[X86][AVX1] Enable *_EXTEND_VECTOR_INREG lowering of 256-bit vectors

[RKSimon]

Summary

As discussed on D52964, this adds 256-bit *_EXTEND_VECTOR_INREG lowering support for AVX1 targets to help improve SimplifyDemandedBits handling.

This is an extension to D52970 - it includes those diffs here, but I'd like to get D52970 dealt with first.

[RKSimon]

rebased

[craig.topper]

test/CodeGen/X86/avg.ll

2399

Is this an increase in code?

[RKSimon]

test/CodeGen/X86/avg.ll

2399

Yes, I enabled AVX1 reuse of SIGN_EXTEND/ZERO_EXTEND 128->256 extensions the same as AVX2 but the codegen falls over more than would be good. I'll remove it for now but this does mean we miss some of the improvements.

[RKSimon]

Don't ignore 128->256 SIGN/ZERO_EXTENDS like we do for AVX2 - avoid regressions.

[craig.topper]

LGTM

lib/Target/X86/X86ISelLowering.cpp

if (!Subtarget.useSoftFloat() && Subtarget.hasAVX()) {
     for (auto VT : { MVT::v32i8, MVT::v16i16, MVT::v8i32 }) {
       setOperationAction(ISD::ABS,  VT, HasInt256 ? Legal : Custom);
       setOperationAction(ISD::SMAX, VT, HasInt256 ? Legal : Custom);
       setOperationAction(ISD::UMAX, VT, HasInt256 ? Legal : Custom);
       setOperationAction(ISD::SMIN, VT, HasInt256 ? Legal : Custom);
       setOperationAction(ISD::UMIN, VT, HasInt256 ? Legal : Custom);
     }
 
-    if (HasInt256) {
-      setOperationAction(ISD::SIGN_EXTEND_VECTOR_INREG, MVT::v4i64,  Custom);
-      setOperationAction(ISD::SIGN_EXTEND_VECTOR_INREG, MVT::v8i32,  Custom);
-      setOperationAction(ISD::SIGN_EXTEND_VECTOR_INREG, MVT::v16i16, Custom);
+    for (auto VT : { MVT::v16i16, MVT::v8i32, MVT::v4i64 }) {
+      setOperationAction(ISD::SIGN_EXTEND_VECTOR_INREG, VT, Custom);
+      setOperationAction(ISD::ZERO_EXTEND_VECTOR_INREG, VT, Custom);
+    }
 
+    if (HasInt256) {
       // The custom lowering for UINT_TO_FP for v8i32 becomes interesting
       // when we have a 256bit-wide blend with immediate.
       setOperationAction(ISD::UINT_TO_FP, MVT::v8i32, Custom);
 
       // AVX2 also has wider vector sign/zero extending loads, VPMOV[SZ]X
       for (auto LoadExtOp : { ISD::SEXTLOAD, ISD::ZEXTLOAD }) {
         setLoadExtAction(LoadExtOp, MVT::v16i16, MVT::v16i8, Legal);
         setLoadExtAction(LoadExtOp, MVT::v8i32,  MVT::v8i8,  Legal);
static SDValue LowerEXTEND_VECTOR_INREG(SDValue Op,
   MVT InSVT = InVT.getVectorElementType();
   assert(SVT.getSizeInBits() > InSVT.getSizeInBits());
 
   if (SVT != MVT::i64 && SVT != MVT::i32 && SVT != MVT::i16)
     return SDValue();
   if (InSVT != MVT::i32 && InSVT != MVT::i16 && InSVT != MVT::i8)
     return SDValue();
   if (!(VT.is128BitVector() && Subtarget.hasSSE2()) &&
-      !(VT.is256BitVector() && Subtarget.hasInt256()) &&
+      !(VT.is256BitVector() && Subtarget.hasAVX()) &&
       !(VT.is512BitVector() && Subtarget.hasAVX512()))
     return SDValue();
 
   SDLoc dl(Op);
+  unsigned Opc = Op.getOpcode();
+  unsigned NumElts = VT.getVectorNumElements();
 
   // For 256-bit vectors, we only need the lower (128-bit) half of the input.
   // For 512-bit vectors, we need 128-bits or 256-bits.
   if (VT.getSizeInBits() > 128) {
     // Input needs to be at least the same number of elements as output, and
     // at least 128-bits.
-    int InSize = InSVT.getSizeInBits() * VT.getVectorNumElements();
+    int InSize = InSVT.getSizeInBits() * NumElts;
     In = extractSubVector(In, 0, DAG, dl, std::max(InSize, 128));
   }
 
-  assert((Op.getOpcode() != ISD::ZERO_EXTEND_VECTOR_INREG ||
-          InVT == MVT::v64i8) && "Zero extend only for v64i8 input!");
-
-  // SSE41 targets can use the pmovsx* instructions directly for 128-bit results,
+  // SSE41 targets can use the pmov[sz]x* instructions directly for 128-bit results,
   // so are legal and shouldn't occur here. AVX2/AVX512 pmovsx* instructions still
   // need to be handled here for 256/512-bit results.
   if (Subtarget.hasInt256()) {
     assert(VT.getSizeInBits() > 128 && "Unexpected 128-bit vector extension");
-    unsigned ExtOpc = Op.getOpcode() == ISD::SIGN_EXTEND_VECTOR_INREG ?
-                        X86ISD::VSEXT : X86ISD::VZEXT;
+    unsigned ExtOpc =
+        Opc == ISD::SIGN_EXTEND_VECTOR_INREG ? X86ISD::VSEXT : X86ISD::VZEXT;
     return DAG.getNode(ExtOpc, dl, VT, In);
   }
 
+  // pre-AVX2 256-bit extensions need to be split into 128-bit instrctions.
+  if (Subtarget.hasAVX()) {
+    assert(VT.is256BitVector() && "256-bit vector expected");
+    int HalfNumElts = NumElts / 2;
+    MVT HalfVT = MVT::getVectorVT(SVT, HalfNumElts);
+
+    InVT = In.getSimpleValueType();
+    unsigned NumSrcElts = InVT.getVectorNumElements();
+    SmallVector<int, 16> HiMask(NumSrcElts, SM_SentinelUndef);
+    for (int i = 0; i != HalfNumElts; ++i)
+      HiMask[i] = HalfNumElts + i;
+
+    SDValue Lo = DAG.getNode(Opc, dl, HalfVT, In);
+    SDValue Hi = DAG.getVectorShuffle(InVT, dl, In, DAG.getUNDEF(InVT), HiMask);
+    Hi = DAG.getNode(Opc, dl, HalfVT, Hi);
+    return DAG.getNode(ISD::CONCAT_VECTORS, dl, VT, Lo, Hi);
+  }
+
   // We should only get here for sign extend.
-  assert(Op.getOpcode() == ISD::SIGN_EXTEND_VECTOR_INREG &&
-         "Unexpected opcode!");
+  assert(Opc == ISD::SIGN_EXTEND_VECTOR_INREG && "Unexpected opcode!");
 
   // pre-SSE41 targets unpack lower lanes and then sign-extend using SRAI.
   SDValue Curr = In;
   MVT CurrVT = InVT;
 
   // As SRAI is only available on i16/i32 types, we expand only up to i32
   // and handle i64 separately.
   while (CurrVT != VT && CurrVT.getVectorElementType() != MVT::i32) {
static SDValue combineToExtendVectorInReg(SDNode *N, SelectionDAG &DAG,
   // Input type must be a vector and we must be extending legal integer types.
   if (!VT.isVector() || VT.getVectorNumElements() < 2)
     return SDValue();
   if (SVT != MVT::i64 && SVT != MVT::i32 && SVT != MVT::i16)
     return SDValue();
   if (InSVT != MVT::i32 && InSVT != MVT::i16 && InSVT != MVT::i8)
     return SDValue();
 
-  // On AVX2+ targets, if the input/output types are both legal then we will be
+  // On AVX+ targets, if the input/output types are both legal then we will be
   // able to use SIGN_EXTEND/ZERO_EXTEND directly.
-  if (Subtarget.hasInt256() && DAG.getTargetLoweringInfo().isTypeLegal(VT) &&
+  if (Subtarget.hasAVX() && DAG.getTargetLoweringInfo().isTypeLegal(VT) &&
       DAG.getTargetLoweringInfo().isTypeLegal(InVT))
     return SDValue();
 
   SDLoc DL(N);
 
   auto ExtendVecSize = [&DAG](const SDLoc &DL, SDValue N, unsigned Size) {
     EVT InVT = N.getValueType();
     EVT OutVT = EVT::getVectorVT(*DAG.getContext(), InVT.getScalarType(),
if (VT.getSizeInBits() < 128 && !(128 % VT.getSizeInBits())) {
     EVT ExVT =
         EVT::getVectorVT(*DAG.getContext(), SVT, 128 / SVT.getSizeInBits());
     SDValue Ex = ExtendVecSize(DL, N0, Scale * InVT.getSizeInBits());
     SDValue SExt = DAG.getNode(Opcode, DL, ExVT, Ex);
     return DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, VT, SExt,
                        DAG.getIntPtrConstant(0, DL));
   }
 
