static SDValue lowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG,
   // without masking.  Avoid matching bit rotates (which are not also element
   // rotates) as slide pairs.  This is a performance heuristic, not a
   // functional check.
-  std::array<std::pair<int, int>, 2> SrcInfo;
+  std::pair<int, int> SrcInfo[2];
   unsigned RotateAmt;
   MVT RotateVT;
-  if (::isMaskedSlidePair(Mask, SrcInfo) &&
+  if (isMaskedSlidePair(Mask, SrcInfo) &&
       (isElementRotate(SrcInfo, NumElts) ||
        !isLegalBitRotate(Mask, VT, Subtarget, RotateVT, RotateAmt))) {
     SDValue Sources[2];
@@ -5942,11 +5964,10 @@ bool RISCVTargetLowering::isShuffleMaskLegal(ArrayRef<int> M, EVT VT) const {
   if (SVT.getScalarType() == MVT::i1)
     return false;
 
-  std::array<std::pair<int, int>, 2> SrcInfo;
+  std::pair<int, int> SrcInfo[2];
   int Dummy1, Dummy2;
   return ShuffleVectorInst::isReverseMask(M, NumElts) ||
-         (::isMaskedSlidePair(M, SrcInfo) &&
-          isElementRotate(SrcInfo, NumElts)) ||
+         (isMaskedSlidePair(M, SrcInfo) && isElementRotate(SrcInfo, NumElts)) ||
          isInterleaveShuffle(M, SVT, Dummy1, Dummy2, Subtarget);
 }
 

diff  --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 6005c067428eb..cde643a250be1 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -475,64 +475,6 @@ costShuffleViaVRegSplitting(RISCVTTIImpl &TTI, MVT LegalVT,
   return InstructionCost::getInvalid();
 }
 
-InstructionCost RISCVTTIImpl::getSlideCost(FixedVectorType *Tp,
-                                           ArrayRef<int> Mask,
-                                           TTI::TargetCostKind CostKind) {
-  // Avoid missing masks and length changing shuffles
-  if (Mask.size() <= 2 || Mask.size() != Tp->getNumElements())
-    return InstructionCost::getInvalid();
-
-  int NumElts = Tp->getNumElements();
-  std::pair<InstructionCost, MVT> LT = getTypeLegalizationCost(Tp);
-  // Avoid scalarization cases
-  if (!LT.second.isFixedLengthVector())
-    return InstructionCost::getInvalid();
-
-  // Requires moving elements between parts, which requires additional
-  // unmodeled instructions.
-  if (LT.first != 1)
-    return InstructionCost::getInvalid();
-
-  auto GetSlideOpcode = [&](int SlideAmt) {
-    assert(SlideAmt != 0);
-    bool IsVI = isUInt<5>(std::abs(SlideAmt));
-    if (SlideAmt < 0)
-      return IsVI ? RISCV::VSLIDEDOWN_VI : RISCV::VSLIDEDOWN_VX;
-    return IsVI ? RISCV::VSLIDEUP_VI : RISCV::VSLIDEUP_VX;
-  };
-
-  std::array<std::pair<int, int>, 2> SrcInfo;
-  if (!isMaskedSlidePair(Mask, NumElts, SrcInfo))
-    return InstructionCost::getInvalid();
-
-  if (SrcInfo[1].second == 0)
-    std::swap(SrcInfo[0], SrcInfo[1]);
-
-  InstructionCost FirstSlideCost = 0;
-  if (SrcInfo[0].second != 0) {
-    unsigned Opcode = GetSlideOpcode(SrcInfo[0].second);
-    FirstSlideCost = getRISCVInstructionCost(Opcode, LT.second, CostKind);
-  }
-
-  if (SrcInfo[1].first == -1)
-    return FirstSlideCost;
-
-  InstructionCost SecondSlideCost = 0;
-  if (SrcInfo[1].second != 0) {
-    unsigned Opcode = GetSlideOpcode(SrcInfo[1].second);
-    SecondSlideCost = getRISCVInstructionCost(Opcode, LT.second, CostKind);
-  } else {
-    SecondSlideCost =
-        getRISCVInstructionCost(RISCV::VMERGE_VVM, LT.second, CostKind);
-  }
-
-  auto EC = Tp->getElementCount();
-  VectorType *MaskTy =
-      VectorType::get(IntegerType::getInt1Ty(Tp->getContext()), EC);
-  InstructionCost MaskCost = getConstantPoolLoadCost(MaskTy, CostKind);
-  return FirstSlideCost + SecondSlideCost + MaskCost;
-}
-
 InstructionCost RISCVTTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
                                              VectorType *Tp, ArrayRef<int> Mask,
                                              TTI::TargetCostKind CostKind,
@@ -545,8 +487,8 @@ InstructionCost RISCVTTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
   // First, handle cases where having a fixed length vector enables us to
   // give a more accurate cost than falling back to generic scalable codegen.
   // TODO: Each of these cases hints at a modeling gap around scalable vectors.
-  if (auto *FVTp = dyn_cast<FixedVectorType>(Tp);
-      FVTp && ST->hasVInstructions() && LT.second.isFixedLengthVector()) {
+  if (ST->hasVInstructions() && isa<FixedVectorType>(Tp) &&
+      LT.second.isFixedLengthVector()) {
     InstructionCost VRegSplittingCost = costShuffleViaVRegSplitting(
         *this, LT.second, ST->getRealVLen(), Tp, Mask, CostKind);
     if (VRegSplittingCost.isValid())
@@ -602,11 +544,6 @@ InstructionCost RISCVTTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
           return Cost;
         }
       }
-
-      if (InstructionCost SlideCost = getSlideCost(FVTp, Mask, CostKind);
-          SlideCost.isValid())
-        return SlideCost;
-
       // vrgather + cost of generating the mask constant.
       // We model this for an unknown mask with a single vrgather.
       if (LT.first == 1 && (LT.second.getScalarSizeInBits() != 8 ||
@@ -621,11 +558,6 @@ InstructionCost RISCVTTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
     }
     case TTI::SK_Transpose:
     case TTI::SK_PermuteTwoSrc: {
-
-      if (InstructionCost SlideCost = getSlideCost(FVTp, Mask, CostKind);
-          SlideCost.isValid())
-        return SlideCost;
-
       // 2 x (vrgather + cost of generating the mask constant) + cost of mask
       // register for the second vrgather. We model this for an unknown
       // (shuffle) mask.

diff  --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
index 3f57560d3c127..134a7333b9b06 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
@@ -63,12 +63,6 @@ class RISCVTTIImpl : public BasicTTIImplBase<RISCVTTIImpl> {
   /// type.
   InstructionCost getConstantPoolLoadCost(Type *Ty,
                                           TTI::TargetCostKind CostKind);
-
-  /// If this shuffle can be lowered as a masked slide pair (at worst),
-  /// return a cost for it.
-  InstructionCost getSlideCost(FixedVectorType *Tp, ArrayRef<int> Mask,
-                               TTI::TargetCostKind CostKind);
-
 public:
   explicit RISCVTTIImpl(const RISCVTargetMachine *TM, const Function &F)
       : BaseT(TM, F.getDataLayout()), ST(TM->getSubtargetImpl(F)),

diff  --git a/llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll b/llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll
index 06c709e4cc879..c951184a31731 100644
--- a/llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll
@@ -186,7 +186,7 @@ define void @insert_subvec() vscale_range(2,2) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %v16i32_4_1 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %v16i32_4_2 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %v16i32_4_3 = shufflevector <16 x i32> poison, <16 x i32> poison, <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 16, i32 17, i32 18, i32 19>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16i32_4_05 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 18, i32 19, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v16i32_4_05 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 18, i32 19, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; CHECK-SIZE-LABEL: 'insert_subvec'
@@ -225,7 +225,7 @@ define void @insert_subvec() vscale_range(2,2) {
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %v16i32_4_1 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %v16i32_4_2 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %v16i32_4_3 = shufflevector <16 x i32> poison, <16 x i32> poison, <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 16, i32 17, i32 18, i32 19>
-; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16i32_4_05 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 18, i32 19, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v16i32_4_05 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 18, i32 19, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %v4i8_2_0 = shufflevector <4 x i8> poison, <4 x i8> poison, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
@@ -737,8 +737,8 @@ define void @multipart() vscale_range(2,2) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16b = shufflevector <8 x i16> poison, <8 x i16> poison, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v16c = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v16d = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32a = shufflevector <4 x i32> poison, <4 x i32> poison, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v32a4 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32a = shufflevector <4 x i32> poison, <4 x i32> poison, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v32a4 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v32idrev = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 15, i32 14, i32 13, i32 12, i32 16, i32 17, i32 18, i32 19, i32 31, i32 30, i32 29, i32 28>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %v32many = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %v32many2 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 1, i32 4, i32 8, i32 12, i32 17, i32 20, i32 24, i32 28, i32 2, i32 6, i32 11, i32 14, i32 18, i32 22, i32 27, i32 30>
@@ -757,8 +757,8 @@ define void @multipart() vscale_range(2,2) {
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16b = shufflevector <8 x i16> poison, <8 x i16> poison, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v16c = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v16d = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
-; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v32a = shufflevector <4 x i32> poison, <4 x i32> poison, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v32a4 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
+; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v32a = shufflevector <4 x i32> poison, <4 x i32> poison, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
+; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v32a4 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v32idrev = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 15, i32 14, i32 13, i32 12, i32 16, i32 17, i32 18, i32 19, i32 31, i32 30, i32 29, i32 28>
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v32many = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30>
 ; CHECK-SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v32many2 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 1, i32 4, i32 8, i32 12, i32 17, i32 20, i32 24, i32 28, i32 2, i32 6, i32 11, i32 14, i32 18, i32 22, i32 27, i32 30>

diff  --git a/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll b/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
index d2bfb61a11b00..e8dd30345cc76 100644
--- a/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
@@ -19,7 +19,7 @@ define void @test_vXf64(<4 x double> %src256, <8 x double> %src512) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_2345 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_4567 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 poison>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 poison>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; VLEN128-LABEL: 'test_vXf64'

diff  --git a/llvm/test/Analysis/CostModel/RISCV/shuffle-transpose.ll b/llvm/test/Analysis/CostModel/RISCV/shuffle-transpose.ll
index ef069fee8526e..8f784a07d3124 100644
--- a/llvm/test/Analysis/CostModel/RISCV/shuffle-transpose.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/shuffle-transpose.ll
@@ -10,11 +10,11 @@ target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
 
 define <8 x i8> @trn1.v8i8(<8 x i8> %v0, <8 x i8> %v1) {
 ; CHECK-LABEL: 'trn1.v8i8'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i8> %tmp0
 ;
 ; SIZE-LABEL: 'trn1.v8i8'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i8> %tmp0
 ;
   %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
@@ -23,11 +23,11 @@ define <8 x i8> @trn1.v8i8(<8 x i8> %v0, <8 x i8> %v1) {
 
 define <8 x i8> @trn2.v8i8(<8 x i8> %v0, <8 x i8> %v1) {
 ; CHECK-LABEL: 'trn2.v8i8'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i8> %tmp0
 ;
 ; SIZE-LABEL: 'trn2.v8i8'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i8> %tmp0
 ;
   %tmp0 = shufflevector <8 x i8> %v0, <8 x i8> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
@@ -36,11 +36,11 @@ define <8 x i8> @trn2.v8i8(<8 x i8> %v0, <8 x i8> %v1) {
 
 define <16 x i8> @trn1.v16i8(<16 x i8> %v0, <16 x i8> %v1) {
 ; CHECK-LABEL: 'trn1.v16i8'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %tmp0
 ;
 ; SIZE-LABEL: 'trn1.v16i8'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %tmp0
 ;
   %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
@@ -49,11 +49,11 @@ define <16 x i8> @trn1.v16i8(<16 x i8> %v0, <16 x i8> %v1) {
 
 define <16 x i8> @trn2.v16i8(<16 x i8> %v0, <16 x i8> %v1) {
 ; CHECK-LABEL: 'trn2.v16i8'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 1, i32 17, i32 3, i32 19, i32 5, i32 21, i32 7, i32 23, i32 9, i32 25, i32 11, i32 27, i32 13, i32 29, i32 15, i32 31>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 1, i32 17, i32 3, i32 19, i32 5, i32 21, i32 7, i32 23, i32 9, i32 25, i32 11, i32 27, i32 13, i32 29, i32 15, i32 31>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %tmp0
 ;
 ; SIZE-LABEL: 'trn2.v16i8'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 1, i32 17, i32 3, i32 19, i32 5, i32 21, i32 7, i32 23, i32 9, i32 25, i32 11, i32 27, i32 13, i32 29, i32 15, i32 31>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 1, i32 17, i32 3, i32 19, i32 5, i32 21, i32 7, i32 23, i32 9, i32 25, i32 11, i32 27, i32 13, i32 29, i32 15, i32 31>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <16 x i8> %tmp0
 ;
   %tmp0 = shufflevector <16 x i8> %v0, <16 x i8> %v1, <16 x i32> <i32 1, i32 17, i32 3, i32 19, i32 5, i32 21, i32 7, i32 23, i32 9, i32 25, i32 11, i32 27, i32 13, i32 29, i32 15, i32 31>
@@ -62,11 +62,11 @@ define <16 x i8> @trn2.v16i8(<16 x i8> %v0, <16 x i8> %v1) {
 
 define <4 x i16> @trn1.v4i16(<4 x i16> %v0, <4 x i16> %v1) {
 ; CHECK-LABEL: 'trn1.v4i16'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i16> %tmp0
 ;
 ; SIZE-LABEL: 'trn1.v4i16'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i16> %tmp0
 ;
   %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
@@ -75,11 +75,11 @@ define <4 x i16> @trn1.v4i16(<4 x i16> %v0, <4 x i16> %v1) {
 
 define <4 x i16> @trn2.v4i16(<4 x i16> %v0, <4 x i16> %v1) {
 ; CHECK-LABEL: 'trn2.v4i16'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i16> %tmp0
 ;
 ; SIZE-LABEL: 'trn2.v4i16'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i16> %tmp0
 ;
   %tmp0 = shufflevector <4 x i16> %v0, <4 x i16> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
@@ -88,11 +88,11 @@ define <4 x i16> @trn2.v4i16(<4 x i16> %v0, <4 x i16> %v1) {
 
 define <8 x i16> @trn1.v8i16(<8 x i16> %v0, <8 x i16> %v1) {
 ; CHECK-LABEL: 'trn1.v8i16'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %tmp0
 ;
 ; SIZE-LABEL: 'trn1.v8i16'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i16> %tmp0
 ;
   %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
@@ -101,11 +101,11 @@ define <8 x i16> @trn1.v8i16(<8 x i16> %v0, <8 x i16> %v1) {
 
 define <8 x i16> @trn2.v8i16(<8 x i16> %v0, <8 x i16> %v1) {
 ; CHECK-LABEL: 'trn2.v8i16'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %tmp0
 ;
 ; SIZE-LABEL: 'trn2.v8i16'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x i16> %tmp0
 ;
   %tmp0 = shufflevector <8 x i16> %v0, <8 x i16> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
@@ -140,11 +140,11 @@ define <2 x i32> @trn2.v2i32(<2 x i32> %v0, <2 x i32> %v1) {
 
 define <4 x i32> @trn1.v4i32(<4 x i32> %v0, <4 x i32> %v1) {
 ; CHECK-LABEL: 'trn1.v4i32'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %tmp0
 ;
 ; SIZE-LABEL: 'trn1.v4i32'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %tmp0
 ;
   %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
@@ -153,11 +153,11 @@ define <4 x i32> @trn1.v4i32(<4 x i32> %v0, <4 x i32> %v1) {
 