-  // If target-size is 128-bits (or 256-bits on AVX2 target), then convert to
+  // If target-size is 128-bits (or 256-bits on AVX target), then convert to
   // ISD::*_EXTEND_VECTOR_INREG which ensures lowering to X86ISD::V*EXT.
   // Also use this if we don't have SSE41 to allow the legalizer do its job.
   if (!Subtarget.hasSSE41() || VT.is128BitVector() ||
-      (VT.is256BitVector() && Subtarget.hasInt256()) ||
+      (VT.is256BitVector() && Subtarget.hasAVX()) ||
       (VT.is512BitVector() && Subtarget.useAVX512Regs())) {
     SDValue ExOp = ExtendVecSize(DL, N0, VT.getSizeInBits());
     return Opcode == ISD::SIGN_EXTEND
                ? DAG.getSignExtendVectorInReg(ExOp, DL, VT)
                : DAG.getZeroExtendVectorInReg(ExOp, DL, VT);
   }
 
   auto SplitAndExtendInReg = [&](unsigned SplitSize) {
for (unsigned i = 0, Offset = 0; i != NumVecs; ++i, Offset += NumSubElts) {
       SrcVec = Opcode == ISD::SIGN_EXTEND
                    ? DAG.getSignExtendVectorInReg(SrcVec, DL, SubVT)
                    : DAG.getZeroExtendVectorInReg(SrcVec, DL, SubVT);
       Opnds.push_back(SrcVec);
     }
     return DAG.getNode(ISD::CONCAT_VECTORS, DL, VT, Opnds);
   };
 
-  // On pre-AVX2 targets, split into 128-bit nodes of
+  // On pre-AVX targets, split into 128-bit nodes of
   // ISD::*_EXTEND_VECTOR_INREG.
-  if (!Subtarget.hasInt256() && !(VT.getSizeInBits() % 128))
+  if (!Subtarget.hasAVX() && !(VT.getSizeInBits() % 128))
     return SplitAndExtendInReg(128);
 
   // On pre-AVX512 targets, split into 256-bit nodes of
   // ISD::*_EXTEND_VECTOR_INREG.
   if (!Subtarget.useAVX512Regs() && !(VT.getSizeInBits() % 256))
     return SplitAndExtendInReg(256);
 
   return SDValue();

test/CodeGen/X86/avg.ll

 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vmovdqa (%rdi), %ymm2
 ; AVX1-NEXT:    vmovdqa 32(%rdi), %ymm5
 ; AVX1-NEXT:    vmovdqa (%rsi), %ymm1
 ; AVX1-NEXT:    vmovdqa 32(%rsi), %ymm0
 ; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm3
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm3[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm3[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm3[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm6 = xmm3[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm6 = xmm6[0],zero,zero,zero,xmm6[1],zero,zero,zero,xmm6[2],zero,zero,zero,xmm6[3],zero,zero,zero
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm11 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm12 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm13 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm15 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm5[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm14 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm5[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm5[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm10 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm5[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
 ; AVX1-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
 ; AVX1-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm5 = xmm2[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm5, %xmm7, %xmm5
 ; AVX1-NEXT:    vmovdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm7 = xmm2[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm7 = xmm2[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = xmm7[0],zero,zero,zero,xmm7[1],zero,zero,zero,xmm7[2],zero,zero,zero,xmm7[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm7, %xmm4, %xmm9
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm7 = xmm2[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = xmm7[0],zero,zero,zero,xmm7[1],zero,zero,zero,xmm7[2],zero,zero,zero,xmm7[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm7, %xmm6, %xmm8
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm2, %xmm11, %xmm11
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm7 = xmm1[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = xmm7[0],zero,zero,zero,xmm7[1],zero,zero,zero,xmm7[2],zero,zero,zero,xmm7[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm7, %xmm12, %xmm12
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm5 = xmm1[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm5 = xmm1[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm5, %xmm13, %xmm13
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm1[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm4, %xmm3, %xmm3
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm1, %xmm15, %xmm15
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm4, %xmm14, %xmm4
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm6 = xmm0[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm6 = xmm0[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm6 = xmm6[0],zero,zero,zero,xmm6[1],zero,zero,zero,xmm6[2],zero,zero,zero,xmm6[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd %xmm6, %xmm10, %xmm6
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 16-byte Folded Reload
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX1-NEXT:    vpaddd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
 ; AVX1-NEXT:    vpcmpeqd %xmm7, %xmm7, %xmm7

 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
 ; AVX1-NEXT:    vmovdqu %xmm0, (%rax)
 ; AVX1-NEXT:    vmovups %ymm1, (%rax)
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: avg_v48i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vmovdqa (%rdi), %ymm1
-; AVX2-NEXT:    vmovdqa 32(%rdi), %ymm2
-; AVX2-NEXT:    vmovdqa (%rsi), %ymm3
-; AVX2-NEXT:    vmovdqa 32(%rsi), %ymm0
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm4
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm5 = xmm4[2,3,0,1]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm6 = xmm4[3,1,2,3]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm6 = xmm6[0],zero,zero,zero,xmm6[1],zero,zero,zero,xmm6[2],zero,zero,zero,xmm6[3],zero,zero,zero
-; AVX2-NEXT:    vinserti128 $1, %xmm6, %ymm5, %ymm5
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm9 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; AVX2-NEXT:    vpand %ymm9, %ymm5, %ymm5
+; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX2-NEXT:    vmovdqa 32(%rdi), %ymm1
+; AVX2-NEXT:    vmovdqa (%rsi), %ymm2
+; AVX2-NEXT:    vmovdqa 32(%rsi), %ymm3
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm4
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero,xmm4[4],zero,zero,zero,xmm4[5],zero,zero,zero,xmm4[6],zero,zero,zero,xmm4[7],zero,zero,zero
+; AVX2-NEXT:    vpermq {{.*#+}} ymm5 = ymm0[3,1,2,3]
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero,xmm5[4],zero,zero,zero,xmm5[5],zero,zero,zero,xmm5[6],zero,zero,zero,xmm5[7],zero,zero,zero
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm6 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
 ; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm7 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero
 ; AVX2-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm8 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero
+; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm1
 ; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm11 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero,xmm4[4],zero,zero,zero,xmm4[5],zero,zero,zero,xmm4[6],zero,zero,zero,xmm4[7],zero,zero,zero
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm8 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero
+; AVX2-NEXT:    vpaddd %ymm1, %ymm4, %ymm1
+; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm2[3,1,2,3]
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero,xmm4[4],zero,zero,zero,xmm4[5],zero,zero,zero,xmm4[6],zero,zero,zero,xmm4[7],zero,zero,zero
+; AVX2-NEXT:    vpaddd %ymm4, %ymm5, %ymm4
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm5 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero
+; AVX2-NEXT:    vpaddd %ymm5, %ymm6, %ymm5
 ; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm10 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero
-; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm6
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm6[2,3,0,1]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm4 = xmm6[3,1,2,3]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
-; AVX2-NEXT:    vinserti128 $1, %xmm4, %ymm2, %ymm2
-; AVX2-NEXT:    vpand %ymm9, %ymm2, %ymm2
-; AVX2-NEXT:    vpaddd %ymm2, %ymm5, %ymm2
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm4 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero,xmm3[4],zero,zero,zero,xmm3[5],zero,zero,zero,xmm3[6],zero,zero,zero,xmm3[7],zero,zero,zero
-; AVX2-NEXT:    vpaddd %ymm4, %ymm7, %ymm4
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero
+; AVX2-NEXT:    vpaddd %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero,xmm3[4],zero,zero,zero,xmm3[5],zero,zero,zero,xmm3[6],zero,zero,zero,xmm3[7],zero,zero,zero
+; AVX2-NEXT:    vpaddd %ymm2, %ymm7, %ymm2
 ; AVX2-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[2,3,0,1]
 ; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero,xmm3[4],zero,zero,zero,xmm3[5],zero,zero,zero,xmm3[6],zero,zero,zero,xmm3[7],zero,zero,zero
-; AVX2-NEXT:    vpaddd %ymm3, %ymm1, %ymm1
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm3 = xmm6[0],zero,zero,zero,xmm6[1],zero,zero,zero,xmm6[2],zero,zero,zero,xmm6[3],zero,zero,zero,xmm6[4],zero,zero,zero,xmm6[5],zero,zero,zero,xmm6[6],zero,zero,zero,xmm6[7],zero,zero,zero
-; AVX2-NEXT:    vpaddd %ymm3, %ymm11, %ymm3
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm5 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
-; AVX2-NEXT:    vpaddd %ymm5, %ymm8, %ymm5
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
-; AVX2-NEXT:    vpaddd %ymm0, %ymm10, %ymm0
+; AVX2-NEXT:    vpaddd %ymm3, %ymm8, %ymm3
 ; AVX2-NEXT:    vpcmpeqd %ymm6, %ymm6, %ymm6
-; AVX2-NEXT:    vpsubd %ymm6, %ymm2, %ymm2
-; AVX2-NEXT:    vpsubd %ymm6, %ymm4, %ymm4
 ; AVX2-NEXT:    vpsubd %ymm6, %ymm1, %ymm1
-; AVX2-NEXT:    vpsubd %ymm6, %ymm3, %ymm3
+; AVX2-NEXT:    vpsubd %ymm6, %ymm4, %ymm4
 ; AVX2-NEXT:    vpsubd %ymm6, %ymm5, %ymm5
-; AVX2-NEXT:    vpsubd %ymm6, %ymm0, %ymm0
-; AVX2-NEXT:    vpsrld $1, %ymm2, %ymm2
+; AVX2-NEXT:    vpsubd %ymm6, %ymm0, %ymm7
+; AVX2-NEXT:    vpsubd %ymm6, %ymm2, %ymm2
+; AVX2-NEXT:    vpsubd %ymm6, %ymm3, %ymm0
 ; AVX2-NEXT:    vpsrld $1, %ymm0, %ymm0
+; AVX2-NEXT:    vpsrld $1, %ymm2, %ymm2
+; AVX2-NEXT:    vpsrld $1, %ymm7, %ymm3
 ; AVX2-NEXT:    vpsrld $1, %ymm5, %ymm5
-; AVX2-NEXT:    vpsrld $1, %ymm3, %ymm3
-; AVX2-NEXT:    vpsrld $1, %ymm1, %ymm1
 ; AVX2-NEXT:    vpsrld $1, %ymm4, %ymm4
-; AVX2-NEXT:    vextracti128 $1, %ymm4, %xmm6
-; AVX2-NEXT:    vpackusdw %xmm6, %xmm4, %xmm4
+; AVX2-NEXT:    vpsrld $1, %ymm1, %ymm1
+; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm6
+; AVX2-NEXT:    vpackusdw %xmm6, %xmm1, %xmm1
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm6 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm6, %xmm4, %xmm4
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm7
-; AVX2-NEXT:    vpackusdw %xmm7, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb %xmm6, %xmm1, %xmm1
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm4[0],xmm1[0]
-; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm4
-; AVX2-NEXT:    vpackusdw %xmm4, %xmm2, %xmm2
-; AVX2-NEXT:    vpshufb %xmm6, %xmm2, %xmm2
-; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm4
-; AVX2-NEXT:    vpackusdw %xmm4, %xmm3, %xmm3
+; AVX2-NEXT:    vextracti128 $1, %ymm4, %xmm7
+; AVX2-NEXT:    vpackusdw %xmm7, %xmm4, %xmm4
+; AVX2-NEXT:    vpshufb %xmm6, %xmm4, %xmm4
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm4[0]
+; AVX2-NEXT:    vextracti128 $1, %ymm5, %xmm4
+; AVX2-NEXT:    vpackusdw %xmm4, %xmm5, %xmm4
+; AVX2-NEXT:    vpshufb %xmm6, %xmm4, %xmm4
+; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm5
+; AVX2-NEXT:    vpackusdw %xmm5, %xmm3, %xmm3
 ; AVX2-NEXT:    vpshufb %xmm6, %xmm3, %xmm3
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
-; AVX2-NEXT:    vinserti128 $1, %xmm2, %ymm1, %ymm1
-; AVX2-NEXT:    vextracti128 $1, %ymm5, %xmm2
-; AVX2-NEXT:    vpackusdw %xmm2, %xmm5, %xmm2
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm4[0],xmm3[0]
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm3, %ymm1
+; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm3
+; AVX2-NEXT:    vpackusdw %xmm3, %xmm2, %xmm2
 ; AVX2-NEXT:    vpshufb %xmm6, %xmm2, %xmm2
 ; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm3
 ; AVX2-NEXT:    vpackusdw %xmm3, %xmm0, %xmm0
 ; AVX2-NEXT:    vpshufb %xmm6, %xmm0, %xmm0
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm2[0],xmm0[0]
 ; AVX2-NEXT:    vmovdqu %xmm0, (%rax)
 ; AVX2-NEXT:    vmovdqu %ymm1, (%rax)
 ; AVX2-NEXT:    vzeroupper

 ; AVX1-LABEL: not_avg_v16i8_wide_constants:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    pushq %rbp
 ; AVX1-NEXT:    pushq %r15
 ; AVX1-NEXT:    pushq %r14
 ; AVX1-NEXT:    pushq %r13
 ; AVX1-NEXT:    pushq %r12
 ; AVX1-NEXT:    pushq %rbx
-; AVX1-NEXT:    subq $24, %rsp
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
+; AVX1-NEXT:    subq $16, %rsp
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm4 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm5 = xmm4[0],zero,xmm4[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm5, %rbx
-; AVX1-NEXT:    vmovq %xmm5, %rbp
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm4[2,3,0,1]
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm4, %rsi
-; AVX1-NEXT:    vmovq %xmm4, %rcx
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm4 = xmm3[0],zero,xmm3[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm4, %r8
-; AVX1-NEXT:    vmovq %xmm4, %r11
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[2,3,0,1]
+; AVX1-NEXT:    movzbl 3(%rdi), %r14d
+; AVX1-NEXT:    movzbl 2(%rdi), %r15d
+; AVX1-NEXT:    movzbl 1(%rdi), %r12d
+; AVX1-NEXT:    movzbl (%rdi), %edx
+; AVX1-NEXT:    movzbl 7(%rdi), %ebx
+; AVX1-NEXT:    movzbl 6(%rdi), %ecx
+; AVX1-NEXT:    movzbl 5(%rdi), %eax
+; AVX1-NEXT:    movzbl 4(%rdi), %edi
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
+; AVX1-NEXT:    movzbl 3(%rsi), %ebp
+; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
+; AVX1-NEXT:    movzbl 2(%rsi), %ebp
+; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
+; AVX1-NEXT:    movzbl 1(%rsi), %ebp
+; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
+; AVX1-NEXT:    movzbl (%rsi), %r9d
+; AVX1-NEXT:    movzbl 7(%rsi), %r13d
+; AVX1-NEXT:    movzbl 6(%rsi), %r10d
+; AVX1-NEXT:    vmovd %edi, %xmm1
+; AVX1-NEXT:    movzbl 5(%rsi), %ebp
+; AVX1-NEXT:    movzbl 4(%rsi), %edi
+; AVX1-NEXT:    vpinsrd $1, %eax, %xmm1, %xmm1
+; AVX1-NEXT:    vpinsrd $2, %ecx, %xmm1, %xmm1
+; AVX1-NEXT:    vpinsrd $3, %ebx, %xmm1, %xmm3
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm4 = xmm3[2],xmm1[2],xmm3[3],xmm1[3]
+; AVX1-NEXT:    vpextrq $1, %xmm4, %rbx
+; AVX1-NEXT:    vmovq %xmm4, %rsi
 ; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm3, %r13
-; AVX1-NEXT:    vmovq %xmm3, %r12
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm4 = xmm3[0],zero,xmm3[1],zero
+; AVX1-NEXT:    vpextrq $1, %xmm3, %rax
+; AVX1-NEXT:    vmovd %edx, %xmm4
+; AVX1-NEXT:    vmovq %xmm3, %r11
+; AVX1-NEXT:    vpinsrd $1, %r12d, %xmm4, %xmm3
+; AVX1-NEXT:    vpinsrd $2, %r15d, %xmm3, %xmm3
+; AVX1-NEXT:    vpinsrd $3, %r14d, %xmm3, %xmm3
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm4 = xmm3[2],xmm1[2],xmm3[3],xmm1[3]
 ; AVX1-NEXT:    vpextrq $1, %xmm4, %r15
-; AVX1-NEXT:    vmovq %xmm4, %rdi
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[2,3,0,1]
+; AVX1-NEXT:    vmovq %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
 ; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero
 ; AVX1-NEXT:    vpextrq $1, %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
-; AVX1-NEXT:    vmovq %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm1[0],zero,xmm1[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
-; AVX1-NEXT:    vmovq %xmm3, %r10
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm4 = xmm3[0],zero,xmm3[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm4, %rdx
-; AVX1-NEXT:    addq %rbx, %rdx
-; AVX1-NEXT:    vmovq %xmm4, %r9
-; AVX1-NEXT:    addq %rbp, %r9
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[2,3,0,1]
+; AVX1-NEXT:    vmovd %edi, %xmm4
+; AVX1-NEXT:    vmovq %xmm3, %rdi
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm5 = xmm3[2],xmm1[2],xmm3[3],xmm1[3]
+; AVX1-NEXT:    vpinsrd $1, %ebp, %xmm4, %xmm4
+; AVX1-NEXT:    vpextrq $1, %xmm5, %r8
+; AVX1-NEXT:    vpinsrd $2, %r10d, %xmm4, %xmm4
+; AVX1-NEXT:    vmovq %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
 ; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm3, %rax
-; AVX1-NEXT:    addq %rsi, %rax
-; AVX1-NEXT:    movq %rax, %r14
-; AVX1-NEXT:    vmovq %xmm3, %rbp
-; AVX1-NEXT:    addq %rcx, %rbp
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm2[0],zero,xmm2[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm3, %rsi
-; AVX1-NEXT:    addq %r8, %rsi
-; AVX1-NEXT:    vmovq %xmm3, %rax
+; AVX1-NEXT:    vpinsrd $3, %r13d, %xmm4, %xmm4
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm5 = xmm4[2],xmm1[2],xmm4[3],xmm1[3]
+; AVX1-NEXT:    vpextrq $1, %xmm5, %r12
+; AVX1-NEXT:    addq %rbx, %r12
+; AVX1-NEXT:    vmovq %xmm5, %r10
+; AVX1-NEXT:    addq %rsi, %r10
+; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero
+; AVX1-NEXT:    vpextrq $1, %xmm4, %rbx
+; AVX1-NEXT:    addq %rax, %rbx
+; AVX1-NEXT:    movq %rbx, %r13
+; AVX1-NEXT:    vmovq %xmm4, %rax
 ; AVX1-NEXT:    addq %r11, %rax
 ; AVX1-NEXT:    movq %rax, %r11
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm2, %rax
-; AVX1-NEXT:    addq %r13, %rax
-; AVX1-NEXT:    movq %rax, %rcx
-; AVX1-NEXT:    vmovq %xmm2, %rax
-; AVX1-NEXT:    addq %r12, %rax
-; AVX1-NEXT:    movq %rax, %r8
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm2[0],zero,xmm2[1],zero
 ; AVX1-NEXT:    vpextrq $1, %xmm3, %rax
-; AVX1-NEXT:    addq %r15, %rax
-; AVX1-NEXT:    movq %rax, %rbx
-; AVX1-NEXT:    vmovq %xmm3, %rax
-; AVX1-NEXT:    addq %rdi, %rax
-; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
+; AVX1-NEXT:    vmovd %r9d, %xmm4
+; AVX1-NEXT:    vmovq %xmm3, %rbx
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm3 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; AVX1-NEXT:    vpinsrd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4, %xmm4 # 4-byte Folded Reload
+; AVX1-NEXT:    vpextrq $1, %xmm3, %rdx
+; AVX1-NEXT:    vpinsrd $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4, %xmm4 # 4-byte Folded Reload
+; AVX1-NEXT:    vmovq %xmm3, %r14
 ; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm2, %rax
-; AVX1-NEXT:    addq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload
-; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    vmovq %xmm2, %rax
-; AVX1-NEXT:    addq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload
+; AVX1-NEXT:    vpinsrd $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4, %xmm3 # 4-byte Folded Reload
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm4 = xmm3[2],xmm1[2],xmm3[3],xmm1[3]
+; AVX1-NEXT:    vpextrq $1, %xmm4, %rcx
+; AVX1-NEXT:    addq %r15, %rcx
+; AVX1-NEXT:    vmovq %xmm4, %rbp
+; AVX1-NEXT:    addq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Folded Reload
+; AVX1-NEXT:    movq %rbp, %r15
+; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero
+; AVX1-NEXT:    vpextrq $1, %xmm3, %rsi
+; AVX1-NEXT:    addq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload
+; AVX1-NEXT:    vmovq %xmm3, %r9
+; AVX1-NEXT:    addq %rdi, %r9
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm4 = xmm3[2],xmm1[2],xmm3[3],xmm1[3]
+; AVX1-NEXT:    vpextrq $1, %xmm4, %rdi
+; AVX1-NEXT:    addq %r8, %rdi
+; AVX1-NEXT:    movq %rdi, %r8
+; AVX1-NEXT:    vmovq %xmm4, %rdi
+; AVX1-NEXT:    addq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Folded Reload
+; AVX1-NEXT:    movq %rdi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero
+; AVX1-NEXT:    vpextrq $1, %xmm3, %rdi
+; AVX1-NEXT:    addq %rax, %rdi
+; AVX1-NEXT:    movq %rdi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    vmovq %xmm3, %rax
+; AVX1-NEXT:    addq %rbx, %rax
 ; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm1 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX1-NEXT:    vpextrq $1, %xmm1, %rbx
+; AVX1-NEXT:    addq %rdx, %rbx
+; AVX1-NEXT:    vmovq %xmm1, %rdi
+; AVX1-NEXT:    addq %r14, %rdi
 ; AVX1-NEXT:    vpextrq $1, %xmm2, %rax
-; AVX1-NEXT:    addq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload
-; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    vmovq %xmm2, %r12
-; AVX1-NEXT:    addq %r10, %r12
-; AVX1-NEXT:    vpextrq $1, %xmm1, %rax
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; AVX1-NEXT:    vpextrq $1, %xmm0, %r10
-; AVX1-NEXT:    addq %rax, %r10
-; AVX1-NEXT:    vmovq %xmm1, %rax
-; AVX1-NEXT:    vmovq %xmm0, %rdi
-; AVX1-NEXT:    addq %rax, %rdi
-; AVX1-NEXT:    addq $-1, %rdx
+; AVX1-NEXT:    vpextrq $1, %xmm0, %rdx
+; AVX1-NEXT:    addq %rax, %rdx
 ; AVX1-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    vmovq %xmm2, %rax
+; AVX1-NEXT:    vmovq %xmm0, %rdx
+; AVX1-NEXT:    addq %rax, %rdx
+; AVX1-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    addq $-1, %r12
+; AVX1-NEXT:    movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
 ; AVX1-NEXT:    movl $0, %eax
 ; AVX1-NEXT:    adcq $-1, %rax
 ; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    addq $-1, %r9
-; AVX1-NEXT:    movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    movl $0, %eax
-; AVX1-NEXT:    adcq $-1, %rax
-; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    addq $-1, %r14
-; AVX1-NEXT:    movq %r14, (%rsp) # 8-byte Spill
-; AVX1-NEXT:    movl $0, %eax