 define <4 x i32> @trn2.v4i32(<4 x i32> %v0, <4 x i32> %v1) {
 ; CHECK-LABEL: 'trn2.v4i32'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %tmp0
 ;
 ; SIZE-LABEL: 'trn2.v4i32'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x i32> %tmp0
 ;
   %tmp0 = shufflevector <4 x i32> %v0, <4 x i32> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
@@ -218,11 +218,11 @@ define <2 x float> @trn2.v2f32(<2 x float> %v0, <2 x float> %v1) {
 
 define <4 x float> @trn1.v4f32(<4 x float> %v0, <4 x float> %v1) {
 ; CHECK-LABEL: 'trn1.v4f32'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %tmp0
 ;
 ; SIZE-LABEL: 'trn1.v4f32'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x float> %tmp0
 ;
   %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
@@ -231,11 +231,11 @@ define <4 x float> @trn1.v4f32(<4 x float> %v0, <4 x float> %v1) {
 
 define <4 x float> @trn2.v4f32(<4 x float> %v0, <4 x float> %v1) {
 ; CHECK-LABEL: 'trn2.v4f32'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %tmp0
 ;
 ; SIZE-LABEL: 'trn2.v4f32'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x float> %tmp0
 ;
   %tmp0 = shufflevector <4 x float> %v0, <4 x float> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
@@ -270,11 +270,11 @@ define <2 x double> @trn2.v2f64(<2 x double> %v0, <2 x double> %v1) {
 
 define <4 x half> @trn1.v4f16(<4 x half> %v0, <4 x half> %v1) {
 ; CHECK-LABEL: 'trn1.v4f16'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x half> %tmp0
 ;
 ; SIZE-LABEL: 'trn1.v4f16'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x half> %tmp0
 ;
   %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
@@ -283,11 +283,11 @@ define <4 x half> @trn1.v4f16(<4 x half> %v0, <4 x half> %v1) {
 
 define <4 x half> @trn2.v4f16(<4 x half> %v0, <4 x half> %v1) {
 ; CHECK-LABEL: 'trn2.v4f16'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x half> %tmp0
 ;
 ; SIZE-LABEL: 'trn2.v4f16'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <4 x half> %tmp0
 ;
   %tmp0 = shufflevector <4 x half> %v0, <4 x half> %v1, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
@@ -296,11 +296,11 @@ define <4 x half> @trn2.v4f16(<4 x half> %v0, <4 x half> %v1) {
 
 define <8 x half> @trn1.v8f16(<8 x half> %v0, <8 x half> %v1) {
 ; CHECK-LABEL: 'trn1.v8f16'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x half> %tmp0
 ;
 ; SIZE-LABEL: 'trn1.v8f16'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x half> %tmp0
 ;
   %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
@@ -309,11 +309,11 @@ define <8 x half> @trn1.v8f16(<8 x half> %v0, <8 x half> %v1) {
 
 define <8 x half> @trn2.v8f16(<8 x half> %v0, <8 x half> %v1) {
 ; CHECK-LABEL: 'trn2.v8f16'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x half> %tmp0
 ;
 ; SIZE-LABEL: 'trn2.v8f16'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
+; SIZE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
 ; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret <8 x half> %tmp0
 ;
   %tmp0 = shufflevector <8 x half> %v0, <8 x half> %v1, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>