-; AVX1-NEXT:    adcq $-1, %rax
-; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    addq $-1, %rbp
-; AVX1-NEXT:    movq %rbp, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    addq $-1, %r10
+; AVX1-NEXT:    movq %r10, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
 ; AVX1-NEXT:    movl $0, %eax
 ; AVX1-NEXT:    adcq $-1, %rax
-; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    addq $-1, %rsi
-; AVX1-NEXT:    movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    movq %rax, (%rsp) # 8-byte Spill
+; AVX1-NEXT:    addq $-1, %r13
+; AVX1-NEXT:    movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
 ; AVX1-NEXT:    movl $0, %eax
 ; AVX1-NEXT:    adcq $-1, %rax
-; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
 ; AVX1-NEXT:    addq $-1, %r11
 ; AVX1-NEXT:    movq %r11, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    movl $0, %eax
-; AVX1-NEXT:    adcq $-1, %rax
-; AVX1-NEXT:    movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    movl $0, %edx
+; AVX1-NEXT:    adcq $-1, %rdx
+; AVX1-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
 ; AVX1-NEXT:    addq $-1, %rcx
 ; AVX1-NEXT:    movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    movl $0, %ecx
+; AVX1-NEXT:    adcq $-1, %rcx
+; AVX1-NEXT:    movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    addq $-1, %r15
+; AVX1-NEXT:    movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    movl $0, %ecx
+; AVX1-NEXT:    adcq $-1, %rcx
+; AVX1-NEXT:    addq $-1, %rsi
+; AVX1-NEXT:    movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
 ; AVX1-NEXT:    movl $0, %ebp
 ; AVX1-NEXT:    adcq $-1, %rbp
+; AVX1-NEXT:    addq $-1, %r9
+; AVX1-NEXT:    movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    movl $0, %r12d
+; AVX1-NEXT:    adcq $-1, %r12
 ; AVX1-NEXT:    addq $-1, %r8
 ; AVX1-NEXT:    movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    movl $0, %edx
+; AVX1-NEXT:    adcq $-1, %rdx
+; AVX1-NEXT:    addq $-1, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
 ; AVX1-NEXT:    movl $0, %r15d
 ; AVX1-NEXT:    adcq $-1, %r15
-; AVX1-NEXT:    addq $-1, %rbx
-; AVX1-NEXT:    movq %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    movl $0, %eax
-; AVX1-NEXT:    adcq $-1, %rax
-; AVX1-NEXT:    movq %rax, %rsi
-; AVX1-NEXT:    addq $-1, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
-; AVX1-NEXT:    movl $0, %r13d
-; AVX1-NEXT:    adcq $-1, %r13
 ; AVX1-NEXT:    addq $-1, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
 ; AVX1-NEXT:    movl $0, %r14d
 ; AVX1-NEXT:    adcq $-1, %r14
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload
-; AVX1-NEXT:    addq $-1, %rdx
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload
+; AVX1-NEXT:    addq $-1, %r13
 ; AVX1-NEXT:    movl $0, %r11d
 ; AVX1-NEXT:    adcq $-1, %r11
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    addq $-1, %rax
-; AVX1-NEXT:    movl $0, %ebx
-; AVX1-NEXT:    adcq $-1, %rbx
-; AVX1-NEXT:    addq $-1, %r12
+; AVX1-NEXT:    addq $-1, %rbx
+; AVX1-NEXT:    movl $0, %r10d
+; AVX1-NEXT:    adcq $-1, %r10
+; AVX1-NEXT:    addq $-1, %rdi
 ; AVX1-NEXT:    movl $0, %r9d
 ; AVX1-NEXT:    adcq $-1, %r9
-; AVX1-NEXT:    addq $-1, %r10
+; AVX1-NEXT:    addq $-1, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
 ; AVX1-NEXT:    movl $0, %r8d
 ; AVX1-NEXT:    adcq $-1, %r8
-; AVX1-NEXT:    addq $-1, %rdi
-; AVX1-NEXT:    movl $0, %ecx
-; AVX1-NEXT:    adcq $-1, %rcx
-; AVX1-NEXT:    shldq $63, %rdi, %rcx
-; AVX1-NEXT:    movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    shldq $63, %r10, %r8
-; AVX1-NEXT:    shldq $63, %r12, %r9
-; AVX1-NEXT:    shldq $63, %rax, %rbx
-; AVX1-NEXT:    shldq $63, %rdx, %r11
+; AVX1-NEXT:    addq $-1, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill
+; AVX1-NEXT:    movl $0, %esi
+; AVX1-NEXT:    adcq $-1, %rsi
+; AVX1-NEXT:    shldq $63, %rdi, %r9
+; AVX1-NEXT:    shldq $63, %rbx, %r10
+; AVX1-NEXT:    shldq $63, %r13, %r11
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rdi, %r14
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rdi, %r15
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rdi, %rdx
+; AVX1-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
 ; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rdx, %r14
+; AVX1-NEXT:    shldq $63, %rdx, %r12
 ; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rdx, %r13
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rax, %rsi
-; AVX1-NEXT:    movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rax, %r15
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rax, %rbp
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Reload
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rax, %rsi
+; AVX1-NEXT:    shldq $63, %rdx, %rbp
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rdx, %rcx
+; AVX1-NEXT:    movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rcx, %rdx
 ; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rax, %rcx
 ; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Reload
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rax, %rdi
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %r12 # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rdi, %rcx
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rdi, %rax
+; AVX1-NEXT:    movq %rax, %r13
 ; AVX1-NEXT:    movq (%rsp), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rax, %r12
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rax, %r10
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload
-; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX1-NEXT:    shldq $63, %rdx, %rax
-; AVX1-NEXT:    vmovq %rax, %xmm8
-; AVX1-NEXT:    vmovq %r10, %xmm0
-; AVX1-NEXT:    vmovq %r12, %xmm1
-; AVX1-NEXT:    vmovq %rdi, %xmm11
-; AVX1-NEXT:    vmovq %rcx, %xmm2
-; AVX1-NEXT:    vmovq %rsi, %xmm13
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rdi, %rax
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Reload
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rbx, %rdi
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rbx, %rsi
+; AVX1-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload
+; AVX1-NEXT:    shldq $63, %rbx, %r8
+; AVX1-NEXT:    vmovq %rdi, %xmm8
+; AVX1-NEXT:    vmovq %rax, %xmm0
+; AVX1-NEXT:    vmovq %r13, %xmm1
+; AVX1-NEXT:    vmovq %rcx, %xmm11
+; AVX1-NEXT:    vmovq %rdx, %xmm2
+; AVX1-NEXT:    vmovq {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 8-byte Folded Reload
+; AVX1-NEXT:    # xmm13 = mem[0],zero
 ; AVX1-NEXT:    vmovq %rbp, %xmm14
-; AVX1-NEXT:    vmovq %r15, %xmm15
+; AVX1-NEXT:    vmovq %r12, %xmm15
 ; AVX1-NEXT:    vmovq {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 8-byte Folded Reload
 ; AVX1-NEXT:    # xmm9 = mem[0],zero
-; AVX1-NEXT:    vmovq %r13, %xmm10
+; AVX1-NEXT:    vmovq %r15, %xmm10
 ; AVX1-NEXT:    vmovq %r14, %xmm12
 ; AVX1-NEXT:    vmovq %r11, %xmm3
-; AVX1-NEXT:    vmovq %rbx, %xmm4
+; AVX1-NEXT:    vmovq %r10, %xmm4
 ; AVX1-NEXT:    vmovq %r9, %xmm5
 ; AVX1-NEXT:    vmovq %r8, %xmm6
-; AVX1-NEXT:    vmovq {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 8-byte Folded Reload
-; AVX1-NEXT:    # xmm7 = mem[0],zero
+; AVX1-NEXT:    vmovq %rsi, %xmm7
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm8 = xmm0[0],xmm8[0]
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm11[0],xmm1[0]
-; AVX1-NEXT:    vshufps {{.*#+}} xmm8 = xmm8[0,2],xmm0[0,2]
+; AVX1-NEXT:    vshufps {{.*#+}} xmm8 = xmm0[0,2],xmm8[0,2]
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm13[0],xmm2[0]
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm15[0],xmm14[0]
-; AVX1-NEXT:    vshufps {{.*#+}} xmm11 = xmm0[0,2],xmm1[0,2]
+; AVX1-NEXT:    vshufps {{.*#+}} xmm11 = xmm1[0,2],xmm0[0,2]
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm8, %xmm0
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm11, %xmm2
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm2[0],xmm0[0]
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm10[0],xmm9[0]
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm12[0]
-; AVX1-NEXT:    vshufps {{.*#+}} xmm2 = xmm2[0,2],xmm3[0,2]
+; AVX1-NEXT:    vshufps {{.*#+}} xmm2 = xmm3[0,2],xmm2[0,2]
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm5[0],xmm4[0]
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
 ; AVX1-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm2
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm5 = xmm7[0],xmm6[0]
-; AVX1-NEXT:    vshufps {{.*#+}} xmm3 = xmm3[0,2],xmm5[0,2]
+; AVX1-NEXT:    vshufps {{.*#+}} xmm3 = xmm5[0,2],xmm3[0,2]
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
 ; AVX1-NEXT:    vpshufb %xmm4, %xmm1, %xmm1
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
 ; AVX1-NEXT:    vmovdqu %xmm0, (%rax)
-; AVX1-NEXT:    addq $24, %rsp
+; AVX1-NEXT:    addq $16, %rsp
 ; AVX1-NEXT:    popq %rbx
 ; AVX1-NEXT:    popq %r12
 ; AVX1-NEXT:    popq %r13
 ; AVX1-NEXT:    popq %r14
 ; AVX1-NEXT:    popq %r15
 ; AVX1-NEXT:    popq %rbp
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: not_avg_v16i8_wide_constants:

test/CodeGen/X86/cast-vsel.ll

 ; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm0 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm5[2,3,0,1]
 ; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: zext:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vcmpltps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm4 = xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm1
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[2,3,0,1]
+; AVX1-NEXT:    vinsertf128 $1, %xmm4, %ymm2, %ymm2
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm2
-; AVX1-NEXT:    vblendvps %ymm0, %ymm1, %ymm2, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm3, %ymm1
+; AVX1-NEXT:    vblendvps %ymm0, %ymm2, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vcmpltps %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm1 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm2 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
 ; AVX2-NEXT:    vblendvps %ymm0, %ymm1, %ymm2, %ymm0

test/CodeGen/X86/madd.ll

 ; SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm1[0,2]
 ; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,3],xmm1[1,3]
 ; SSE2-NEXT:    paddd %xmm2, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; AVX1-LABEL: pmaddwd_negative1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm1[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; AVX1-NEXT:    vpmulld %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
+; AVX1-NEXT:    vpmulld %xmm2, %xmm3, %xmm2
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero
 ; AVX1-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vphaddd %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX256-LABEL: pmaddwd_negative1:
 ; AVX256:       # %bb.0:
 ; AVX256-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero

test/CodeGen/X86/pmaddubsw.ll

 ; SSE-NEXT:    paddd %xmm3, %xmm1
 ; SSE-NEXT:    packssdw %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: pmaddubsw_bad_extend:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vmovdqa (%rdi), %xmm0
 ; AVX1-NEXT:    vmovdqa (%rsi), %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = <8,10,12,14,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm3
+; AVX1-NEXT:    vpmovsxbd %xmm3, %xmm4
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[1,1,2,3]
 ; AVX1-NEXT:    vpmovsxbd %xmm3, %xmm3
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,2,4,6,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm4, %xmm0, %xmm5
-; AVX1-NEXT:    vpmovsxbd %xmm5, %xmm5
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm2
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
+; AVX1-NEXT:    vpmulld %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
 ; AVX1-NEXT:    vpmulld %xmm2, %xmm3, %xmm2
-; AVX1-NEXT:    vpshufb %xmm4, %xmm1, %xmm3
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
-; AVX1-NEXT:    vpmulld %xmm3, %xmm5, %xmm3
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = <9,11,13,15,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm4, %xmm0, %xmm5
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = <1,3,5,7,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm6, %xmm0, %xmm0
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = <1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u>
+; AVX1-NEXT:    vpshufb %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
-; AVX1-NEXT:    vpshufb %xmm4, %xmm1, %xmm4
-; AVX1-NEXT:    vpmovsxbd %xmm4, %xmm4
-; AVX1-NEXT:    vpmulld %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vpaddd %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpshufb %xmm6, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovsxbd %xmm1, %xmm3
+; AVX1-NEXT:    vpmulld %xmm3, %xmm5, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm4, %xmm3
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,2,3]
 ; AVX1-NEXT:    vpmovsxbd %xmm1, %xmm1
 ; AVX1-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpaddd %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpaddd %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm0, %xmm3, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: pmaddubsw_bad_extend:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vmovdqa (%rdi), %xmm0
 ; AVX2-NEXT:    vmovdqa (%rsi), %xmm1
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm3

 ; SSE-NEXT:    paddd %xmm3, %xmm1
 ; SSE-NEXT:    packssdw %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: pmaddubsw_bad_indices:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vmovdqa (%rdi), %xmm0
 ; AVX1-NEXT:    vmovdqa (%rsi), %xmm1
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[9,10,13,14,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[1,2,5,6,9,10,13,14,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpmovsxbd %xmm2, %xmm3
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX1-NEXT:    vpmovsxbd %xmm2, %xmm2
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm3 = xmm0[1,2,5,6,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX1-NEXT:    vpmovsxbd %xmm3, %xmm3
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm4 = xmm1[8,10,12,14,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm4 = xmm1[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
+; AVX1-NEXT:    vpmulld %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm4[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
 ; AVX1-NEXT:    vpmulld %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm4 = xmm1[0,2,4,6,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero
-; AVX1-NEXT:    vpmulld %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm4 = xmm0[8,11,12,15,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX1-NEXT:    vpmovsxbd %xmm4, %xmm4
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,3,4,7,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,3,4,7,8,11,12,15,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpmovsxbd %xmm0, %xmm4
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmovsxbd %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm5 = xmm1[9,11,13,15,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
 ; AVX1-NEXT:    vpmulld %xmm5, %xmm4, %xmm4
-; AVX1-NEXT:    vpaddd %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[1,3,5,7,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpaddd %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
 ; AVX1-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpaddd %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpaddd %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm0, %xmm3, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: pmaddubsw_bad_indices:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vmovdqa (%rdi), %xmm0
 ; AVX2-NEXT:    vmovdqa (%rsi), %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[1,2,5,6,9,10,13,14,u,u,u,u,u,u,u,u]
 ; AVX2-NEXT:    vpmovsxbd %xmm2, %ymm2