diff  --git a/llvm/test/Transforms/SLPVectorizer/RISCV/complex-loads.ll b/llvm/test/Transforms/SLPVectorizer/RISCV/complex-loads.ll
index 18acae5835724..11fa3337544a1 100644
--- a/llvm/test/Transforms/SLPVectorizer/RISCV/complex-loads.ll
+++ b/llvm/test/Transforms/SLPVectorizer/RISCV/complex-loads.ll
@@ -6,175 +6,663 @@ define i32 @test(ptr %pix1, ptr %pix2, i64 %idx.ext, i64 %idx.ext63, ptr %add.pt
 ; CHECK-LABEL: define i32 @test(
 ; CHECK-SAME: ptr [[PIX1:%.*]], ptr [[PIX2:%.*]], i64 [[IDX_EXT:%.*]], i64 [[IDX_EXT63:%.*]], ptr [[ADD_PTR:%.*]], ptr [[ADD_PTR64:%.*]]) #[[ATTR0:[0-9]+]] {
 ; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i8, ptr [[PIX1]], align 1
+; CHECK-NEXT:    [[CONV1:%.*]] = zext i8 [[TMP0]] to i32
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr i8, ptr [[PIX1]], i64 4
 ; CHECK-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr i8, ptr [[PIX2]], i64 4
+; CHECK-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr i8, ptr [[PIX1]], i64 1
+; CHECK-NEXT:    [[ARRAYIDX32:%.*]] = getelementptr i8, ptr [[PIX1]], i64 3
+; CHECK-NEXT:    [[TMP10:%.*]] = load i8, ptr [[ARRAYIDX32]], align 1
+; CHECK-NEXT:    [[CONV33:%.*]] = zext i8 [[TMP10]] to i32
 ; CHECK-NEXT:    [[ADD_PTR3:%.*]] = getelementptr i8, ptr [[PIX1]], i64 [[IDX_EXT]]
 ; CHECK-NEXT:    [[ADD_PTR644:%.*]] = getelementptr i8, ptr [[PIX2]], i64 [[IDX_EXT63]]
+; CHECK-NEXT:    [[TMP11:%.*]] = load i8, ptr [[ADD_PTR3]], align 1
+; CHECK-NEXT:    [[CONV_1:%.*]] = zext i8 [[TMP11]] to i32
 ; CHECK-NEXT:    [[ARRAYIDX3_1:%.*]] = getelementptr i8, ptr [[ADD_PTR3]], i64 4
 ; CHECK-NEXT:    [[ARRAYIDX5_1:%.*]] = getelementptr i8, ptr [[ADD_PTR644]], i64 4
+; CHECK-NEXT:    [[ARRAYIDX8_1:%.*]] = getelementptr i8, ptr [[ADD_PTR3]], i64 1
+; CHECK-NEXT:    [[ARRAYIDX27_1:%.*]] = getelementptr i8, ptr [[ADD_PTR3]], i64 3
+; CHECK-NEXT:    [[TMP5:%.*]] = load i8, ptr [[ARRAYIDX27_1]], align 1
+; CHECK-NEXT:    [[CONV33_1:%.*]] = zext i8 [[TMP5]] to i32
 ; CHECK-NEXT:    [[ADD_PTR_1:%.*]] = getelementptr i8, ptr [[ADD_PTR]], i64 [[IDX_EXT]]
 ; CHECK-NEXT:    [[ADD_PTR64_1:%.*]] = getelementptr i8, ptr [[ADD_PTR64]], i64 [[IDX_EXT63]]
 ; CHECK-NEXT:    [[ARRAYIDX3_2:%.*]] = getelementptr i8, ptr [[ADD_PTR_1]], i64 4
 ; CHECK-NEXT:    [[ARRAYIDX5_2:%.*]] = getelementptr i8, ptr [[ADD_PTR64_1]], i64 4
+; CHECK-NEXT:    [[TMP4:%.*]] = load <4 x i8>, ptr [[ADD_PTR_1]], align 1
+; CHECK-NEXT:    [[TMP19:%.*]] = shufflevector <4 x i8> [[TMP4]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP21:%.*]] = zext <2 x i8> [[TMP19]] to <2 x i32>
+; CHECK-NEXT:    [[TMP9:%.*]] = load <4 x i8>, ptr [[ADD_PTR64_1]], align 1
+; CHECK-NEXT:    [[TMP22:%.*]] = shufflevector <4 x i8> [[TMP9]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP31:%.*]] = zext <2 x i8> [[TMP22]] to <2 x i32>
+; CHECK-NEXT:    [[TMP23:%.*]] = sub <2 x i32> [[TMP21]], [[TMP31]]
+; CHECK-NEXT:    [[TMP13:%.*]] = load <4 x i8>, ptr [[ARRAYIDX3_2]], align 1
+; CHECK-NEXT:    [[TMP49:%.*]] = shufflevector <4 x i8> [[TMP13]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP26:%.*]] = zext <2 x i8> [[TMP49]] to <2 x i32>
+; CHECK-NEXT:    [[TMP16:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5_2]], align 1
+; CHECK-NEXT:    [[TMP27:%.*]] = shufflevector <4 x i8> [[TMP16]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP50:%.*]] = zext <2 x i8> [[TMP27]] to <2 x i32>
+; CHECK-NEXT:    [[TMP24:%.*]] = sub <2 x i32> [[TMP26]], [[TMP50]]
+; CHECK-NEXT:    [[TMP25:%.*]] = shl <2 x i32> [[TMP24]], splat (i32 16)
+; CHECK-NEXT:    [[TMP30:%.*]] = add <2 x i32> [[TMP25]], [[TMP23]]
+; CHECK-NEXT:    [[TMP32:%.*]] = shufflevector <4 x i8> [[TMP4]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP51:%.*]] = zext <2 x i8> [[TMP32]] to <2 x i32>
+; CHECK-NEXT:    [[TMP56:%.*]] = shufflevector <4 x i8> [[TMP9]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP57:%.*]] = zext <2 x i8> [[TMP56]] to <2 x i32>
+; CHECK-NEXT:    [[TMP35:%.*]] = sub <2 x i32> [[TMP51]], [[TMP57]]
+; CHECK-NEXT:    [[TMP38:%.*]] = shufflevector <4 x i8> [[TMP13]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP39:%.*]] = zext <2 x i8> [[TMP38]] to <2 x i32>
+; CHECK-NEXT:    [[TMP40:%.*]] = shufflevector <4 x i8> [[TMP16]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP61:%.*]] = zext <2 x i8> [[TMP40]] to <2 x i32>
+; CHECK-NEXT:    [[TMP36:%.*]] = sub <2 x i32> [[TMP39]], [[TMP61]]
+; CHECK-NEXT:    [[TMP37:%.*]] = shl <2 x i32> [[TMP36]], splat (i32 16)
+; CHECK-NEXT:    [[TMP42:%.*]] = add <2 x i32> [[TMP37]], [[TMP35]]
+; CHECK-NEXT:    [[TMP34:%.*]] = add <2 x i32> [[TMP42]], [[TMP30]]
+; CHECK-NEXT:    [[TMP44:%.*]] = sub <2 x i32> [[TMP30]], [[TMP42]]
+; CHECK-NEXT:    [[TMP43:%.*]] = extractelement <2 x i32> [[TMP34]], i32 0
+; CHECK-NEXT:    [[CONV_2:%.*]] = extractelement <2 x i32> [[TMP34]], i32 1
+; CHECK-NEXT:    [[ADD48_2:%.*]] = add i32 [[CONV_2]], [[TMP43]]
+; CHECK-NEXT:    [[TMP46:%.*]] = extractelement <2 x i32> [[TMP44]], i32 0
+; CHECK-NEXT:    [[TMP47:%.*]] = extractelement <2 x i32> [[TMP44]], i32 1
+; CHECK-NEXT:    [[ADD55_2:%.*]] = add i32 [[TMP47]], [[TMP46]]
 ; CHECK-NEXT:    [[ARRAYIDX5_3:%.*]] = getelementptr i8, ptr null, i64 4
+; CHECK-NEXT:    [[TMP53:%.*]] = load <2 x i8>, ptr null, align 1
 ; CHECK-NEXT:    [[TMP52:%.*]] = load i8, ptr null, align 1
-; CHECK-NEXT:    [[TMP3:%.*]] = load i8, ptr null, align 1
-; CHECK-NEXT:    [[TMP2:%.*]] = load <4 x i8>, ptr [[PIX1]], align 1
+; CHECK-NEXT:    [[TMP62:%.*]] = zext <2 x i8> [[TMP53]] to <2 x i32>
+; CHECK-NEXT:    [[TMP77:%.*]] = zext i8 [[TMP52]] to i32
+; CHECK-NEXT:    [[TMP54:%.*]] = load <2 x i8>, ptr null, align 1
+; CHECK-NEXT:    [[TMP55:%.*]] = zext <2 x i8> [[TMP54]] to <2 x i32>
+; CHECK-NEXT:    [[TMP59:%.*]] = sub <2 x i32> [[TMP62]], [[TMP55]]
+; CHECK-NEXT:    [[TMP41:%.*]] = call <2 x i8> @llvm.experimental.vp.strided.load.v2i8.p0.i64(ptr align 1 null, i64 4, <2 x i1> splat (i1 true), i32 2)
+; CHECK-NEXT:    [[TMP58:%.*]] = zext <2 x i8> [[TMP41]] to <2 x i32>
+; CHECK-NEXT:    [[TMP48:%.*]] = shufflevector <2 x i32> [[TMP58]], <2 x i32> poison, <2 x i32> <i32 1, i32 0>
+; CHECK-NEXT:    [[TMP63:%.*]] = load <2 x i8>, ptr [[ARRAYIDX5_3]], align 1
+; CHECK-NEXT:    [[TMP76:%.*]] = zext <2 x i8> [[TMP63]] to <2 x i32>
+; CHECK-NEXT:    [[TMP81:%.*]] = sub <2 x i32> [[TMP48]], [[TMP76]]
+; CHECK-NEXT:    [[TMP167:%.*]] = shl <2 x i32> [[TMP81]], splat (i32 16)
+; CHECK-NEXT:    [[TMP75:%.*]] = add <2 x i32> [[TMP167]], [[TMP59]]
+; CHECK-NEXT:    [[ARRAYIDX20_3:%.*]] = getelementptr i8, ptr null, i64 2
+; CHECK-NEXT:    [[ARRAYIDX22_3:%.*]] = getelementptr i8, ptr null, i64 2
+; CHECK-NEXT:    [[ARRAYIDX27_3:%.*]] = getelementptr i8, ptr null, i64 6
+; CHECK-NEXT:    [[TMP64:%.*]] = load <2 x i8>, ptr [[ARRAYIDX20_3]], align 1
+; CHECK-NEXT:    [[TMP79:%.*]] = zext <2 x i8> [[TMP64]] to <2 x i32>
+; CHECK-NEXT:    [[TMP82:%.*]] = load <2 x i8>, ptr [[ARRAYIDX22_3]], align 1
+; CHECK-NEXT:    [[TMP91:%.*]] = zext <2 x i8> [[TMP82]] to <2 x i32>
+; CHECK-NEXT:    [[TMP65:%.*]] = sub <2 x i32> [[TMP79]], [[TMP91]]
+; CHECK-NEXT:    [[TMP170:%.*]] = call <2 x i8> @llvm.masked.gather.v2i8.v2p0(<2 x ptr> zeroinitializer, i32 1, <2 x i1> splat (i1 true), <2 x i8> poison)
+; CHECK-NEXT:    [[TMP171:%.*]] = zext <2 x i8> [[TMP170]] to <2 x i32>
+; CHECK-NEXT:    [[TMP172:%.*]] = load <2 x i8>, ptr [[ARRAYIDX27_3]], align 1
+; CHECK-NEXT:    [[TMP173:%.*]] = zext <2 x i8> [[TMP172]] to <2 x i32>
+; CHECK-NEXT:    [[TMP66:%.*]] = sub <2 x i32> [[TMP171]], [[TMP173]]
+; CHECK-NEXT:    [[TMP67:%.*]] = shl <2 x i32> [[TMP66]], splat (i32 16)
+; CHECK-NEXT:    [[TMP69:%.*]] = add <2 x i32> [[TMP67]], [[TMP65]]
+; CHECK-NEXT:    [[TMP176:%.*]] = extractelement <2 x i32> [[TMP75]], i32 0
+; CHECK-NEXT:    [[TMP197:%.*]] = extractelement <2 x i32> [[TMP75]], i32 1
+; CHECK-NEXT:    [[SUB59:%.*]] = add i32 [[TMP197]], [[TMP176]]
+; CHECK-NEXT:    [[SUB45_3:%.*]] = sub i32 [[TMP176]], [[TMP197]]
+; CHECK-NEXT:    [[ADD112_2:%.*]] = extractelement <2 x i32> [[TMP69]], i32 0
+; CHECK-NEXT:    [[XOR_I63_2:%.*]] = extractelement <2 x i32> [[TMP69]], i32 1
+; CHECK-NEXT:    [[SUB59_1:%.*]] = add i32 [[XOR_I63_2]], [[ADD112_2]]
+; CHECK-NEXT:    [[SUB47_3:%.*]] = sub i32 [[ADD112_2]], [[XOR_I63_2]]
+; CHECK-NEXT:    [[ADD94:%.*]] = add i32 [[SUB59_1]], [[SUB59]]
+; CHECK-NEXT:    [[TMP70:%.*]] = shufflevector <2 x i32> [[TMP34]], <2 x i32> poison, <2 x i32> <i32 poison, i32 0>
+; CHECK-NEXT:    [[TMP71:%.*]] = insertelement <2 x i32> [[TMP70]], i32 [[SUB59]], i32 0
+; CHECK-NEXT:    [[TMP72:%.*]] = insertelement <2 x i32> [[TMP34]], i32 [[SUB59_1]], i32 0
+; CHECK-NEXT:    [[TMP222:%.*]] = sub <2 x i32> [[TMP71]], [[TMP72]]
+; CHECK-NEXT:    [[ADD55_3:%.*]] = add i32 [[SUB47_3]], [[SUB45_3]]
+; CHECK-NEXT:    [[TMP74:%.*]] = shufflevector <2 x i32> [[TMP44]], <2 x i32> poison, <2 x i32> <i32 poison, i32 0>
+; CHECK-NEXT:    [[TMP78:%.*]] = insertelement <2 x i32> [[TMP74]], i32 [[SUB45_3]], i32 0
+; CHECK-NEXT:    [[TMP80:%.*]] = insertelement <2 x i32> [[TMP44]], i32 [[SUB47_3]], i32 0
+; CHECK-NEXT:    [[TMP85:%.*]] = sub <2 x i32> [[TMP78]], [[TMP80]]
+; CHECK-NEXT:    [[ADD95:%.*]] = add i32 [[ADD94]], [[ADD48_2]]
+; CHECK-NEXT:    [[SUB86_3:%.*]] = sub i32 [[ADD48_2]], [[ADD94]]
+; CHECK-NEXT:    [[SHR_I_1:%.*]] = lshr i32 [[TMP77]], 15
+; CHECK-NEXT:    [[AND_I_1:%.*]] = and i32 [[SHR_I_1]], 65537
+; CHECK-NEXT:    [[MUL_I_1:%.*]] = mul i32 [[AND_I_1]], 65535
+; CHECK-NEXT:    [[SHR_I49_1:%.*]] = lshr i32 [[CONV_2]], 15
+; CHECK-NEXT:    [[AND_I50_1:%.*]] = and i32 [[SHR_I49_1]], 65537
+; CHECK-NEXT:    [[MUL_I51_1:%.*]] = mul i32 [[AND_I50_1]], 65535
+; CHECK-NEXT:    [[TMP86:%.*]] = extractelement <2 x i32> [[TMP222]], i32 0
+; CHECK-NEXT:    [[TMP87:%.*]] = extractelement <2 x i32> [[TMP222]], i32 1
+; CHECK-NEXT:    [[ADD94_3:%.*]] = add i32 [[TMP86]], [[TMP87]]
+; CHECK-NEXT:    [[ADD112_1:%.*]] = sub i32 [[TMP87]], [[TMP86]]
+; CHECK-NEXT:    [[SHR_I49_2:%.*]] = lshr i32 [[CONV_1]], 15
+; CHECK-NEXT:    [[AND_I50_2:%.*]] = and i32 [[SHR_I49_2]], 65537
+; CHECK-NEXT:    [[MUL_I51_2:%.*]] = mul i32 [[AND_I50_2]], 65535
+; CHECK-NEXT:    [[TMP88:%.*]] = extractelement <2 x i32> [[TMP85]], i32 0
+; CHECK-NEXT:    [[TMP89:%.*]] = extractelement <2 x i32> [[TMP85]], i32 1
+; CHECK-NEXT:    [[ADD94_4:%.*]] = add i32 [[TMP88]], [[TMP89]]
+; CHECK-NEXT:    [[SUB102_3:%.*]] = sub i32 [[TMP89]], [[TMP88]]
+; CHECK-NEXT:    [[SHR_I49_3:%.*]] = lshr i32 [[CONV1]], 15
+; CHECK-NEXT:    [[AND_I50_3:%.*]] = and i32 [[SHR_I49_3]], 65537
+; CHECK-NEXT:    [[MUL_I51_3:%.*]] = mul i32 [[AND_I50_3]], 65535
+; CHECK-NEXT:    [[TMP90:%.*]] = load <2 x i8>, ptr [[ARRAYIDX8]], align 1
+; CHECK-NEXT:    [[TMP102:%.*]] = zext <2 x i8> [[TMP90]] to <2 x i32>
 ; CHECK-NEXT:    [[TMP92:%.*]] = load <4 x i8>, ptr [[PIX2]], align 1
+; CHECK-NEXT:    [[TMP93:%.*]] = shufflevector <4 x i8> [[TMP92]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP94:%.*]] = zext <2 x i8> [[TMP93]] to <2 x i32>
 ; CHECK-NEXT:    [[TMP95:%.*]] = load <4 x i8>, ptr [[TMP1]], align 1
+; CHECK-NEXT:    [[TMP96:%.*]] = shufflevector <4 x i8> [[TMP95]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP97:%.*]] = zext <2 x i8> [[TMP96]] to <2 x i32>
 ; CHECK-NEXT:    [[TMP98:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5]], align 1
-; CHECK-NEXT:    [[TMP6:%.*]] = load <4 x i8>, ptr [[ADD_PTR3]], align 1
+; CHECK-NEXT:    [[TMP99:%.*]] = shufflevector <4 x i8> [[TMP98]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP100:%.*]] = zext <2 x i8> [[TMP99]] to <2 x i32>
+; CHECK-NEXT:    [[TMP101:%.*]] = sub <2 x i32> [[TMP97]], [[TMP100]]
+; CHECK-NEXT:    [[TMP224:%.*]] = shl <2 x i32> [[TMP101]], splat (i32 16)
+; CHECK-NEXT:    [[TMP103:%.*]] = shufflevector <4 x i8> [[TMP92]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP104:%.*]] = zext <2 x i8> [[TMP103]] to <2 x i32>
+; CHECK-NEXT:    [[TMP105:%.*]] = shufflevector <4 x i8> [[TMP95]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP106:%.*]] = zext <2 x i8> [[TMP105]] to <2 x i32>
+; CHECK-NEXT:    [[TMP107:%.*]] = shufflevector <4 x i8> [[TMP98]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP108:%.*]] = zext <2 x i8> [[TMP107]] to <2 x i32>
+; CHECK-NEXT:    [[TMP109:%.*]] = sub <2 x i32> [[TMP106]], [[TMP108]]
+; CHECK-NEXT:    [[TMP110:%.*]] = shl <2 x i32> [[TMP109]], splat (i32 16)
+; CHECK-NEXT:    [[TMP111:%.*]] = insertelement <2 x i32> [[TMP102]], i32 [[CONV33]], i32 1
+; CHECK-NEXT:    [[TMP112:%.*]] = sub <2 x i32> [[TMP111]], [[TMP104]]
+; CHECK-NEXT:    [[TMP113:%.*]] = add <2 x i32> [[TMP110]], [[TMP112]]
+; CHECK-NEXT:    [[TMP114:%.*]] = insertelement <2 x i32> [[TMP102]], i32 [[CONV1]], i32 0
+; CHECK-NEXT:    [[TMP115:%.*]] = sub <2 x i32> [[TMP114]], [[TMP94]]
+; CHECK-NEXT:    [[TMP116:%.*]] = add <2 x i32> [[TMP224]], [[TMP115]]
+; CHECK-NEXT:    [[TMP117:%.*]] = shufflevector <2 x i32> [[TMP113]], <2 x i32> [[TMP116]], <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP126:%.*]] = add <2 x i32> [[TMP113]], [[TMP116]]
+; CHECK-NEXT:    [[TMP119:%.*]] = sub <2 x i32> [[TMP116]], [[TMP113]]
+; CHECK-NEXT:    [[TMP120:%.*]] = extractelement <2 x i32> [[TMP126]], i32 0
+; CHECK-NEXT:    [[TMP127:%.*]] = extractelement <2 x i32> [[TMP126]], i32 1
+; CHECK-NEXT:    [[ADD48:%.*]] = add i32 [[TMP127]], [[TMP120]]
+; CHECK-NEXT:    [[TMP166:%.*]] = sub i32 [[TMP120]], [[TMP127]]
+; CHECK-NEXT:    [[TMP128:%.*]] = extractelement <2 x i32> [[TMP119]], i32 0
+; CHECK-NEXT:    [[TMP129:%.*]] = extractelement <2 x i32> [[TMP119]], i32 1
+; CHECK-NEXT:    [[ADD55:%.*]] = add i32 [[TMP129]], [[TMP128]]
+; CHECK-NEXT:    [[SUB60:%.*]] = sub i32 [[TMP128]], [[TMP129]]
+; CHECK-NEXT:    [[SHR_I59:%.*]] = lshr i32 [[TMP127]], 15
+; CHECK-NEXT:    [[AND_I60:%.*]] = and i32 [[SHR_I59]], 65537
+; CHECK-NEXT:    [[MUL_I61:%.*]] = mul i32 [[AND_I60]], 65535
+; CHECK-NEXT:    [[SHR_I59_1:%.*]] = lshr i32 [[TMP129]], 15
+; CHECK-NEXT:    [[AND_I60_1:%.*]] = and i32 [[SHR_I59_1]], 65537
+; CHECK-NEXT:    [[MUL_I61_1:%.*]] = mul i32 [[AND_I60_1]], 65535
+; CHECK-NEXT:    [[TMP130:%.*]] = load <2 x i8>, ptr [[ARRAYIDX8_1]], align 1
+; CHECK-NEXT:    [[TMP131:%.*]] = zext <2 x i8> [[TMP130]] to <2 x i32>
 ; CHECK-NEXT:    [[TMP132:%.*]] = load <4 x i8>, ptr [[ADD_PTR644]], align 1
+; CHECK-NEXT:    [[TMP133:%.*]] = shufflevector <4 x i8> [[TMP132]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP134:%.*]] = zext <2 x i8> [[TMP133]] to <2 x i32>
 ; CHECK-NEXT:    [[TMP135:%.*]] = load <4 x i8>, ptr [[ARRAYIDX3_1]], align 1
+; CHECK-NEXT:    [[TMP136:%.*]] = shufflevector <4 x i8> [[TMP135]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP137:%.*]] = zext <2 x i8> [[TMP136]] to <2 x i32>
 ; CHECK-NEXT:    [[TMP138:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5_1]], align 1
-; CHECK-NEXT:    [[TMP10:%.*]] = load <4 x i8>, ptr [[ADD_PTR_1]], align 1
-; CHECK-NEXT:    [[TMP11:%.*]] = load <4 x i8>, ptr [[ADD_PTR64_1]], align 1
-; CHECK-NEXT:    [[TMP12:%.*]] = load <4 x i8>, ptr [[ARRAYIDX3_2]], align 1
-; CHECK-NEXT:    [[TMP13:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5_2]], align 1
-; CHECK-NEXT:    [[TMP14:%.*]] = load <4 x i8>, ptr null, align 1
-; CHECK-NEXT:    [[TMP15:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> poison, <4 x i8> [[TMP10]], i64 0)
-; CHECK-NEXT:    [[TMP16:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP15]], <4 x i8> [[TMP14]], i64 4)
-; CHECK-NEXT:    [[TMP17:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP16]], <4 x i8> [[TMP2]], i64 8)
-; CHECK-NEXT:    [[TMP18:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP17]], <4 x i8> [[TMP6]], i64 12)
-; CHECK-NEXT:    [[TMP19:%.*]] = zext <16 x i8> [[TMP18]] to <16 x i32>
-; CHECK-NEXT:    [[TMP20:%.*]] = load <4 x i8>, ptr null, align 1
-; CHECK-NEXT:    [[TMP21:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> poison, <4 x i8> [[TMP11]], i64 0)
-; CHECK-NEXT:    [[TMP22:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP21]], <4 x i8> [[TMP20]], i64 4)
-; CHECK-NEXT:    [[TMP23:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP22]], <4 x i8> [[TMP92]], i64 8)
-; CHECK-NEXT:    [[TMP24:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP23]], <4 x i8> [[TMP132]], i64 12)
-; CHECK-NEXT:    [[TMP25:%.*]] = zext <16 x i8> [[TMP24]] to <16 x i32>
-; CHECK-NEXT:    [[TMP26:%.*]] = sub <16 x i32> [[TMP19]], [[TMP25]]
-; CHECK-NEXT:    [[TMP27:%.*]] = shufflevector <16 x i32> [[TMP26]], <16 x i32> poison, <16 x i32> <i32 3, i32 7, i32 15, i32 11, i32 2, i32 6, i32 14, i32 10, i32 1, i32 5, i32 13, i32 9, i32 0, i32 4, i32 12, i32 8>
-; CHECK-NEXT:    [[TMP28:%.*]] = call <2 x i8> @llvm.experimental.vp.strided.load.v2i8.p0.i64(ptr align 1 null, i64 4, <2 x i1> splat (i1 true), i32 2)
-; CHECK-NEXT:    [[TMP29:%.*]] = shufflevector <2 x i8> [[TMP28]], <2 x i8> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison>
-; CHECK-NEXT:    [[TMP30:%.*]] = shufflevector <4 x i8> [[TMP12]], <4 x i8> [[TMP29]], <16 x i32> <i32 3, i32 4, i32 poison, i32 poison, i32 2, i32 poison, i32 poison, i32 poison, i32 1, i32 poison, i32 poison, i32 poison, i32 0, i32 5, i32 poison, i32 poison>
-; CHECK-NEXT:    [[TMP31:%.*]] = shufflevector <4 x i8> [[TMP135]], <4 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; CHECK-NEXT:    [[TMP32:%.*]] = shufflevector <16 x i8> [[TMP30]], <16 x i8> [[TMP31]], <16 x i32> <i32 0, i32 1, i32 19, i32 poison, i32 4, i32 poison, i32 18, i32 poison, i32 8, i32 poison, i32 17, i32 poison, i32 12, i32 13, i32 16, i32 poison>
-; CHECK-NEXT:    [[TMP33:%.*]] = shufflevector <4 x i8> [[TMP95]], <4 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; CHECK-NEXT:    [[TMP34:%.*]] = shufflevector <16 x i8> [[TMP32]], <16 x i8> [[TMP33]], <16 x i32> <i32 0, i32 1, i32 2, i32 19, i32 4, i32 poison, i32 6, i32 18, i32 8, i32 poison, i32 10, i32 17, i32 12, i32 13, i32 14, i32 16>
-; CHECK-NEXT:    [[TMP35:%.*]] = insertelement <16 x i8> [[TMP34]], i8 [[TMP3]], i32 5
-; CHECK-NEXT:    [[TMP36:%.*]] = insertelement <16 x i8> [[TMP35]], i8 [[TMP52]], i32 9
-; CHECK-NEXT:    [[TMP37:%.*]] = zext <16 x i8> [[TMP36]] to <16 x i32>
-; CHECK-NEXT:    [[TMP38:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5_3]], align 1
-; CHECK-NEXT:    [[TMP39:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> poison, <4 x i8> [[TMP13]], i64 0)
-; CHECK-NEXT:    [[TMP40:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP39]], <4 x i8> [[TMP38]], i64 4)
-; CHECK-NEXT:    [[TMP41:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP40]], <4 x i8> [[TMP98]], i64 8)
-; CHECK-NEXT:    [[TMP42:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP41]], <4 x i8> [[TMP138]], i64 12)
-; CHECK-NEXT:    [[TMP43:%.*]] = zext <16 x i8> [[TMP42]] to <16 x i32>
-; CHECK-NEXT:    [[TMP44:%.*]] = shufflevector <16 x i32> [[TMP43]], <16 x i32> poison, <16 x i32> <i32 3, i32 7, i32 15, i32 11, i32 2, i32 6, i32 14, i32 10, i32 1, i32 5, i32 13, i32 9, i32 0, i32 4, i32 12, i32 8>
-; CHECK-NEXT:    [[TMP45:%.*]] = sub <16 x i32> [[TMP37]], [[TMP44]]
-; CHECK-NEXT:    [[TMP46:%.*]] = shl <16 x i32> [[TMP45]], splat (i32 16)
-; CHECK-NEXT:    [[TMP47:%.*]] = add <16 x i32> [[TMP46]], [[TMP27]]
-; CHECK-NEXT:    [[TMP48:%.*]] = shufflevector <16 x i32> [[TMP47]], <16 x i32> poison, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 12, i32 13, i32 14, i32 15, i32 8, i32 9, i32 10, i32 11>
-; CHECK-NEXT:    [[TMP49:%.*]] = add <16 x i32> [[TMP47]], [[TMP48]]
-; CHECK-NEXT:    [[TMP50:%.*]] = sub <16 x i32> [[TMP47]], [[TMP48]]
-; CHECK-NEXT:    [[TMP51:%.*]] = shufflevector <16 x i32> [[TMP49]], <16 x i32> [[TMP50]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 28, i32 29, i32 30, i32 31>
-; CHECK-NEXT:    [[TMP70:%.*]] = shufflevector <16 x i32> [[TMP51]], <16 x i32> poison, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEXT:    [[TMP53:%.*]] = add <16 x i32> [[TMP51]], [[TMP70]]
-; CHECK-NEXT:    [[TMP54:%.*]] = sub <16 x i32> [[TMP51]], [[TMP70]]
-; CHECK-NEXT:    [[TMP55:%.*]] = shufflevector <16 x i32> [[TMP53]], <16 x i32> [[TMP54]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
-; CHECK-NEXT:    [[TMP56:%.*]] = shufflevector <16 x i32> [[TMP55]], <16 x i32> poison, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
-; CHECK-NEXT:    [[TMP57:%.*]] = sub <16 x i32> [[TMP55]], [[TMP56]]
-; CHECK-NEXT:    [[TMP58:%.*]] = add <16 x i32> [[TMP55]], [[TMP56]]
-; CHECK-NEXT:    [[TMP59:%.*]] = shufflevector <16 x i32> [[TMP57]], <16 x i32> [[TMP58]], <16 x i32> <i32 0, i32 17, i32 18, i32 3, i32 4, i32 21, i32 22, i32 7, i32 8, i32 25, i32 26, i32 11, i32 12, i32 29, i32 30, i32 15>
-; CHECK-NEXT:    [[TMP60:%.*]] = shufflevector <16 x i32> [[TMP59]], <16 x i32> poison, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4, i32 11, i32 10, i32 9, i32 8, i32 15, i32 14, i32 13, i32 12>
-; CHECK-NEXT:    [[TMP61:%.*]] = add <16 x i32> [[TMP59]], [[TMP60]]
-; CHECK-NEXT:    [[TMP62:%.*]] = sub <16 x i32> [[TMP59]], [[TMP60]]
-; CHECK-NEXT:    [[TMP63:%.*]] = shufflevector <16 x i32> [[TMP61]], <16 x i32> [[TMP62]], <16 x i32> <i32 0, i32 1, i32 18, i32 19, i32 4, i32 5, i32 22, i32 23, i32 8, i32 9, i32 26, i32 27, i32 12, i32 13, i32 30, i32 31>
-; CHECK-NEXT:    [[TMP64:%.*]] = shufflevector <16 x i32> [[TMP51]], <16 x i32> [[TMP19]], <16 x i32> <i32 0, i32 20, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 28, i32 29, i32 30, i32 11, i32 24, i32 25, i32 26, i32 27>
-; CHECK-NEXT:    [[TMP65:%.*]] = lshr <16 x i32> [[TMP64]], splat (i32 15)
-; CHECK-NEXT:    [[TMP66:%.*]] = and <16 x i32> [[TMP65]], splat (i32 65537)
-; CHECK-NEXT:    [[TMP67:%.*]] = mul <16 x i32> [[TMP66]], splat (i32 65535)
-; CHECK-NEXT:    [[TMP68:%.*]] = add <16 x i32> [[TMP67]], [[TMP63]]
-; CHECK-NEXT:    [[TMP69:%.*]] = xor <16 x i32> [[TMP68]], [[TMP64]]
-; CHECK-NEXT:    [[ADD113_3:%.*]] = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> [[TMP69]])
+; CHECK-NEXT:    [[TMP139:%.*]] = shufflevector <4 x i8> [[TMP138]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[TMP140:%.*]] = zext <2 x i8> [[TMP139]] to <2 x i32>
+; CHECK-NEXT:    [[TMP141:%.*]] = sub <2 x i32> [[TMP137]], [[TMP140]]
+; CHECK-NEXT:    [[TMP142:%.*]] = shl <2 x i32> [[TMP141]], splat (i32 16)
+; CHECK-NEXT:    [[TMP143:%.*]] = shufflevector <4 x i8> [[TMP132]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP144:%.*]] = zext <2 x i8> [[TMP143]] to <2 x i32>
+; CHECK-NEXT:    [[TMP145:%.*]] = shufflevector <4 x i8> [[TMP135]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP146:%.*]] = zext <2 x i8> [[TMP145]] to <2 x i32>
+; CHECK-NEXT:    [[TMP147:%.*]] = shufflevector <4 x i8> [[TMP138]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT:    [[TMP148:%.*]] = zext <2 x i8> [[TMP147]] to <2 x i32>
+; CHECK-NEXT:    [[TMP149:%.*]] = sub <2 x i32> [[TMP146]], [[TMP148]]
+; CHECK-NEXT:    [[TMP150:%.*]] = shl <2 x i32> [[TMP149]], splat (i32 16)
+; CHECK-NEXT:    [[TMP151:%.*]] = insertelement <2 x i32> [[TMP131]], i32 [[CONV33_1]], i32 1
+; CHECK-NEXT:    [[TMP225:%.*]] = sub <2 x i32> [[TMP151]], [[TMP144]]
+; CHECK-NEXT:    [[TMP153:%.*]] = add <2 x i32> [[TMP150]], [[TMP225]]
+; CHECK-NEXT:    [[TMP154:%.*]] = insertelement <2 x i32> [[TMP131]], i32 [[CONV_1]], i32 0
+; CHECK-NEXT:    [[TMP155:%.*]] = sub <2 x i32> [[TMP154]], [[TMP134]]
+; CHECK-NEXT:    [[TMP156:%.*]] = add <2 x i32> [[TMP142]], [[TMP155]]
+; CHECK-NEXT:    [[TMP157:%.*]] = add <2 x i32> [[TMP153]], [[TMP156]]
+; CHECK-NEXT:    [[TMP158:%.*]] = sub <2 x i32> [[TMP156]], [[TMP153]]
+; CHECK-NEXT:    [[TMP159:%.*]] = extractelement <2 x i32> [[TMP157]], i32 0
+; CHECK-NEXT:    [[TMP160:%.*]] = extractelement <2 x i32> [[TMP157]], i32 1
+; CHECK-NEXT:    [[ADD48_1:%.*]] = add i32 [[TMP160]], [[TMP159]]
+; CHECK-NEXT:    [[SUB51_1:%.*]] = sub i32 [[TMP159]], [[TMP160]]
+; CHECK-NEXT:    [[TMP161:%.*]] = extractelement <2 x i32> [[TMP158]], i32 0
+; CHECK-NEXT:    [[TMP162:%.*]] = extractelement <2 x i32> [[TMP158]], i32 1
+; CHECK-NEXT:    [[ADD55_1:%.*]] = add i32 [[TMP162]], [[TMP161]]
+; CHECK-NEXT:    [[SUB59_2:%.*]] = sub i32 [[TMP161]], [[TMP162]]
+; CHECK-NEXT:    [[SHR_I54:%.*]] = lshr i32 [[TMP160]], 15
+; CHECK-NEXT:    [[AND_I55:%.*]] = and i32 [[SHR_I54]], 65537
+; CHECK-NEXT:    [[MUL_I56:%.*]] = mul i32 [[AND_I55]], 65535
+; CHECK-NEXT:    [[SHR_I54_1:%.*]] = lshr i32 [[TMP162]], 15
+; CHECK-NEXT:    [[AND_I55_1:%.*]] = and i32 [[SHR_I54_1]], 65537
+; CHECK-NEXT:    [[MUL_I56_1:%.*]] = mul i32 [[AND_I55_1]], 65535
+; CHECK-NEXT:    [[TMP163:%.*]] = lshr <2 x i32> [[TMP131]], splat (i32 15)
+; CHECK-NEXT:    [[TMP164:%.*]] = and <2 x i32> [[TMP163]], splat (i32 65537)
+; CHECK-NEXT:    [[TMP165:%.*]] = mul <2 x i32> [[TMP164]], splat (i32 65535)
+; CHECK-NEXT:    [[ADD78:%.*]] = add i32 [[ADD48_1]], [[ADD48]]
+; CHECK-NEXT:    [[SUB86:%.*]] = sub i32 [[ADD48]], [[ADD48_1]]
+; CHECK-NEXT:    [[ADD103:%.*]] = add i32 [[ADD95]], [[ADD78]]
+; CHECK-NEXT:    [[SUB104:%.*]] = sub i32 [[ADD78]], [[ADD95]]
+; CHECK-NEXT:    [[ADD105:%.*]] = add i32 [[SUB86_3]], [[SUB86]]
+; CHECK-NEXT:    [[SUB106:%.*]] = sub i32 [[SUB86]], [[SUB86_3]]
+; CHECK-NEXT:    [[ADD_I:%.*]] = add i32 [[MUL_I_1]], [[ADD103]]
+; CHECK-NEXT:    [[XOR_I:%.*]] = xor i32 [[ADD_I]], [[TMP77]]
+; CHECK-NEXT:    [[ADD_I52:%.*]] = add i32 [[MUL_I51_1]], [[ADD105]]
+; CHECK-NEXT:    [[XOR_I53:%.*]] = xor i32 [[ADD_I52]], [[CONV_2]]
+; CHECK-NEXT:    [[ADD_I57:%.*]] = add i32 [[MUL_I56]], [[SUB104]]
+; CHECK-NEXT:    [[XOR_I58:%.*]] = xor i32 [[ADD_I57]], [[TMP160]]
+; CHECK-NEXT:    [[ADD_I62:%.*]] = add i32 [[MUL_I61]], [[SUB106]]
+; CHECK-NEXT:    [[XOR_I63:%.*]] = xor i32 [[ADD_I62]], [[TMP127]]
+; CHECK-NEXT:    [[ADD110:%.*]] = add i32 [[XOR_I53]], [[XOR_I]]
+; CHECK-NEXT:    [[ADD112:%.*]] = add i32 [[ADD110]], [[XOR_I58]]
+; CHECK-NEXT:    [[ADD105_3:%.*]] = add i32 [[ADD112]], [[XOR_I63]]
+; CHECK-NEXT:    [[TMP169:%.*]] = load <2 x i8>, ptr [[ADD_PTR_1]], align 1
+; CHECK-NEXT:    [[TMP181:%.*]] = zext <2 x i8> [[TMP169]] to <2 x i32>
+; CHECK-NEXT:    [[TMP152:%.*]] = insertelement <2 x i32> poison, i32 [[ADD55_2]], i32 0