test/CodeGen/X86/pr15267.ll

; CHECK-NEXT:    retq
   ret <4 x i1> %ret
 }
 
 define <4 x i64> @test3(<4 x i1>* %in) nounwind {
 ; CHECK-LABEL: test3:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    movzbl (%rdi), %eax
 ; CHECK-NEXT:    movq %rax, %rcx
-; CHECK-NEXT:    shlq $62, %rcx
+; CHECK-NEXT:    shlq $60, %rcx
 ; CHECK-NEXT:    sarq $63, %rcx
-; CHECK-NEXT:    movq %rax, %rdx
-; CHECK-NEXT:    shlq $63, %rdx
-; CHECK-NEXT:    sarq $63, %rdx
-; CHECK-NEXT:    vmovd %edx, %xmm0
-; CHECK-NEXT:    vpinsrd $1, %ecx, %xmm0, %xmm0
+; CHECK-NEXT:    vmovq %rcx, %xmm0
 ; CHECK-NEXT:    movq %rax, %rcx
 ; CHECK-NEXT:    shlq $61, %rcx
 ; CHECK-NEXT:    sarq $63, %rcx
-; CHECK-NEXT:    vpinsrd $2, %ecx, %xmm0, %xmm0
-; CHECK-NEXT:    shlq $60, %rax
+; CHECK-NEXT:    vmovq %rcx, %xmm1
+; CHECK-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; CHECK-NEXT:    movq %rax, %rcx
+; CHECK-NEXT:    shlq $62, %rcx
+; CHECK-NEXT:    sarq $63, %rcx
+; CHECK-NEXT:    vmovq %rcx, %xmm1
+; CHECK-NEXT:    shlq $63, %rax
 ; CHECK-NEXT:    sarq $63, %rax
-; CHECK-NEXT:    vpinsrd $3, %eax, %xmm0, %xmm0
-; CHECK-NEXT:    vpmovsxdq %xmm0, %xmm1
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; CHECK-NEXT:    vpmovsxdq %xmm0, %xmm0
+; CHECK-NEXT:    vmovq %rax, %xmm2
+; CHECK-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
 ; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; CHECK-NEXT:    retq
   %wide.load35 = load <4 x i1>, <4 x i1>* %in, align 1
   %sext = sext <4 x i1> %wide.load35 to <4 x i64>
   ret <4 x i64> %sext
 }
 
 define <16 x i4> @test4(<16 x i4>* %in) nounwind {

test/CodeGen/X86/pr35443.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown -mcpu=skx | FileCheck %s
 
 @ac = external local_unnamed_addr global [20 x i8], align 16
 @ai3 = external local_unnamed_addr global [20 x i32], align 16
 
 ; Function Attrs: norecurse nounwind uwtable
 define void @pr35443() {
 ; CHECK-LABEL: pr35443:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    movzbl ac+{{.*}}(%rip), %eax
-; CHECK-NEXT:    vmovd %eax, %xmm0
+; CHECK-NEXT:    vpmovzxbq {{.*#+}} ymm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero,mem[2],zero,zero,zero,zero,zero,zero,zero,mem[3],zero,zero,zero,zero,zero,zero,zero
 ; CHECK-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; CHECK-NEXT:    vpsubq %ymm0, %ymm1, %ymm0
 ; CHECK-NEXT:    vpmovqd %ymm0, ai3+{{.*}}(%rip)
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
 entry:
   %wide.masked.load66 = call <4 x i8> @llvm.masked.load.v4i8.p0v4i8(<4 x i8>* bitcast (i8* getelementptr inbounds ([20 x i8], [20 x i8]* @ac, i64 0, i64 4) to <4 x i8>*), i32 1, <4 x i1> <i1 true, i1 false, i1 false, i1 false>, <4 x i8> undef)
   %0 = zext <4 x i8> %wide.masked.load66 to <4 x i64>

test/CodeGen/X86/psubus.ll

 ; SSE41-NEXT:    pshufb %xmm6, %xmm4
 ; SSE41-NEXT:    pshufb %xmm6, %xmm3
 ; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm3[0]
 ; SSE41-NEXT:    pandn %xmm4, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test13:
 ; AVX1:       # %bb.0: # %vector.ph
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpmaxud %xmm1, %xmm0, %xmm3
-; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm0, %xmm3
-; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpmaxud %xmm5, %xmm2, %xmm6
-; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm2, %xmm6
-; AVX1-NEXT:    vpxor %xmm4, %xmm6, %xmm4
-; AVX1-NEXT:    vpackssdw %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpsubd %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
+; AVX1-NEXT:    vpmaxud %xmm3, %xmm2, %xmm4
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm2, %xmm4
+; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm5, %xmm5
+; AVX1-NEXT:    vpxor %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpmaxud %xmm1, %xmm0, %xmm6
+; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm0, %xmm6
+; AVX1-NEXT:    vpxor %xmm5, %xmm6, %xmm5
+; AVX1-NEXT:    vpackssdw %xmm4, %xmm5, %xmm4
 ; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpsubd %xmm3, %xmm2, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; AVX1-NEXT:    vpandn %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpandn %xmm0, %xmm4, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test13:
 ; AVX2:       # %bb.0: # %vector.ph
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpmaxud %ymm1, %ymm0, %ymm2
 ; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm0, %ymm2

 ;
 ; AVX1-LABEL: test14:
 ; AVX1:       # %bb.0: # %vector.ph
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm8 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm9 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm5 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm11 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm6
 ; AVX1-NEXT:    vpmaxud %xmm0, %xmm6, %xmm7
 ; AVX1-NEXT:    vpcmpeqd %xmm7, %xmm6, %xmm7
 ; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
 ; AVX1-NEXT:    vpxor %xmm3, %xmm7, %xmm7
 ; AVX1-NEXT:    vpmaxud %xmm11, %xmm2, %xmm4
 ; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm2, %xmm4

 ; SSE41-NEXT:    pshufb %xmm6, %xmm0
 ; SSE41-NEXT:    pshufb %xmm6, %xmm3
 ; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm3[0]
 ; SSE41-NEXT:    pand %xmm4, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test15:
 ; AVX1:       # %bb.0: # %vector.ph
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpminud %xmm1, %xmm0, %xmm3
-; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm0, %xmm3
-; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpminud %xmm5, %xmm2, %xmm6
-; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm2, %xmm6
-; AVX1-NEXT:    vpxor %xmm4, %xmm6, %xmm4
-; AVX1-NEXT:    vpackssdw %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpsubd %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
+; AVX1-NEXT:    vpminud %xmm3, %xmm2, %xmm4
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm2, %xmm4
+; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm5, %xmm5
+; AVX1-NEXT:    vpxor %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpminud %xmm1, %xmm0, %xmm6
+; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm0, %xmm6
+; AVX1-NEXT:    vpxor %xmm5, %xmm6, %xmm5
+; AVX1-NEXT:    vpackssdw %xmm4, %xmm5, %xmm4
 ; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpsubd %xmm3, %xmm2, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; AVX1-NEXT:    vpand %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpand %xmm0, %xmm4, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test15:
 ; AVX2:       # %bb.0: # %vector.ph
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpminud %ymm1, %ymm0, %ymm2
 ; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm0, %ymm2

 ; SSE41-NEXT:    pshufb %xmm6, %xmm0
 ; SSE41-NEXT:    pshufb %xmm6, %xmm3
 ; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm3[0]
 ; SSE41-NEXT:    pand %xmm4, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test16:
 ; AVX1:       # %bb.0: # %vector.ph
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpmaxud %xmm0, %xmm1, %xmm3
-; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm1, %xmm3
-; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
-; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpmaxud %xmm2, %xmm5, %xmm6
-; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm5, %xmm6
-; AVX1-NEXT:    vpxor %xmm4, %xmm6, %xmm4
-; AVX1-NEXT:    vpackssdw %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpsubd %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
+; AVX1-NEXT:    vpmaxud %xmm2, %xmm3, %xmm4
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm3, %xmm4
+; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm5, %xmm5
+; AVX1-NEXT:    vpxor %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpmaxud %xmm0, %xmm1, %xmm6
+; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm1, %xmm6
+; AVX1-NEXT:    vpxor %xmm5, %xmm6, %xmm5
+; AVX1-NEXT:    vpackssdw %xmm4, %xmm5, %xmm4
 ; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpsubd %xmm3, %xmm2, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; AVX1-NEXT:    vpand %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpand %xmm0, %xmm4, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test16:
 ; AVX2:       # %bb.0: # %vector.ph
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpmaxud %ymm0, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm1, %ymm2

 ; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm4
 ; AVX1-NEXT:    vpcmpgtq %xmm4, %xmm5, %xmm4
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm4, %ymm3
 ; AVX1-NEXT:    vblendvpd %ymm3, %ymm1, %ymm6, %ymm1
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
 ; AVX1-NEXT:    vpackusdw %xmm3, %xmm1, %xmm1
 ; AVX1-NEXT:    vpackusdw %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpackusdw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: psubus_8i64_max:
 ; AVX2:       # %bb.0: # %vector.ph
 ; AVX2-NEXT:    vpbroadcastq {{.*#+}} ymm3 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808]
 ; AVX2-NEXT:    vpxor %ymm3, %ymm2, %ymm4
 ; AVX2-NEXT:    vpbroadcastq {{.*#+}} ymm5 = [9223372036854841343,9223372036854841343,9223372036854841343,9223372036854841343]

test/CodeGen/X86/shrink_vmul.ll

 ; X86-AVX1-NEXT:    pushl %ebp
 ; X86-AVX1-NEXT:    .cfi_def_cfa_offset 8
 ; X86-AVX1-NEXT:    pushl %ebx
 ; X86-AVX1-NEXT:    .cfi_def_cfa_offset 12
 ; X86-AVX1-NEXT:    pushl %edi
 ; X86-AVX1-NEXT:    .cfi_def_cfa_offset 16
 ; X86-AVX1-NEXT:    pushl %esi
 ; X86-AVX1-NEXT:    .cfi_def_cfa_offset 20
-; X86-AVX1-NEXT:    subl $16, %esp
-; X86-AVX1-NEXT:    .cfi_def_cfa_offset 36
+; X86-AVX1-NEXT:    subl $8, %esp
+; X86-AVX1-NEXT:    .cfi_def_cfa_offset 28
 ; X86-AVX1-NEXT:    .cfi_offset %esi, -20
 ; X86-AVX1-NEXT:    .cfi_offset %edi, -16
 ; X86-AVX1-NEXT:    .cfi_offset %ebx, -12
 ; X86-AVX1-NEXT:    .cfi_offset %ebp, -8
 ; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-AVX1-NEXT:    vmovdqa (%eax), %ymm2
 ; X86-AVX1-NEXT:    vmovdqa (%ecx), %ymm1
-; X86-AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
-; X86-AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; X86-AVX1-NEXT:    vpxor %xmm0, %xmm0, %xmm0
+; X86-AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
 ; X86-AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
 ; X86-AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm2
 ; X86-AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
 ; X86-AVX1-NEXT:    vmovd %xmm2, %eax
 ; X86-AVX1-NEXT:    xorl %edx, %edx
 ; X86-AVX1-NEXT:    divl 32(%ecx)
 ; X86-AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
 ; X86-AVX1-NEXT:    vpextrd $3, %xmm1, %ecx
 ; X86-AVX1-NEXT:    vpextrd $3, %xmm3, %eax
 ; X86-AVX1-NEXT:    xorl %edx, %edx
 ; X86-AVX1-NEXT:    divl %ecx
-; X86-AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-AVX1-NEXT:    movl %edx, (%esp) # 4-byte Spill
 ; X86-AVX1-NEXT:    vpextrd $2, %xmm1, %ecx
 ; X86-AVX1-NEXT:    vpextrd $2, %xmm3, %eax
 ; X86-AVX1-NEXT:    xorl %edx, %edx
 ; X86-AVX1-NEXT:    divl %ecx
-; X86-AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
+; X86-AVX1-NEXT:    movl %edx, %edi
 ; X86-AVX1-NEXT:    vpextrd $1, %xmm1, %ecx
 ; X86-AVX1-NEXT:    vpextrd $1, %xmm3, %eax
 ; X86-AVX1-NEXT:    xorl %edx, %edx
 ; X86-AVX1-NEXT:    divl %ecx
-; X86-AVX1-NEXT:    movl %edx, (%esp) # 4-byte Spill
+; X86-AVX1-NEXT:    movl %edx, %ebx
 ; X86-AVX1-NEXT:    vmovd %xmm1, %ecx
 ; X86-AVX1-NEXT:    vmovd %xmm3, %eax
 ; X86-AVX1-NEXT:    xorl %edx, %edx
 ; X86-AVX1-NEXT:    divl %ecx
 ; X86-AVX1-NEXT:    movl %edx, %ebp
+; X86-AVX1-NEXT:    vpextrd $3, %xmm0, %eax
 ; X86-AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
-; X86-AVX1-NEXT:    vpextrd $3, %xmm1, %ecx
 ; X86-AVX1-NEXT:    xorl %edx, %edx
-; X86-AVX1-NEXT:    vpextrd $3, %xmm0, %eax
+; X86-AVX1-NEXT:    vpextrd $3, %xmm1, %ecx
 ; X86-AVX1-NEXT:    divl %ecx
-; X86-AVX1-NEXT:    movl %edx, %ebx
+; X86-AVX1-NEXT:    movl %edx, %ecx
 ; X86-AVX1-NEXT:    xorl %edx, %edx
-; X86-AVX1-NEXT:    vpextrd $2, %xmm1, %esi
 ; X86-AVX1-NEXT:    vpextrd $2, %xmm0, %eax
+; X86-AVX1-NEXT:    vpextrd $2, %xmm1, %esi
 ; X86-AVX1-NEXT:    divl %esi
 ; X86-AVX1-NEXT:    movl %edx, %esi
+; X86-AVX1-NEXT:    vmovd %ebp, %xmm2
 ; X86-AVX1-NEXT:    xorl %edx, %edx
-; X86-AVX1-NEXT:    vpextrd $1, %xmm1, %edi
 ; X86-AVX1-NEXT:    vpextrd $1, %xmm0, %eax
-; X86-AVX1-NEXT:    divl %edi
-; X86-AVX1-NEXT:    movl %edx, %edi
-; X86-AVX1-NEXT:    xorl %edx, %edx
-; X86-AVX1-NEXT:    vmovd %xmm1, %ecx
+; X86-AVX1-NEXT:    vpextrd $1, %xmm1, %ebp
+; X86-AVX1-NEXT:    divl %ebp
+; X86-AVX1-NEXT:    movl %edx, %ebp
+; X86-AVX1-NEXT:    vpinsrd $1, %ebx, %xmm2, %xmm2
 ; X86-AVX1-NEXT:    vmovd %xmm0, %eax
-; X86-AVX1-NEXT:    divl %ecx
-; X86-AVX1-NEXT:    vmovd %edx, %xmm0
-; X86-AVX1-NEXT:    vpinsrd $1, %edi, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vpinsrd $2, %esi, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vpinsrd $3, %ebx, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovd %ebp, %xmm1
-; X86-AVX1-NEXT:    vpinsrd $1, (%esp), %xmm1, %xmm1 # 4-byte Folded Reload
-; X86-AVX1-NEXT:    vpinsrd $2, {{[-0-9]+}}(%e{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; X86-AVX1-NEXT:    vpinsrd $3, {{[-0-9]+}}(%e{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
+; X86-AVX1-NEXT:    vpinsrd $2, %edi, %xmm2, %xmm0
+; X86-AVX1-NEXT:    vmovd %xmm1, %edi
+; X86-AVX1-NEXT:    vpinsrd $3, (%esp), %xmm0, %xmm0 # 4-byte Folded Reload
+; X86-AVX1-NEXT:    xorl %edx, %edx
+; X86-AVX1-NEXT:    divl %edi
+; X86-AVX1-NEXT:    vmovd %edx, %xmm1
+; X86-AVX1-NEXT:    vpinsrd $1, %ebp, %xmm1, %xmm1
+; X86-AVX1-NEXT:    vpinsrd $2, %esi, %xmm1, %xmm1
+; X86-AVX1-NEXT:    vpinsrd $3, %ecx, %xmm1, %xmm1
 ; X86-AVX1-NEXT:    vmovd {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 4-byte Folded Reload
 ; X86-AVX1-NEXT:    # xmm2 = mem[0],zero,zero,zero
 ; X86-AVX1-NEXT:    movl $8199, %eax # imm = 0x2007
 ; X86-AVX1-NEXT:    vmovd %eax, %xmm3
 ; X86-AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [8199,8199,8199,8199]
 ; X86-AVX1-NEXT:    vpmulld %xmm4, %xmm0, %xmm0
 ; X86-AVX1-NEXT:    vpmulld %xmm4, %xmm1, %xmm1
-; X86-AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; X86-AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; X86-AVX1-NEXT:    vpmulld %xmm3, %xmm2, %xmm1
 ; X86-AVX1-NEXT:    vmovd %xmm1, (%eax)
 ; X86-AVX1-NEXT:    vmovaps %ymm0, (%eax)
-; X86-AVX1-NEXT:    addl $16, %esp
+; X86-AVX1-NEXT:    addl $8, %esp
 ; X86-AVX1-NEXT:    .cfi_def_cfa_offset 20
 ; X86-AVX1-NEXT:    popl %esi
 ; X86-AVX1-NEXT:    .cfi_def_cfa_offset 16
 ; X86-AVX1-NEXT:    popl %edi
 ; X86-AVX1-NEXT:    .cfi_def_cfa_offset 12
 ; X86-AVX1-NEXT:    popl %ebx
 ; X86-AVX1-NEXT:    .cfi_def_cfa_offset 8
 ; X86-AVX1-NEXT:    popl %ebp

 ; X64-AVX1-NEXT:    pushq %rbp
 ; X64-AVX1-NEXT:    .cfi_def_cfa_offset 16
 ; X64-AVX1-NEXT:    pushq %rbx
 ; X64-AVX1-NEXT:    .cfi_def_cfa_offset 24
 ; X64-AVX1-NEXT:    .cfi_offset %rbx, -24
 ; X64-AVX1-NEXT:    .cfi_offset %rbp, -16
 ; X64-AVX1-NEXT:    vmovdqa (%rdi), %ymm2
 ; X64-AVX1-NEXT:    vmovdqa (%rsi), %ymm1
-; X64-AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
-; X64-AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; X64-AVX1-NEXT:    vpxor %xmm0, %xmm0, %xmm0
+; X64-AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
 ; X64-AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
 ; X64-AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm2
 ; X64-AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
 ; X64-AVX1-NEXT:    vmovd %xmm2, %eax
 ; X64-AVX1-NEXT:    xorl %edx, %edx
 ; X64-AVX1-NEXT:    divl 32(%rsi)
 ; X64-AVX1-NEXT:    movl %edx, %r8d
 ; X64-AVX1-NEXT:    vpextrd $3, %xmm1, %ecx