+; CHECK-NEXT:    [[TMP182:%.*]] = shufflevector <2 x i32> [[TMP152]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP183:%.*]] = insertelement <2 x i32> poison, i32 [[ADD55_3]], i32 0
+; CHECK-NEXT:    [[TMP184:%.*]] = shufflevector <2 x i32> [[TMP183]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP191:%.*]] = sub <2 x i32> [[TMP182]], [[TMP184]]
+; CHECK-NEXT:    [[TMP192:%.*]] = add <2 x i32> [[TMP182]], [[TMP184]]
+; CHECK-NEXT:    [[TMP194:%.*]] = shufflevector <2 x i32> [[TMP191]], <2 x i32> [[TMP192]], <2 x i32> <i32 0, i32 3>
+; CHECK-NEXT:    [[TMP195:%.*]] = lshr <2 x i32> [[TMP181]], splat (i32 15)
+; CHECK-NEXT:    [[TMP196:%.*]] = and <2 x i32> [[TMP195]], splat (i32 65537)
+; CHECK-NEXT:    [[TMP198:%.*]] = mul <2 x i32> [[TMP196]], splat (i32 65535)
+; CHECK-NEXT:    [[TMP202:%.*]] = insertelement <2 x i32> poison, i32 [[ADD55]], i32 0
+; CHECK-NEXT:    [[TMP203:%.*]] = shufflevector <2 x i32> [[TMP202]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP205:%.*]] = insertelement <2 x i32> poison, i32 [[ADD55_1]], i32 0
+; CHECK-NEXT:    [[TMP206:%.*]] = shufflevector <2 x i32> [[TMP205]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP207:%.*]] = sub <2 x i32> [[TMP203]], [[TMP206]]
+; CHECK-NEXT:    [[TMP210:%.*]] = add <2 x i32> [[TMP203]], [[TMP206]]
+; CHECK-NEXT:    [[TMP168:%.*]] = shufflevector <2 x i32> [[TMP207]], <2 x i32> [[TMP210]], <2 x i32> <i32 0, i32 3>
+; CHECK-NEXT:    [[ADD94_1:%.*]] = extractelement <2 x i32> [[TMP194]], i32 1
+; CHECK-NEXT:    [[ADD78_1:%.*]] = extractelement <2 x i32> [[TMP168]], i32 1
+; CHECK-NEXT:    [[SUB104_1:%.*]] = sub i32 [[ADD78_1]], [[ADD94_1]]
+; CHECK-NEXT:    [[TMP220:%.*]] = add <2 x i32> [[TMP194]], [[TMP168]]
+; CHECK-NEXT:    [[SUB102_1:%.*]] = extractelement <2 x i32> [[TMP194]], i32 0
+; CHECK-NEXT:    [[SUB86_1:%.*]] = extractelement <2 x i32> [[TMP168]], i32 0
+; CHECK-NEXT:    [[TMP174:%.*]] = shufflevector <2 x i32> [[TMP168]], <2 x i32> [[TMP194]], <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT:    [[SUB106_1:%.*]] = sub i32 [[SUB86_1]], [[SUB102_1]]
+; CHECK-NEXT:    [[TMP175:%.*]] = add <2 x i32> [[TMP198]], [[TMP220]]
+; CHECK-NEXT:    [[TMP221:%.*]] = xor <2 x i32> [[TMP175]], [[TMP181]]
+; CHECK-NEXT:    [[ADD_I57_1:%.*]] = add i32 [[MUL_I56_1]], [[SUB104_1]]
+; CHECK-NEXT:    [[XOR_I58_1:%.*]] = xor i32 [[ADD_I57_1]], [[TMP162]]
+; CHECK-NEXT:    [[ADD_I62_1:%.*]] = add i32 [[MUL_I61_1]], [[SUB106_1]]
+; CHECK-NEXT:    [[XOR_I63_1:%.*]] = xor i32 [[ADD_I62_1]], [[TMP129]]
+; CHECK-NEXT:    [[XOR_I53_1:%.*]] = extractelement <2 x i32> [[TMP221]], i32 0
+; CHECK-NEXT:    [[ADD108_1:%.*]] = add i32 [[XOR_I53_1]], [[ADD105_3]]
+; CHECK-NEXT:    [[XOR_I_1:%.*]] = extractelement <2 x i32> [[TMP221]], i32 1
+; CHECK-NEXT:    [[ADD110_1:%.*]] = add i32 [[ADD108_1]], [[XOR_I_1]]
+; CHECK-NEXT:    [[ADD112_5:%.*]] = add i32 [[ADD110_1]], [[XOR_I58_1]]
+; CHECK-NEXT:    [[ADD113_2:%.*]] = add i32 [[ADD112_5]], [[XOR_I63_1]]
+; CHECK-NEXT:    [[ADD78_3:%.*]] = add i32 [[SUB51_1]], [[TMP166]]
+; CHECK-NEXT:    [[TMP204:%.*]] = sub i32 [[TMP166]], [[SUB51_1]]
+; CHECK-NEXT:    [[TMP177:%.*]] = insertelement <2 x i32> poison, i32 [[ADD78_3]], i32 0
+; CHECK-NEXT:    [[TMP178:%.*]] = shufflevector <2 x i32> [[TMP177]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP179:%.*]] = insertelement <2 x i32> poison, i32 [[ADD94_3]], i32 0
+; CHECK-NEXT:    [[TMP180:%.*]] = shufflevector <2 x i32> [[TMP179]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP199:%.*]] = add <2 x i32> [[TMP178]], [[TMP180]]
+; CHECK-NEXT:    [[TMP200:%.*]] = sub <2 x i32> [[TMP178]], [[TMP180]]
+; CHECK-NEXT:    [[TMP201:%.*]] = shufflevector <2 x i32> [[TMP199]], <2 x i32> [[TMP200]], <2 x i32> <i32 0, i32 3>
+; CHECK-NEXT:    [[ADD113_1:%.*]] = add i32 [[ADD112_1]], [[TMP204]]
+; CHECK-NEXT:    [[SUB106_2:%.*]] = sub i32 [[TMP204]], [[ADD112_1]]
+; CHECK-NEXT:    [[ADD_I52_2:%.*]] = add i32 [[MUL_I51_2]], [[ADD113_1]]
+; CHECK-NEXT:    [[XOR_I53_2:%.*]] = xor i32 [[ADD_I52_2]], [[CONV_1]]
+; CHECK-NEXT:    [[TMP208:%.*]] = add <2 x i32> [[TMP165]], [[TMP201]]
+; CHECK-NEXT:    [[TMP209:%.*]] = xor <2 x i32> [[TMP208]], [[TMP131]]
+; CHECK-NEXT:    [[SHR_I59_2:%.*]] = lshr i32 [[TMP120]], 15
+; CHECK-NEXT:    [[AND_I60_2:%.*]] = and i32 [[SHR_I59_2]], 65537
+; CHECK-NEXT:    [[MUL_I61_2:%.*]] = mul i32 [[AND_I60_2]], 65535
+; CHECK-NEXT:    [[ADD_I62_2:%.*]] = add i32 [[MUL_I61_2]], [[SUB106_2]]
+; CHECK-NEXT:    [[XOR_I63_4:%.*]] = xor i32 [[ADD_I62_2]], [[TMP120]]
+; CHECK-NEXT:    [[ADD108_2:%.*]] = add i32 [[XOR_I53_2]], [[ADD113_2]]
+; CHECK-NEXT:    [[TMP211:%.*]] = extractelement <2 x i32> [[TMP209]], i32 0
+; CHECK-NEXT:    [[ADD110_2:%.*]] = add i32 [[ADD108_2]], [[TMP211]]
+; CHECK-NEXT:    [[TMP212:%.*]] = extractelement <2 x i32> [[TMP209]], i32 1
+; CHECK-NEXT:    [[ADD112_4:%.*]] = add i32 [[ADD110_2]], [[TMP212]]
+; CHECK-NEXT:    [[ADD113_4:%.*]] = add i32 [[ADD112_4]], [[XOR_I63_4]]
+; CHECK-NEXT:    [[ADD78_4:%.*]] = add i32 [[SUB59_2]], [[SUB60]]
+; CHECK-NEXT:    [[SUB86_4:%.*]] = sub i32 [[SUB60]], [[SUB59_2]]
+; CHECK-NEXT:    [[TMP213:%.*]] = insertelement <2 x i32> poison, i32 [[ADD78_4]], i32 0
+; CHECK-NEXT:    [[TMP214:%.*]] = shufflevector <2 x i32> [[TMP213]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP215:%.*]] = insertelement <2 x i32> poison, i32 [[ADD94_4]], i32 0
+; CHECK-NEXT:    [[TMP216:%.*]] = shufflevector <2 x i32> [[TMP215]], <2 x i32> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP217:%.*]] = add <2 x i32> [[TMP214]], [[TMP216]]
+; CHECK-NEXT:    [[TMP218:%.*]] = sub <2 x i32> [[TMP214]], [[TMP216]]
+; CHECK-NEXT:    [[TMP219:%.*]] = shufflevector <2 x i32> [[TMP217]], <2 x i32> [[TMP218]], <2 x i32> <i32 0, i32 3>
+; CHECK-NEXT:    [[ADD105_4:%.*]] = add i32 [[SUB102_3]], [[SUB86_4]]
+; CHECK-NEXT:    [[SUB106_3:%.*]] = sub i32 [[SUB86_4]], [[SUB102_3]]
+; CHECK-NEXT:    [[ADD_I52_4:%.*]] = add i32 [[MUL_I51_3]], [[ADD105_4]]
+; CHECK-NEXT:    [[XOR_I53_3:%.*]] = xor i32 [[ADD_I52_4]], [[CONV1]]
+; CHECK-NEXT:    [[TMP185:%.*]] = lshr <2 x i32> [[TMP102]], splat (i32 15)
+; CHECK-NEXT:    [[TMP193:%.*]] = and <2 x i32> [[TMP185]], splat (i32 65537)
+; CHECK-NEXT:    [[TMP186:%.*]] = mul <2 x i32> [[TMP193]], splat (i32 65535)
+; CHECK-NEXT:    [[TMP187:%.*]] = add <2 x i32> [[TMP186]], [[TMP219]]
+; CHECK-NEXT:    [[TMP188:%.*]] = xor <2 x i32> [[TMP187]], [[TMP102]]
+; CHECK-NEXT:    [[SHR_I59_3:%.*]] = lshr i32 [[CONV33]], 15
+; CHECK-NEXT:    [[AND_I60_3:%.*]] = and i32 [[SHR_I59_3]], 65537
+; CHECK-NEXT:    [[MUL_I61_3:%.*]] = mul i32 [[AND_I60_3]], 65535
+; CHECK-NEXT:    [[ADD_I62_3:%.*]] = add i32 [[MUL_I61_3]], [[SUB106_3]]
+; CHECK-NEXT:    [[XOR_I63_3:%.*]] = xor i32 [[ADD_I62_3]], [[CONV33]]
+; CHECK-NEXT:    [[ADD108_3:%.*]] = add i32 [[XOR_I53_3]], [[ADD113_4]]
+; CHECK-NEXT:    [[TMP189:%.*]] = extractelement <2 x i32> [[TMP188]], i32 0
+; CHECK-NEXT:    [[ADD110_3:%.*]] = add i32 [[ADD108_3]], [[TMP189]]
+; CHECK-NEXT:    [[TMP190:%.*]] = extractelement <2 x i32> [[TMP188]], i32 1
+; CHECK-NEXT:    [[ADD112_3:%.*]] = add i32 [[ADD110_3]], [[TMP190]]
+; CHECK-NEXT:    [[ADD113_3:%.*]] = add i32 [[ADD112_3]], [[XOR_I63_3]]
 ; CHECK-NEXT:    ret i32 [[ADD113_3]]
 ;
 ; THR15-LABEL: define i32 @test(
 ; THR15-SAME: ptr [[PIX1:%.*]], ptr [[PIX2:%.*]], i64 [[IDX_EXT:%.*]], i64 [[IDX_EXT63:%.*]], ptr [[ADD_PTR:%.*]], ptr [[ADD_PTR64:%.*]]) #[[ATTR0:[0-9]+]] {
 ; THR15-NEXT:  entry:
+; THR15-NEXT:    [[TMP0:%.*]] = load i8, ptr [[PIX1]], align 1
+; THR15-NEXT:    [[CONV:%.*]] = zext i8 [[TMP0]] to i32
 ; THR15-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr i8, ptr [[PIX1]], i64 4
 ; THR15-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr i8, ptr [[PIX2]], i64 4
+; THR15-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr i8, ptr [[PIX1]], i64 1
+; THR15-NEXT:    [[ARRAYIDX32:%.*]] = getelementptr i8, ptr [[PIX1]], i64 3
+; THR15-NEXT:    [[TMP1:%.*]] = load i8, ptr [[ARRAYIDX32]], align 1
+; THR15-NEXT:    [[CONV33:%.*]] = zext i8 [[TMP1]] to i32
 ; THR15-NEXT:    [[ADD_PTR3:%.*]] = getelementptr i8, ptr [[PIX1]], i64 [[IDX_EXT]]
 ; THR15-NEXT:    [[ADD_PTR644:%.*]] = getelementptr i8, ptr [[PIX2]], i64 [[IDX_EXT63]]
+; THR15-NEXT:    [[TMP2:%.*]] = load i8, ptr [[ADD_PTR3]], align 1
+; THR15-NEXT:    [[CONV_1:%.*]] = zext i8 [[TMP2]] to i32
 ; THR15-NEXT:    [[ARRAYIDX3_1:%.*]] = getelementptr i8, ptr [[ADD_PTR3]], i64 4
 ; THR15-NEXT:    [[ARRAYIDX5_1:%.*]] = getelementptr i8, ptr [[ADD_PTR644]], i64 4
+; THR15-NEXT:    [[ARRAYIDX8_1:%.*]] = getelementptr i8, ptr [[ADD_PTR3]], i64 1
+; THR15-NEXT:    [[ARRAYIDX27_1:%.*]] = getelementptr i8, ptr [[ADD_PTR3]], i64 3
+; THR15-NEXT:    [[TMP5:%.*]] = load i8, ptr [[ARRAYIDX27_1]], align 1
+; THR15-NEXT:    [[CONV33_1:%.*]] = zext i8 [[TMP5]] to i32
 ; THR15-NEXT:    [[ADD_PTR_1:%.*]] = getelementptr i8, ptr [[ADD_PTR]], i64 [[IDX_EXT]]
 ; THR15-NEXT:    [[ADD_PTR64_1:%.*]] = getelementptr i8, ptr [[ADD_PTR64]], i64 [[IDX_EXT63]]
 ; THR15-NEXT:    [[ARRAYIDX3_2:%.*]] = getelementptr i8, ptr [[ADD_PTR_1]], i64 4
 ; THR15-NEXT:    [[ARRAYIDX5_2:%.*]] = getelementptr i8, ptr [[ADD_PTR64_1]], i64 4
+; THR15-NEXT:    [[ARRAYIDX8_2:%.*]] = getelementptr i8, ptr [[ADD_PTR_1]], i64 1
+; THR15-NEXT:    [[TMP4:%.*]] = load <4 x i8>, ptr [[ADD_PTR_1]], align 1
+; THR15-NEXT:    [[TMP7:%.*]] = load i8, ptr [[ARRAYIDX8_2]], align 1
+; THR15-NEXT:    [[TMP6:%.*]] = load i8, ptr [[ADD_PTR_1]], align 1
+; THR15-NEXT:    [[TMP19:%.*]] = shufflevector <4 x i8> [[TMP4]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP20:%.*]] = zext <2 x i8> [[TMP19]] to <2 x i32>
+; THR15-NEXT:    [[TMP87:%.*]] = zext i8 [[TMP6]] to i32
+; THR15-NEXT:    [[TMP9:%.*]] = load <4 x i8>, ptr [[ADD_PTR64_1]], align 1
+; THR15-NEXT:    [[TMP21:%.*]] = shufflevector <4 x i8> [[TMP9]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP22:%.*]] = zext <2 x i8> [[TMP21]] to <2 x i32>
+; THR15-NEXT:    [[TMP23:%.*]] = sub <2 x i32> [[TMP20]], [[TMP22]]
+; THR15-NEXT:    [[TMP13:%.*]] = load <4 x i8>, ptr [[ARRAYIDX3_2]], align 1
+; THR15-NEXT:    [[TMP24:%.*]] = shufflevector <4 x i8> [[TMP13]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP25:%.*]] = zext <2 x i8> [[TMP24]] to <2 x i32>
+; THR15-NEXT:    [[TMP16:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5_2]], align 1
+; THR15-NEXT:    [[TMP26:%.*]] = shufflevector <4 x i8> [[TMP16]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP27:%.*]] = zext <2 x i8> [[TMP26]] to <2 x i32>
+; THR15-NEXT:    [[TMP28:%.*]] = sub <2 x i32> [[TMP25]], [[TMP27]]
+; THR15-NEXT:    [[TMP29:%.*]] = shl <2 x i32> [[TMP28]], splat (i32 16)
+; THR15-NEXT:    [[TMP59:%.*]] = add <2 x i32> [[TMP29]], [[TMP23]]
+; THR15-NEXT:    [[TMP31:%.*]] = shufflevector <4 x i8> [[TMP4]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP32:%.*]] = zext <2 x i8> [[TMP31]] to <2 x i32>
+; THR15-NEXT:    [[TMP86:%.*]] = zext i8 [[TMP7]] to i32
+; THR15-NEXT:    [[TMP33:%.*]] = shufflevector <4 x i8> [[TMP9]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP34:%.*]] = zext <2 x i8> [[TMP33]] to <2 x i32>
+; THR15-NEXT:    [[TMP35:%.*]] = sub <2 x i32> [[TMP32]], [[TMP34]]
+; THR15-NEXT:    [[TMP36:%.*]] = shufflevector <4 x i8> [[TMP13]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP37:%.*]] = zext <2 x i8> [[TMP36]] to <2 x i32>
+; THR15-NEXT:    [[TMP38:%.*]] = shufflevector <4 x i8> [[TMP16]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP39:%.*]] = zext <2 x i8> [[TMP38]] to <2 x i32>
+; THR15-NEXT:    [[TMP40:%.*]] = sub <2 x i32> [[TMP37]], [[TMP39]]
+; THR15-NEXT:    [[TMP41:%.*]] = shl <2 x i32> [[TMP40]], splat (i32 16)
+; THR15-NEXT:    [[TMP76:%.*]] = add <2 x i32> [[TMP41]], [[TMP35]]
+; THR15-NEXT:    [[TMP30:%.*]] = add <2 x i32> [[TMP76]], [[TMP59]]
+; THR15-NEXT:    [[TMP42:%.*]] = sub <2 x i32> [[TMP59]], [[TMP76]]
+; THR15-NEXT:    [[TMP43:%.*]] = extractelement <2 x i32> [[TMP30]], i32 0
+; THR15-NEXT:    [[TMP44:%.*]] = extractelement <2 x i32> [[TMP30]], i32 1
+; THR15-NEXT:    [[ADD44_2:%.*]] = add i32 [[TMP44]], [[TMP43]]
+; THR15-NEXT:    [[TMP45:%.*]] = extractelement <2 x i32> [[TMP42]], i32 0
+; THR15-NEXT:    [[TMP46:%.*]] = extractelement <2 x i32> [[TMP42]], i32 1
+; THR15-NEXT:    [[ADD46_2:%.*]] = add i32 [[TMP46]], [[TMP45]]
 ; THR15-NEXT:    [[ARRAYIDX5_3:%.*]] = getelementptr i8, ptr null, i64 4
+; THR15-NEXT:    [[TMP47:%.*]] = load <2 x i8>, ptr null, align 1
 ; THR15-NEXT:    [[TMP48:%.*]] = load i8, ptr null, align 1
-; THR15-NEXT:    [[TMP1:%.*]] = load i8, ptr null, align 1
-; THR15-NEXT:    [[TMP2:%.*]] = load <4 x i8>, ptr [[PIX1]], align 1
+; THR15-NEXT:    [[TMP49:%.*]] = zext <2 x i8> [[TMP47]] to <2 x i32>
+; THR15-NEXT:    [[CONV_3:%.*]] = zext i8 [[TMP48]] to i32
+; THR15-NEXT:    [[TMP50:%.*]] = load <2 x i8>, ptr null, align 1
+; THR15-NEXT:    [[TMP51:%.*]] = zext <2 x i8> [[TMP50]] to <2 x i32>
+; THR15-NEXT:    [[TMP52:%.*]] = sub <2 x i32> [[TMP49]], [[TMP51]]
+; THR15-NEXT:    [[TMP53:%.*]] = call <2 x i8> @llvm.experimental.vp.strided.load.v2i8.p0.i64(ptr align 1 null, i64 4, <2 x i1> splat (i1 true), i32 2)
+; THR15-NEXT:    [[TMP54:%.*]] = zext <2 x i8> [[TMP53]] to <2 x i32>
+; THR15-NEXT:    [[TMP77:%.*]] = shufflevector <2 x i32> [[TMP54]], <2 x i32> poison, <2 x i32> <i32 1, i32 0>
+; THR15-NEXT:    [[TMP55:%.*]] = load <2 x i8>, ptr [[ARRAYIDX5_3]], align 1
+; THR15-NEXT:    [[TMP56:%.*]] = zext <2 x i8> [[TMP55]] to <2 x i32>
+; THR15-NEXT:    [[TMP57:%.*]] = sub <2 x i32> [[TMP77]], [[TMP56]]
+; THR15-NEXT:    [[TMP58:%.*]] = shl <2 x i32> [[TMP57]], splat (i32 16)
+; THR15-NEXT:    [[TMP72:%.*]] = add <2 x i32> [[TMP58]], [[TMP52]]
+; THR15-NEXT:    [[ARRAYIDX20_3:%.*]] = getelementptr i8, ptr null, i64 2
+; THR15-NEXT:    [[ARRAYIDX22_3:%.*]] = getelementptr i8, ptr null, i64 2
+; THR15-NEXT:    [[ARRAYIDX27_3:%.*]] = getelementptr i8, ptr null, i64 6
+; THR15-NEXT:    [[TMP60:%.*]] = load <2 x i8>, ptr [[ARRAYIDX20_3]], align 1
+; THR15-NEXT:    [[TMP61:%.*]] = zext <2 x i8> [[TMP60]] to <2 x i32>
+; THR15-NEXT:    [[TMP62:%.*]] = load <2 x i8>, ptr [[ARRAYIDX22_3]], align 1
+; THR15-NEXT:    [[TMP63:%.*]] = zext <2 x i8> [[TMP62]] to <2 x i32>
+; THR15-NEXT:    [[TMP64:%.*]] = sub <2 x i32> [[TMP61]], [[TMP63]]
+; THR15-NEXT:    [[TMP65:%.*]] = call <2 x i8> @llvm.masked.gather.v2i8.v2p0(<2 x ptr> zeroinitializer, i32 1, <2 x i1> splat (i1 true), <2 x i8> poison)
+; THR15-NEXT:    [[TMP66:%.*]] = zext <2 x i8> [[TMP65]] to <2 x i32>
+; THR15-NEXT:    [[TMP67:%.*]] = load <2 x i8>, ptr [[ARRAYIDX27_3]], align 1
+; THR15-NEXT:    [[TMP68:%.*]] = zext <2 x i8> [[TMP67]] to <2 x i32>
+; THR15-NEXT:    [[TMP69:%.*]] = sub <2 x i32> [[TMP66]], [[TMP68]]
+; THR15-NEXT:    [[TMP70:%.*]] = shl <2 x i32> [[TMP69]], splat (i32 16)
+; THR15-NEXT:    [[TMP73:%.*]] = add <2 x i32> [[TMP70]], [[TMP64]]
+; THR15-NEXT:    [[TMP74:%.*]] = extractelement <2 x i32> [[TMP72]], i32 0
+; THR15-NEXT:    [[TMP75:%.*]] = extractelement <2 x i32> [[TMP72]], i32 1
+; THR15-NEXT:    [[ADD48_3:%.*]] = add i32 [[TMP75]], [[TMP74]]
+; THR15-NEXT:    [[SUB45_3:%.*]] = sub i32 [[TMP74]], [[TMP75]]
+; THR15-NEXT:    [[TMP80:%.*]] = extractelement <2 x i32> [[TMP73]], i32 0
+; THR15-NEXT:    [[TMP81:%.*]] = extractelement <2 x i32> [[TMP73]], i32 1
+; THR15-NEXT:    [[ADD55_3:%.*]] = add i32 [[TMP81]], [[TMP80]]
+; THR15-NEXT:    [[SUB47_3:%.*]] = sub i32 [[TMP80]], [[TMP81]]
+; THR15-NEXT:    [[ADD48_4:%.*]] = add i32 [[ADD55_3]], [[ADD48_3]]
+; THR15-NEXT:    [[TMP78:%.*]] = shufflevector <2 x i32> [[TMP30]], <2 x i32> poison, <2 x i32> <i32 poison, i32 0>
+; THR15-NEXT:    [[TMP71:%.*]] = insertelement <2 x i32> [[TMP78]], i32 [[ADD48_3]], i32 0
+; THR15-NEXT:    [[TMP83:%.*]] = insertelement <2 x i32> [[TMP30]], i32 [[ADD55_3]], i32 0
+; THR15-NEXT:    [[TMP79:%.*]] = sub <2 x i32> [[TMP71]], [[TMP83]]
+; THR15-NEXT:    [[ADD55_4:%.*]] = add i32 [[SUB47_3]], [[SUB45_3]]
+; THR15-NEXT:    [[TMP137:%.*]] = shufflevector <2 x i32> [[TMP42]], <2 x i32> poison, <2 x i32> <i32 poison, i32 0>
+; THR15-NEXT:    [[TMP82:%.*]] = insertelement <2 x i32> [[TMP137]], i32 [[SUB45_3]], i32 0
+; THR15-NEXT:    [[TMP84:%.*]] = insertelement <2 x i32> [[TMP42]], i32 [[SUB47_3]], i32 0
+; THR15-NEXT:    [[TMP85:%.*]] = sub <2 x i32> [[TMP82]], [[TMP84]]
+; THR15-NEXT:    [[ADD94:%.*]] = add i32 [[ADD48_4]], [[ADD44_2]]
+; THR15-NEXT:    [[SUB102:%.*]] = sub i32 [[ADD44_2]], [[ADD48_4]]
+; THR15-NEXT:    [[SHR_I:%.*]] = lshr i32 [[CONV_3]], 15
+; THR15-NEXT:    [[AND_I:%.*]] = and i32 [[SHR_I]], 65537
+; THR15-NEXT:    [[MUL_I:%.*]] = mul i32 [[AND_I]], 65535
+; THR15-NEXT:    [[SHR_I49:%.*]] = lshr i32 [[TMP44]], 15
+; THR15-NEXT:    [[AND_I50:%.*]] = and i32 [[SHR_I49]], 65537
+; THR15-NEXT:    [[MUL_I51:%.*]] = mul i32 [[AND_I50]], 65535
+; THR15-NEXT:    [[ADD94_1:%.*]] = add i32 [[ADD55_4]], [[ADD46_2]]
+; THR15-NEXT:    [[SUB102_1:%.*]] = sub i32 [[ADD46_2]], [[ADD55_4]]
+; THR15-NEXT:    [[SHR_I_1:%.*]] = lshr i32 [[TMP86]], 15
+; THR15-NEXT:    [[AND_I_1:%.*]] = and i32 [[SHR_I_1]], 65537
+; THR15-NEXT:    [[MUL_I_1:%.*]] = mul i32 [[AND_I_1]], 65535
+; THR15-NEXT:    [[SHR_I49_1:%.*]] = lshr i32 [[TMP87]], 15
+; THR15-NEXT:    [[AND_I50_1:%.*]] = and i32 [[SHR_I49_1]], 65537
+; THR15-NEXT:    [[MUL_I51_1:%.*]] = mul i32 [[AND_I50_1]], 65535
+; THR15-NEXT:    [[TMP88:%.*]] = extractelement <2 x i32> [[TMP79]], i32 0
+; THR15-NEXT:    [[TMP89:%.*]] = extractelement <2 x i32> [[TMP79]], i32 1
+; THR15-NEXT:    [[ADD94_2:%.*]] = add i32 [[TMP88]], [[TMP89]]
+; THR15-NEXT:    [[SUB102_2:%.*]] = sub i32 [[TMP89]], [[TMP88]]
+; THR15-NEXT:    [[SHR_I49_2:%.*]] = lshr i32 [[CONV_1]], 15
+; THR15-NEXT:    [[AND_I50_2:%.*]] = and i32 [[SHR_I49_2]], 65537
+; THR15-NEXT:    [[MUL_I51_2:%.*]] = mul i32 [[AND_I50_2]], 65535
+; THR15-NEXT:    [[TMP90:%.*]] = extractelement <2 x i32> [[TMP85]], i32 0
+; THR15-NEXT:    [[TMP91:%.*]] = extractelement <2 x i32> [[TMP85]], i32 1
+; THR15-NEXT:    [[ADD94_3:%.*]] = add i32 [[TMP90]], [[TMP91]]
+; THR15-NEXT:    [[SUB102_3:%.*]] = sub i32 [[TMP91]], [[TMP90]]
+; THR15-NEXT:    [[SHR_I49_3:%.*]] = lshr i32 [[CONV]], 15
+; THR15-NEXT:    [[AND_I50_3:%.*]] = and i32 [[SHR_I49_3]], 65537
+; THR15-NEXT:    [[MUL_I51_3:%.*]] = mul i32 [[AND_I50_3]], 65535
+; THR15-NEXT:    [[TMP92:%.*]] = load <2 x i8>, ptr [[ARRAYIDX8]], align 1
+; THR15-NEXT:    [[TMP93:%.*]] = zext <2 x i8> [[TMP92]] to <2 x i32>
 ; THR15-NEXT:    [[TMP143:%.*]] = load <4 x i8>, ptr [[PIX2]], align 1
+; THR15-NEXT:    [[TMP94:%.*]] = shufflevector <4 x i8> [[TMP143]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP95:%.*]] = zext <2 x i8> [[TMP94]] to <2 x i32>
 ; THR15-NEXT:    [[TMP146:%.*]] = load <4 x i8>, ptr [[ARRAYIDX3]], align 1
+; THR15-NEXT:    [[TMP96:%.*]] = shufflevector <4 x i8> [[TMP146]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP97:%.*]] = zext <2 x i8> [[TMP96]] to <2 x i32>
 ; THR15-NEXT:    [[TMP147:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5]], align 1
-; THR15-NEXT:    [[TMP6:%.*]] = load <4 x i8>, ptr [[ADD_PTR3]], align 1
+; THR15-NEXT:    [[TMP98:%.*]] = shufflevector <4 x i8> [[TMP147]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP99:%.*]] = zext <2 x i8> [[TMP98]] to <2 x i32>
+; THR15-NEXT:    [[TMP100:%.*]] = sub <2 x i32> [[TMP97]], [[TMP99]]
+; THR15-NEXT:    [[TMP101:%.*]] = shl <2 x i32> [[TMP100]], splat (i32 16)
+; THR15-NEXT:    [[TMP102:%.*]] = shufflevector <4 x i8> [[TMP143]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP103:%.*]] = zext <2 x i8> [[TMP102]] to <2 x i32>
+; THR15-NEXT:    [[TMP104:%.*]] = shufflevector <4 x i8> [[TMP146]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP105:%.*]] = zext <2 x i8> [[TMP104]] to <2 x i32>
+; THR15-NEXT:    [[TMP106:%.*]] = shufflevector <4 x i8> [[TMP147]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP107:%.*]] = zext <2 x i8> [[TMP106]] to <2 x i32>
+; THR15-NEXT:    [[TMP108:%.*]] = sub <2 x i32> [[TMP105]], [[TMP107]]
+; THR15-NEXT:    [[TMP109:%.*]] = shl <2 x i32> [[TMP108]], splat (i32 16)
+; THR15-NEXT:    [[TMP110:%.*]] = insertelement <2 x i32> [[TMP93]], i32 [[CONV33]], i32 1
+; THR15-NEXT:    [[TMP111:%.*]] = sub <2 x i32> [[TMP110]], [[TMP103]]
+; THR15-NEXT:    [[TMP112:%.*]] = add <2 x i32> [[TMP109]], [[TMP111]]
+; THR15-NEXT:    [[TMP113:%.*]] = insertelement <2 x i32> [[TMP93]], i32 [[CONV]], i32 0
+; THR15-NEXT:    [[TMP114:%.*]] = sub <2 x i32> [[TMP113]], [[TMP95]]
+; THR15-NEXT:    [[TMP115:%.*]] = add <2 x i32> [[TMP101]], [[TMP114]]
+; THR15-NEXT:    [[TMP116:%.*]] = shufflevector <2 x i32> [[TMP112]], <2 x i32> [[TMP115]], <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP117:%.*]] = add <2 x i32> [[TMP112]], [[TMP115]]
+; THR15-NEXT:    [[TMP118:%.*]] = sub <2 x i32> [[TMP115]], [[TMP112]]
+; THR15-NEXT:    [[TMP119:%.*]] = extractelement <2 x i32> [[TMP117]], i32 0
+; THR15-NEXT:    [[TMP120:%.*]] = extractelement <2 x i32> [[TMP117]], i32 1
+; THR15-NEXT:    [[ADD48:%.*]] = add i32 [[TMP120]], [[TMP119]]
+; THR15-NEXT:    [[SUB51:%.*]] = sub i32 [[TMP119]], [[TMP120]]
+; THR15-NEXT:    [[TMP121:%.*]] = extractelement <2 x i32> [[TMP118]], i32 0
+; THR15-NEXT:    [[TMP122:%.*]] = extractelement <2 x i32> [[TMP118]], i32 1
+; THR15-NEXT:    [[ADD55:%.*]] = add i32 [[TMP122]], [[TMP121]]
+; THR15-NEXT:    [[SUB59:%.*]] = sub i32 [[TMP121]], [[TMP122]]
+; THR15-NEXT:    [[SHR_I59:%.*]] = lshr i32 [[TMP120]], 15
+; THR15-NEXT:    [[AND_I60:%.*]] = and i32 [[SHR_I59]], 65537
+; THR15-NEXT:    [[MUL_I61:%.*]] = mul i32 [[AND_I60]], 65535
+; THR15-NEXT:    [[SHR_I59_1:%.*]] = lshr i32 [[TMP122]], 15
+; THR15-NEXT:    [[AND_I60_1:%.*]] = and i32 [[SHR_I59_1]], 65537
+; THR15-NEXT:    [[MUL_I61_1:%.*]] = mul i32 [[AND_I60_1]], 65535
+; THR15-NEXT:    [[TMP123:%.*]] = load <2 x i8>, ptr [[ARRAYIDX8_1]], align 1
+; THR15-NEXT:    [[TMP124:%.*]] = zext <2 x i8> [[TMP123]] to <2 x i32>
 ; THR15-NEXT:    [[TMP148:%.*]] = load <4 x i8>, ptr [[ADD_PTR644]], align 1
+; THR15-NEXT:    [[TMP125:%.*]] = shufflevector <4 x i8> [[TMP148]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP126:%.*]] = zext <2 x i8> [[TMP125]] to <2 x i32>
 ; THR15-NEXT:    [[TMP152:%.*]] = load <4 x i8>, ptr [[ARRAYIDX3_1]], align 1
+; THR15-NEXT:    [[TMP127:%.*]] = shufflevector <4 x i8> [[TMP152]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP128:%.*]] = zext <2 x i8> [[TMP127]] to <2 x i32>
 ; THR15-NEXT:    [[TMP153:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5_1]], align 1
-; THR15-NEXT:    [[TMP10:%.*]] = load <4 x i8>, ptr [[ADD_PTR_1]], align 1
-; THR15-NEXT:    [[TMP11:%.*]] = load <4 x i8>, ptr [[ADD_PTR64_1]], align 1
-; THR15-NEXT:    [[TMP12:%.*]] = load <4 x i8>, ptr [[ARRAYIDX3_2]], align 1
-; THR15-NEXT:    [[TMP13:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5_2]], align 1
-; THR15-NEXT:    [[TMP14:%.*]] = load <4 x i8>, ptr null, align 1
-; THR15-NEXT:    [[TMP15:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> poison, <4 x i8> [[TMP10]], i64 0)
-; THR15-NEXT:    [[TMP16:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP15]], <4 x i8> [[TMP14]], i64 4)
-; THR15-NEXT:    [[TMP17:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP16]], <4 x i8> [[TMP2]], i64 8)
-; THR15-NEXT:    [[TMP18:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP17]], <4 x i8> [[TMP6]], i64 12)
-; THR15-NEXT:    [[TMP19:%.*]] = zext <16 x i8> [[TMP18]] to <16 x i32>
-; THR15-NEXT:    [[TMP20:%.*]] = load <4 x i8>, ptr null, align 1
-; THR15-NEXT:    [[TMP21:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> poison, <4 x i8> [[TMP11]], i64 0)
-; THR15-NEXT:    [[TMP22:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP21]], <4 x i8> [[TMP20]], i64 4)
-; THR15-NEXT:    [[TMP23:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP22]], <4 x i8> [[TMP143]], i64 8)
-; THR15-NEXT:    [[TMP24:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP23]], <4 x i8> [[TMP148]], i64 12)
-; THR15-NEXT:    [[TMP25:%.*]] = zext <16 x i8> [[TMP24]] to <16 x i32>
-; THR15-NEXT:    [[TMP26:%.*]] = sub <16 x i32> [[TMP19]], [[TMP25]]
-; THR15-NEXT:    [[TMP27:%.*]] = shufflevector <16 x i32> [[TMP26]], <16 x i32> poison, <16 x i32> <i32 3, i32 7, i32 15, i32 11, i32 2, i32 6, i32 14, i32 10, i32 1, i32 5, i32 13, i32 9, i32 0, i32 4, i32 12, i32 8>
-; THR15-NEXT:    [[TMP28:%.*]] = call <2 x i8> @llvm.experimental.vp.strided.load.v2i8.p0.i64(ptr align 1 null, i64 4, <2 x i1> splat (i1 true), i32 2)
-; THR15-NEXT:    [[TMP29:%.*]] = shufflevector <2 x i8> [[TMP28]], <2 x i8> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison>
-; THR15-NEXT:    [[TMP30:%.*]] = shufflevector <4 x i8> [[TMP12]], <4 x i8> [[TMP29]], <16 x i32> <i32 3, i32 4, i32 poison, i32 poison, i32 2, i32 poison, i32 poison, i32 poison, i32 1, i32 poison, i32 poison, i32 poison, i32 0, i32 5, i32 poison, i32 poison>
-; THR15-NEXT:    [[TMP31:%.*]] = shufflevector <4 x i8> [[TMP152]], <4 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; THR15-NEXT:    [[TMP32:%.*]] = shufflevector <16 x i8> [[TMP30]], <16 x i8> [[TMP31]], <16 x i32> <i32 0, i32 1, i32 19, i32 poison, i32 4, i32 poison, i32 18, i32 poison, i32 8, i32 poison, i32 17, i32 poison, i32 12, i32 13, i32 16, i32 poison>
-; THR15-NEXT:    [[TMP33:%.*]] = shufflevector <4 x i8> [[TMP146]], <4 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; THR15-NEXT:    [[TMP34:%.*]] = shufflevector <16 x i8> [[TMP32]], <16 x i8> [[TMP33]], <16 x i32> <i32 0, i32 1, i32 2, i32 19, i32 4, i32 poison, i32 6, i32 18, i32 8, i32 poison, i32 10, i32 17, i32 12, i32 13, i32 14, i32 16>
-; THR15-NEXT:    [[TMP35:%.*]] = insertelement <16 x i8> [[TMP34]], i8 [[TMP1]], i32 5
-; THR15-NEXT:    [[TMP36:%.*]] = insertelement <16 x i8> [[TMP35]], i8 [[TMP48]], i32 9
-; THR15-NEXT:    [[TMP37:%.*]] = zext <16 x i8> [[TMP36]] to <16 x i32>
-; THR15-NEXT:    [[TMP38:%.*]] = load <4 x i8>, ptr [[ARRAYIDX5_3]], align 1
-; THR15-NEXT:    [[TMP39:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> poison, <4 x i8> [[TMP13]], i64 0)
-; THR15-NEXT:    [[TMP40:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP39]], <4 x i8> [[TMP38]], i64 4)
-; THR15-NEXT:    [[TMP41:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP40]], <4 x i8> [[TMP147]], i64 8)
-; THR15-NEXT:    [[TMP42:%.*]] = call <16 x i8> @llvm.vector.insert.v16i8.v4i8(<16 x i8> [[TMP41]], <4 x i8> [[TMP153]], i64 12)
-; THR15-NEXT:    [[TMP43:%.*]] = zext <16 x i8> [[TMP42]] to <16 x i32>
-; THR15-NEXT:    [[TMP44:%.*]] = shufflevector <16 x i32> [[TMP43]], <16 x i32> poison, <16 x i32> <i32 3, i32 7, i32 15, i32 11, i32 2, i32 6, i32 14, i32 10, i32 1, i32 5, i32 13, i32 9, i32 0, i32 4, i32 12, i32 8>
-; THR15-NEXT:    [[TMP45:%.*]] = sub <16 x i32> [[TMP37]], [[TMP44]]
-; THR15-NEXT:    [[TMP46:%.*]] = shl <16 x i32> [[TMP45]], splat (i32 16)
-; THR15-NEXT:    [[TMP47:%.*]] = add <16 x i32> [[TMP46]], [[TMP27]]
-; THR15-NEXT:    [[TMP70:%.*]] = shufflevector <16 x i32> [[TMP47]], <16 x i32> poison, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 12, i32 13, i32 14, i32 15, i32 8, i32 9, i32 10, i32 11>
-; THR15-NEXT:    [[TMP49:%.*]] = add <16 x i32> [[TMP47]], [[TMP70]]
-; THR15-NEXT:    [[TMP50:%.*]] = sub <16 x i32> [[TMP47]], [[TMP70]]
-; THR15-NEXT:    [[TMP51:%.*]] = shufflevector <16 x i32> [[TMP49]], <16 x i32> [[TMP50]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 28, i32 29, i32 30, i32 31>
-; THR15-NEXT:    [[TMP52:%.*]] = shufflevector <16 x i32> [[TMP51]], <16 x i32> poison, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; THR15-NEXT:    [[TMP53:%.*]] = add <16 x i32> [[TMP51]], [[TMP52]]
-; THR15-NEXT:    [[TMP54:%.*]] = sub <16 x i32> [[TMP51]], [[TMP52]]
-; THR15-NEXT:    [[TMP55:%.*]] = shufflevector <16 x i32> [[TMP53]], <16 x i32> [[TMP54]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
-; THR15-NEXT:    [[TMP56:%.*]] = shufflevector <16 x i32> [[TMP55]], <16 x i32> poison, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
-; THR15-NEXT:    [[TMP57:%.*]] = sub <16 x i32> [[TMP55]], [[TMP56]]
-; THR15-NEXT:    [[TMP58:%.*]] = add <16 x i32> [[TMP55]], [[TMP56]]
-; THR15-NEXT:    [[TMP59:%.*]] = shufflevector <16 x i32> [[TMP57]], <16 x i32> [[TMP58]], <16 x i32> <i32 0, i32 17, i32 18, i32 3, i32 4, i32 21, i32 22, i32 7, i32 8, i32 25, i32 26, i32 11, i32 12, i32 29, i32 30, i32 15>
-; THR15-NEXT:    [[TMP60:%.*]] = shufflevector <16 x i32> [[TMP59]], <16 x i32> poison, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4, i32 11, i32 10, i32 9, i32 8, i32 15, i32 14, i32 13, i32 12>
-; THR15-NEXT:    [[TMP61:%.*]] = add <16 x i32> [[TMP59]], [[TMP60]]
-; THR15-NEXT:    [[TMP62:%.*]] = sub <16 x i32> [[TMP59]], [[TMP60]]
-; THR15-NEXT:    [[TMP63:%.*]] = shufflevector <16 x i32> [[TMP61]], <16 x i32> [[TMP62]], <16 x i32> <i32 0, i32 1, i32 18, i32 19, i32 4, i32 5, i32 22, i32 23, i32 8, i32 9, i32 26, i32 27, i32 12, i32 13, i32 30, i32 31>
-; THR15-NEXT:    [[TMP64:%.*]] = shufflevector <16 x i32> [[TMP51]], <16 x i32> [[TMP19]], <16 x i32> <i32 0, i32 20, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 28, i32 29, i32 30, i32 11, i32 24, i32 25, i32 26, i32 27>
-; THR15-NEXT:    [[TMP65:%.*]] = lshr <16 x i32> [[TMP64]], splat (i32 15)
-; THR15-NEXT:    [[TMP66:%.*]] = and <16 x i32> [[TMP65]], splat (i32 65537)
-; THR15-NEXT:    [[TMP67:%.*]] = mul <16 x i32> [[TMP66]], splat (i32 65535)
-; THR15-NEXT:    [[TMP68:%.*]] = add <16 x i32> [[TMP67]], [[TMP63]]
-; THR15-NEXT:    [[TMP69:%.*]] = xor <16 x i32> [[TMP68]], [[TMP64]]
-; THR15-NEXT:    [[ADD113_3:%.*]] = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> [[TMP69]])
+; THR15-NEXT:    [[TMP129:%.*]] = shufflevector <4 x i8> [[TMP153]], <4 x i8> poison, <2 x i32> <i32 0, i32 2>
+; THR15-NEXT:    [[TMP130:%.*]] = zext <2 x i8> [[TMP129]] to <2 x i32>
+; THR15-NEXT:    [[TMP131:%.*]] = sub <2 x i32> [[TMP128]], [[TMP130]]
+; THR15-NEXT:    [[TMP132:%.*]] = shl <2 x i32> [[TMP131]], splat (i32 16)
+; THR15-NEXT:    [[TMP138:%.*]] = shufflevector <4 x i8> [[TMP148]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP139:%.*]] = zext <2 x i8> [[TMP138]] to <2 x i32>
+; THR15-NEXT:    [[TMP154:%.*]] = shufflevector <4 x i8> [[TMP152]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP155:%.*]] = zext <2 x i8> [[TMP154]] to <2 x i32>
+; THR15-NEXT:    [[TMP133:%.*]] = shufflevector <4 x i8> [[TMP153]], <4 x i8> poison, <2 x i32> <i32 1, i32 3>
+; THR15-NEXT:    [[TMP134:%.*]] = zext <2 x i8> [[TMP133]] to <2 x i32>
+; THR15-NEXT:    [[TMP135:%.*]] = sub <2 x i32> [[TMP155]], [[TMP134]]
+; THR15-NEXT:    [[TMP170:%.*]] = shl <2 x i32> [[TMP135]], splat (i32 16)
+; THR15-NEXT:    [[TMP140:%.*]] = insertelement <2 x i32> [[TMP124]], i32 [[CONV33_1]], i32 1
+; THR15-NEXT:    [[TMP141:%.*]] = sub <2 x i32> [[TMP140]], [[TMP139]]
+; THR15-NEXT:    [[TMP171:%.*]] = add <2 x i32> [[TMP170]], [[TMP141]]
+; THR15-NEXT:    [[TMP186:%.*]] = insertelement <2 x i32> [[TMP124]], i32 [[CONV_1]], i32 0
+; THR15-NEXT:    [[TMP187:%.*]] = sub <2 x i32> [[TMP186]], [[TMP126]]
+; THR15-NEXT:    [[TMP142:%.*]] = add <2 x i32> [[TMP132]], [[TMP187]]
+; THR15-NEXT:    [[TMP136:%.*]] = add <2 x i32> [[TMP171]], [[TMP142]]
+; THR15-NEXT:    [[TMP149:%.*]] = sub <2 x i32> [[TMP142]], [[TMP171]]
+; THR15-NEXT:    [[TMP144:%.*]] = extractelement <2 x i32> [[TMP136]], i32 0
+; THR15-NEXT:    [[TMP145:%.*]] = extractelement <2 x i32> [[TMP136]], i32 1
+; THR15-NEXT:    [[ADD48_2:%.*]] = add i32 [[TMP145]], [[TMP144]]
+; THR15-NEXT:    [[SUB45_1:%.*]] = sub i32 [[TMP144]], [[TMP145]]
+; THR15-NEXT:    [[TMP150:%.*]] = extractelement <2 x i32> [[TMP149]], i32 0
+; THR15-NEXT:    [[TMP151:%.*]] = extractelement <2 x i32> [[TMP149]], i32 1
+; THR15-NEXT:    [[ADD48_1:%.*]] = add i32 [[TMP151]], [[TMP150]]
+; THR15-NEXT:    [[SUB51_1:%.*]] = sub i32 [[TMP150]], [[TMP151]]
+; THR15-NEXT:    [[SHR_I54:%.*]] = lshr i32 [[TMP145]], 15
+; THR15-NEXT:    [[AND_I55:%.*]] = and i32 [[SHR_I54]], 65537
+; THR15-NEXT:    [[MUL_I56:%.*]] = mul i32 [[AND_I55]], 65535
+; THR15-NEXT:    [[SHR_I54_1:%.*]] = lshr i32 [[TMP151]], 15
+; THR15-NEXT:    [[AND_I55_1:%.*]] = and i32 [[SHR_I54_1]], 65537
+; THR15-NEXT:    [[MUL_I56_1:%.*]] = mul i32 [[AND_I55_1]], 65535
+; THR15-NEXT:    [[TMP156:%.*]] = lshr <2 x i32> [[TMP124]], splat (i32 15)
+; THR15-NEXT:    [[TMP157:%.*]] = and <2 x i32> [[TMP156]], splat (i32 65537)
+; THR15-NEXT:    [[TMP158:%.*]] = mul <2 x i32> [[TMP157]], splat (i32 65535)
+; THR15-NEXT:    [[ADD78:%.*]] = add i32 [[ADD48_2]], [[ADD48]]
+; THR15-NEXT:    [[SUB86:%.*]] = sub i32 [[ADD48]], [[ADD48_2]]
+; THR15-NEXT:    [[ADD103:%.*]] = add i32 [[ADD94]], [[ADD78]]
+; THR15-NEXT:    [[SUB104:%.*]] = sub i32 [[ADD78]], [[ADD94]]
+; THR15-NEXT:    [[ADD105:%.*]] = add i32 [[SUB102]], [[SUB86]]
+; THR15-NEXT:    [[SUB106:%.*]] = sub i32 [[SUB86]], [[SUB102]]
+; THR15-NEXT:    [[ADD_I:%.*]] = add i32 [[MUL_I]], [[ADD103]]
+; THR15-NEXT:    [[XOR_I:%.*]] = xor i32 [[ADD_I]], [[CONV_3]]
+; THR15-NEXT:    [[ADD_I52:%.*]] = add i32 [[MUL_I51]], [[ADD105]]
+; THR15-NEXT:    [[XOR_I53:%.*]] = xor i32 [[ADD_I52]], [[TMP44]]
+; THR15-NEXT:    [[ADD_I57:%.*]] = add i32 [[MUL_I56]], [[SUB104]]
+; THR15-NEXT:    [[XOR_I58:%.*]] = xor i32 [[ADD_I57]], [[TMP145]]
+; THR15-NEXT:    [[ADD_I62:%.*]] = add i32 [[MUL_I61]], [[SUB106]]
+; THR15-NEXT:    [[XOR_I63:%.*]] = xor i32 [[ADD_I62]], [[TMP120]]
+; THR15-NEXT:    [[ADD110:%.*]] = add i32 [[XOR_I53]], [[XOR_I]]
+; THR15-NEXT:    [[ADD112:%.*]] = add i32 [[ADD110]], [[XOR_I58]]
+; THR15-NEXT:    [[ADD113:%.*]] = add i32 [[ADD112]], [[XOR_I63]]
+; THR15-NEXT:    [[ADD78_1:%.*]] = add i32 [[ADD48_1]], [[ADD55]]
+; THR15-NEXT:    [[SUB86_1:%.*]] = sub i32 [[ADD55]], [[ADD48_1]]
+; THR15-NEXT:    [[ADD103_1:%.*]] = add i32 [[ADD94_1]], [[ADD78_1]]
+; THR15-NEXT:    [[SUB104_1:%.*]] = sub i32 [[ADD78_1]], [[ADD94_1]]
+; THR15-NEXT:    [[ADD105_1:%.*]] = add i32 [[SUB102_1]], [[SUB86_1]]
+; THR15-NEXT:    [[SUB106_1:%.*]] = sub i32 [[SUB86_1]], [[SUB102_1]]
+; THR15-NEXT:    [[ADD_I_1:%.*]] = add i32 [[MUL_I_1]], [[ADD103_1]]
+; THR15-NEXT:    [[XOR_I_1:%.*]] = xor i32 [[ADD_I_1]], [[TMP86]]
+; THR15-NEXT:    [[ADD_I52_1:%.*]] = add i32 [[MUL_I51_1]], [[ADD105_1]]
+; THR15-NEXT:    [[XOR_I53_1:%.*]] = xor i32 [[ADD_I52_1]], [[TMP87]]
+; THR15-NEXT:    [[ADD_I57_1:%.*]] = add i32 [[MUL_I56_1]], [[SUB104_1]]
+; THR15-NEXT:    [[XOR_I58_1:%.*]] = xor i32 [[ADD_I57_1]], [[TMP151]]
+; THR15-NEXT:    [[ADD_I62_1:%.*]] = add i32 [[MUL_I61_1]], [[SUB106_1]]
+; THR15-NEXT:    [[XOR_I63_1:%.*]] = xor i32 [[ADD_I62_1]], [[TMP122]]
+; THR15-NEXT:    [[ADD108_1:%.*]] = add i32 [[XOR_I53_1]], [[ADD113]]
+; THR15-NEXT:    [[ADD110_1:%.*]] = add i32 [[ADD108_1]], [[XOR_I_1]]
+; THR15-NEXT:    [[ADD112_1:%.*]] = add i32 [[ADD110_1]], [[XOR_I58_1]]
+; THR15-NEXT:    [[ADD113_1:%.*]] = add i32 [[ADD112_1]], [[XOR_I63_1]]
+; THR15-NEXT:    [[ADD78_2:%.*]] = add i32 [[SUB45_1]], [[SUB51]]
+; THR15-NEXT:    [[SUB86_2:%.*]] = sub i32 [[SUB51]], [[SUB45_1]]
+; THR15-NEXT:    [[TMP159:%.*]] = insertelement <2 x i32> poison, i32 [[ADD78_2]], i32 0
+; THR15-NEXT:    [[TMP160:%.*]] = shufflevector <2 x i32> [[TMP159]], <2 x i32> poison, <2 x i32> zeroinitializer
+; THR15-NEXT:    [[TMP161:%.*]] = insertelement <2 x i32> poison, i32 [[ADD94_2]], i32 0
+; THR15-NEXT:    [[TMP162:%.*]] = shufflevector <2 x i32> [[TMP161]], <2 x i32> poison, <2 x i32> zeroinitializer
+; THR15-NEXT:    [[TMP163:%.*]] = add <2 x i32> [[TMP160]], [[TMP162]]
+; THR15-NEXT:    [[TMP164:%.*]] = sub <2 x i32> [[TMP160]], [[TMP162]]
+; THR15-NEXT:    [[TMP165:%.*]] = shufflevector <2 x i32> [[TMP163]], <2 x i32> [[TMP164]], <2 x i32> <i32 0, i32 3>
+; THR15-NEXT:    [[ADD105_2:%.*]] = add i32 [[SUB102_2]], [[SUB86_2]]
+; THR15-NEXT:    [[SUB106_2:%.*]] = sub i32 [[SUB86_2]], [[SUB102_2]]
+; THR15-NEXT:    [[ADD_I52_2:%.*]] = add i32 [[MUL_I51_2]], [[ADD105_2]]
+; THR15-NEXT:    [[XOR_I53_2:%.*]] = xor i32 [[ADD_I52_2]], [[CONV_1]]
+; THR15-NEXT:    [[TMP166:%.*]] = add <2 x i32> [[TMP158]], [[TMP165]]
+; THR15-NEXT:    [[TMP167:%.*]] = xor <2 x i32> [[TMP166]], [[TMP124]]
+; THR15-NEXT:    [[SHR_I59_2:%.*]] = lshr i32 [[TMP119]], 15
+; THR15-NEXT:    [[AND_I60_2:%.*]] = and i32 [[SHR_I59_2]], 65537
+; THR15-NEXT:    [[MUL_I61_2:%.*]] = mul i32 [[AND_I60_2]], 65535
+; THR15-NEXT:    [[ADD_I62_2:%.*]] = add i32 [[MUL_I61_2]], [[SUB106_2]]
+; THR15-NEXT:    [[XOR_I63_2:%.*]] = xor i32 [[ADD_I62_2]], [[TMP119]]
+; THR15-NEXT:    [[ADD108_2:%.*]] = add i32 [[XOR_I53_2]], [[ADD113_1]]
+; THR15-NEXT:    [[TMP168:%.*]] = extractelement <2 x i32> [[TMP167]], i32 0
+; THR15-NEXT:    [[ADD110_2:%.*]] = add i32 [[ADD108_2]], [[TMP168]]
+; THR15-NEXT:    [[TMP169:%.*]] = extractelement <2 x i32> [[TMP167]], i32 1
+; THR15-NEXT:    [[ADD112_2:%.*]] = add i32 [[ADD110_2]], [[TMP169]]
+; THR15-NEXT:    [[ADD113_2:%.*]] = add i32 [[ADD112_2]], [[XOR_I63_2]]
+; THR15-NEXT:    [[ADD78_3:%.*]] = add i32 [[SUB51_1]], [[SUB59]]
+; THR15-NEXT:    [[SUB86_3:%.*]] = sub i32 [[SUB59]], [[SUB51_1]]
+; THR15-NEXT:    [[TMP172:%.*]] = insertelement <2 x i32> poison, i32 [[ADD78_3]], i32 0
+; THR15-NEXT:    [[TMP173:%.*]] = shufflevector <2 x i32> [[TMP172]], <2 x i32> poison, <2 x i32> zeroinitializer
+; THR15-NEXT:    [[TMP174:%.*]] = insertelement <2 x i32> poison, i32 [[ADD94_3]], i32 0
+; THR15-NEXT:    [[TMP175:%.*]] = shufflevector <2 x i32> [[TMP174]], <2 x i32> poison, <2 x i32> zeroinitializer
+; THR15-NEXT:    [[TMP176:%.*]] = add <2 x i32> [[TMP173]], [[TMP175]]
+; THR15-NEXT:    [[TMP177:%.*]] = sub <2 x i32> [[TMP173]], [[TMP175]]
+; THR15-NEXT:    [[TMP178:%.*]] = shufflevector <2 x i32> [[TMP176]], <2 x i32> [[TMP177]], <2 x i32> <i32 0, i32 3>
+; THR15-NEXT:    [[ADD105_3:%.*]] = add i32 [[SUB102_3]], [[SUB86_3]]
+; THR15-NEXT:    [[SUB106_3:%.*]] = sub i32 [[SUB86_3]], [[SUB102_3]]
+; THR15-NEXT:    [[ADD_I52_3:%.*]] = add i32 [[MUL_I51_3]], [[ADD105_3]]
+; THR15-NEXT:    [[XOR_I53_3:%.*]] = xor i32 [[ADD_I52_3]], [[CONV]]
+; THR15-NEXT:    [[TMP179:%.*]] = lshr <2 x i32> [[TMP93]], splat (i32 15)
+; THR15-NEXT:    [[TMP180:%.*]] = and <2 x i32> [[TMP179]], splat (i32 65537)
+; THR15-NEXT:    [[TMP181:%.*]] = mul <2 x i32> [[TMP180]], splat (i32 65535)
+; THR15-NEXT:    [[TMP182:%.*]] = add <2 x i32> [[TMP181]], [[TMP178]]
+; THR15-NEXT:    [[TMP183:%.*]] = xor <2 x i32> [[TMP182]], [[TMP93]]
+; THR15-NEXT:    [[SHR_I59_3:%.*]] = lshr i32 [[CONV33]], 15
+; THR15-NEXT:    [[AND_I60_3:%.*]] = and i32 [[SHR_I59_3]], 65537
+; THR15-NEXT:    [[MUL_I61_3:%.*]] = mul i32 [[AND_I60_3]], 65535
+; THR15-NEXT:    [[ADD_I62_3:%.*]] = add i32 [[MUL_I61_3]], [[SUB106_3]]
+; THR15-NEXT:    [[XOR_I63_3:%.*]] = xor i32 [[ADD_I62_3]], [[CONV33]]
+; THR15-NEXT:    [[ADD108_3:%.*]] = add i32 [[XOR_I53_3]], [[ADD113_2]]
+; THR15-NEXT:    [[TMP184:%.*]] = extractelement <2 x i32> [[TMP183]], i32 0
+; THR15-NEXT:    [[ADD110_3:%.*]] = add i32 [[ADD108_3]], [[TMP184]]
+; THR15-NEXT:    [[TMP185:%.*]] = extractelement <2 x i32> [[TMP183]], i32 1
+; THR15-NEXT:    [[ADD112_3:%.*]] = add i32 [[ADD110_3]], [[TMP185]]
+; THR15-NEXT:    [[ADD113_3:%.*]] = add i32 [[ADD112_3]], [[XOR_I63_3]]
 ; THR15-NEXT:    ret i32 [[ADD113_3]]
 ;
 entry:

diff  --git a/llvm/test/Transforms/SLPVectorizer/RISCV/reductions.ll b/llvm/test/Transforms/SLPVectorizer/RISCV/reductions.ll
index 7723746dda301..5b0f4a69de4c3 100644
--- a/llvm/test/Transforms/SLPVectorizer/RISCV/reductions.ll
+++ b/llvm/test/Transforms/SLPVectorizer/RISCV/reductions.ll
@@ -342,8 +342,8 @@ define void @reduce_or_2() {
 ; ZVFHMIN-NEXT:    [[TMP4:%.*]] = insertelement <16 x i64> <i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 poison, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0>, i64 [[TMP1]], i32 6
 ; ZVFHMIN-NEXT:    [[TMP5:%.*]] = icmp ult <16 x i64> [[TMP4]], zeroinitializer
 ; ZVFHMIN-NEXT:    [[RDX_OP:%.*]] = or <16 x i1> [[TMP3]], [[TMP5]]
-; ZVFHMIN-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> [[RDX_OP]])
-; ZVFHMIN-NEXT:    br i1 [[TMP6]], label [[TMP8:%.*]], label [[TMP7:%.*]]
+; ZVFHMIN-NEXT:    [[OP_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> [[RDX_OP]])
+; ZVFHMIN-NEXT:    br i1 [[OP_RDX]], label [[TMP9:%.*]], label [[TMP8:%.*]]
 ; ZVFHMIN:       7:
 ; ZVFHMIN-NEXT:    ret void
 ; ZVFHMIN:       8:
@@ -356,8 +356,8 @@ define void @reduce_or_2() {
 ; ZVL128-NEXT:    [[TMP4:%.*]] = insertelement <16 x i64> <i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 poison, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0>, i64 [[TMP1]], i32 6
 ; ZVL128-NEXT:    [[TMP5:%.*]] = icmp ult <16 x i64> [[TMP4]], zeroinitializer
 ; ZVL128-NEXT:    [[RDX_OP:%.*]] = or <16 x i1> [[TMP3]], [[TMP5]]
-; ZVL128-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> [[RDX_OP]])
-; ZVL128-NEXT:    br i1 [[TMP6]], label [[TMP8:%.*]], label [[TMP7:%.*]]
+; ZVL128-NEXT:    [[OP_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> [[RDX_OP]])
+; ZVL128-NEXT:    br i1 [[OP_RDX]], label [[TMP9:%.*]], label [[TMP8:%.*]]
 ; ZVL128:       7:
 ; ZVL128-NEXT:    ret void
 ; ZVL128:       8:
@@ -365,15 +365,17 @@ define void @reduce_or_2() {
 ;
 ; ZVL256-LABEL: @reduce_or_2(
 ; ZVL256-NEXT:    [[TMP1:%.*]] = shl i64 0, 0
-; ZVL256-NEXT:    [[TMP2:%.*]] = insertelement <32 x i64> <i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 poison, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 poison, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0>, i64 [[TMP1]], i32 15
-; ZVL256-NEXT:    [[TMP3:%.*]] = shufflevector <32 x i64> [[TMP2]], <32 x i64> poison, <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 15, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
-; ZVL256-NEXT:    [[TMP4:%.*]] = icmp ult <32 x i64> [[TMP3]], zeroinitializer
-; ZVL256-NEXT:    [[TMP5:%.*]] = call i1 @llvm.vector.reduce.or.v32i1(<32 x i1> [[TMP4]])
-; ZVL256-NEXT:    br i1 [[TMP5]], label [[TMP7:%.*]], label [[TMP6:%.*]]
-; ZVL256:       6:
-; ZVL256-NEXT:    ret void
+; ZVL256-NEXT:    [[TMP2:%.*]] = insertelement <16 x i64> <i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 poison>, i64 [[TMP1]], i32 15
+; ZVL256-NEXT:    [[TMP3:%.*]] = icmp ult <16 x i64> [[TMP2]], zeroinitializer
+; ZVL256-NEXT:    [[TMP4:%.*]] = insertelement <16 x i64> <i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 poison, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0, i64 0>, i64 [[TMP1]], i32 6
+; ZVL256-NEXT:    [[TMP5:%.*]] = icmp ult <16 x i64> [[TMP4]], zeroinitializer
+; ZVL256-NEXT:    [[RDX_OP:%.*]] = or <16 x i1> [[TMP3]], [[TMP5]]
+; ZVL256-NEXT:    [[OP_RDX:%.*]] = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> [[RDX_OP]])
+; ZVL256-NEXT:    br i1 [[OP_RDX]], label [[TMP9:%.*]], label [[TMP8:%.*]]
 ; ZVL256:       7:
 ; ZVL256-NEXT:    ret void
+; ZVL256:       8:
+; ZVL256-NEXT:    ret void
 ;
 ; ZVL512-LABEL: @reduce_or_2(
 ; ZVL512-NEXT:    [[TMP1:%.*]] = shl i64 0, 0


        


More information about the llvm-commits mailing list