 ; X64-AVX1-NEXT:    xorl %edx, %edx
 ; X64-AVX1-NEXT:    divl %ecx
 ; X64-AVX1-NEXT:    movl %edx, %r11d
 ; X64-AVX1-NEXT:    vmovd %xmm1, %ecx
 ; X64-AVX1-NEXT:    vmovd %xmm3, %eax
 ; X64-AVX1-NEXT:    xorl %edx, %edx
 ; X64-AVX1-NEXT:    divl %ecx
 ; X64-AVX1-NEXT:    movl %edx, %esi
+; X64-AVX1-NEXT:    vpextrd $3, %xmm0, %eax
 ; X64-AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
 ; X64-AVX1-NEXT:    vpextrd $3, %xmm1, %ecx
-; X64-AVX1-NEXT:    vpextrd $3, %xmm0, %eax
 ; X64-AVX1-NEXT:    xorl %edx, %edx
 ; X64-AVX1-NEXT:    divl %ecx
 ; X64-AVX1-NEXT:    movl %edx, %edi
-; X64-AVX1-NEXT:    vpextrd $2, %xmm1, %ecx
 ; X64-AVX1-NEXT:    vpextrd $2, %xmm0, %eax
+; X64-AVX1-NEXT:    vpextrd $2, %xmm1, %ecx
 ; X64-AVX1-NEXT:    xorl %edx, %edx
 ; X64-AVX1-NEXT:    divl %ecx
 ; X64-AVX1-NEXT:    movl %edx, %ecx
-; X64-AVX1-NEXT:    vpextrd $1, %xmm1, %ebx
 ; X64-AVX1-NEXT:    vpextrd $1, %xmm0, %eax
+; X64-AVX1-NEXT:    vpextrd $1, %xmm1, %ebx
 ; X64-AVX1-NEXT:    xorl %edx, %edx
 ; X64-AVX1-NEXT:    divl %ebx
 ; X64-AVX1-NEXT:    movl %edx, %ebx
-; X64-AVX1-NEXT:    vmovd %xmm1, %ebp
 ; X64-AVX1-NEXT:    vmovd %xmm0, %eax
+; X64-AVX1-NEXT:    vmovd %xmm1, %ebp
 ; X64-AVX1-NEXT:    xorl %edx, %edx
 ; X64-AVX1-NEXT:    divl %ebp
-; X64-AVX1-NEXT:    vmovd %edx, %xmm0
-; X64-AVX1-NEXT:    vpinsrd $1, %ebx, %xmm0, %xmm0
-; X64-AVX1-NEXT:    vpinsrd $2, %ecx, %xmm0, %xmm0
-; X64-AVX1-NEXT:    vpinsrd $3, %edi, %xmm0, %xmm0
+; X64-AVX1-NEXT:    vmovd %esi, %xmm0
+; X64-AVX1-NEXT:    vpinsrd $1, %r11d, %xmm0, %xmm0
+; X64-AVX1-NEXT:    vpinsrd $2, %r10d, %xmm0, %xmm0
+; X64-AVX1-NEXT:    vpinsrd $3, %r9d, %xmm0, %xmm0
 ; X64-AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [8199,8199,8199,8199]
 ; X64-AVX1-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
-; X64-AVX1-NEXT:    vmovd %esi, %xmm2
-; X64-AVX1-NEXT:    vpinsrd $1, %r11d, %xmm2, %xmm2
-; X64-AVX1-NEXT:    vpinsrd $2, %r10d, %xmm2, %xmm2
-; X64-AVX1-NEXT:    vpinsrd $3, %r9d, %xmm2, %xmm2
+; X64-AVX1-NEXT:    vmovd %edx, %xmm2
+; X64-AVX1-NEXT:    vpinsrd $1, %ebx, %xmm2, %xmm2
+; X64-AVX1-NEXT:    vpinsrd $2, %ecx, %xmm2, %xmm2
+; X64-AVX1-NEXT:    vpinsrd $3, %edi, %xmm2, %xmm2
 ; X64-AVX1-NEXT:    vpmulld %xmm1, %xmm2, %xmm1
-; X64-AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; X64-AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; X64-AVX1-NEXT:    vmovd %r8d, %xmm1
 ; X64-AVX1-NEXT:    movl $8199, %eax # imm = 0x2007
 ; X64-AVX1-NEXT:    vmovd %eax, %xmm2
 ; X64-AVX1-NEXT:    vpmulld %xmm2, %xmm1, %xmm1
 ; X64-AVX1-NEXT:    vmovd %xmm1, (%rax)
 ; X64-AVX1-NEXT:    vmovaps %ymm0, (%rax)
 ; X64-AVX1-NEXT:    popq %rbx
 ; X64-AVX1-NEXT:    .cfi_def_cfa_offset 16

test/CodeGen/X86/v8i1-masks.ll

 
 define <8 x i32> @and_mask_constant(<8 x i32> %v0, <8 x i32> %v1) {
 ; X32-LABEL: and_mask_constant:
 ; X32:       ## %bb.0:
 ; X32-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; X32-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; X32-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
 ; X32-NEXT:    vpcmpeqd %xmm2, %xmm0, %xmm0
-; X32-NEXT:    vpackssdw %xmm1, %xmm0, %xmm0
-; X32-NEXT:    vpand LCPI2_0, %xmm0, %xmm0
-; X32-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; X32-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; X32-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; X32-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; X32-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; X32-NEXT:    vandps LCPI2_0, %ymm0, %ymm0
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: and_mask_constant:
 ; X64:       ## %bb.0:
 ; X64-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; X64-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; X64-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
 ; X64-NEXT:    vpcmpeqd %xmm2, %xmm0, %xmm0
-; X64-NEXT:    vpackssdw %xmm1, %xmm0, %xmm0
-; X64-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
-; X64-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; X64-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; X64-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; X64-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; X64-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; X64-NEXT:    vandps {{.*}}(%rip), %ymm0, %ymm0
 ; X64-NEXT:    retq
 ;
 ; X32-AVX2-LABEL: and_mask_constant:
 ; X32-AVX2:       ## %bb.0:
 ; X32-AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; X32-AVX2-NEXT:    vpcmpeqd %ymm1, %ymm0, %ymm0
 ; X32-AVX2-NEXT:    vpand LCPI2_0, %ymm0, %ymm0
 ; X32-AVX2-NEXT:    retl

test/CodeGen/X86/vec_cast2.ll

; CHECK-WIDE-NEXT:    retl
   %res = sitofp <4 x i16> %src to <4 x float>
   ret <4 x float> %res
 }
 
 define <8 x float> @cvt_v8u8_v8f32(<8 x i8> %src) {
 ; CHECK-LABEL: cvt_v8u8_v8f32:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    vpand LCPI4_0, %xmm0, %xmm0
-; CHECK-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; CHECK-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; CHECK-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; CHECK-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; CHECK-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; CHECK-NEXT:    retl
 ;
 ; CHECK-WIDE-LABEL: cvt_v8u8_v8f32:
 ; CHECK-WIDE:       ## %bb.0:
 ; CHECK-WIDE-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; CHECK-WIDE-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; CHECK-WIDE-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; CHECK-WIDE-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; CHECK-WIDE-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; CHECK-WIDE-NEXT:    retl
   %res = uitofp <8 x i8> %src to <8 x float>
   ret <8 x float> %res
 }
 
 define <8 x float> @cvt_v8u16_v8f32(<8 x i16> %src) {
 ; CHECK-LABEL: cvt_v8u16_v8f32:
 ; CHECK:       ## %bb.0:
-; CHECK-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; CHECK-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; CHECK-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; CHECK-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; CHECK-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; CHECK-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; CHECK-NEXT:    retl
 ;
 ; CHECK-WIDE-LABEL: cvt_v8u16_v8f32:
 ; CHECK-WIDE:       ## %bb.0:
-; CHECK-WIDE-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; CHECK-WIDE-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; CHECK-WIDE-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; CHECK-WIDE-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; CHECK-WIDE-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; CHECK-WIDE-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; CHECK-WIDE-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; CHECK-WIDE-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; CHECK-WIDE-NEXT:    retl
   %res = uitofp <8 x i16> %src to <8 x float>
   ret <8 x float> %res
 }
 
 define <4 x float> @cvt_v4u8_v4f32(<4 x i8> %src) {
 ; CHECK-LABEL: cvt_v4u8_v4f32:

test/CodeGen/X86/vec_int_to_fp.ll

 ; SSE41-LABEL: uitofp_8i16_to_4f32:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; SSE41-NEXT:    cvtdq2ps %xmm0, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: uitofp_8i16_to_4f32:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; AVX1-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: uitofp_8i16_to_4f32:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero

 ; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; SSE41-NEXT:    cvtdq2ps %xmm0, %xmm1
 ; SSE41-NEXT:    movaps %xmm2, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: uitofp_8i16_to_8f32:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: uitofp_8i16_to_8f32:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; AVX2-NEXT:    retq

 ; SSE41-NEXT:    cvtdq2ps %xmm0, %xmm0
 ; SSE41-NEXT:    movaps %xmm0, (%rax)
 ; SSE41-NEXT:    movaps %xmm1, 16(%rax)
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: aggregate_sitofp_8i16_to_8f32:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    movq 24(%rdi), %rax
-; AVX1-NEXT:    vmovdqu 8(%rdi), %xmm0
-; AVX1-NEXT:    vpmovsxwd %xmm0, %xmm1
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovsxwd %xmm0, %xmm0
+; AVX1-NEXT:    vpmovsxwd 16(%rdi), %xmm0
+; AVX1-NEXT:    vpmovsxwd 8(%rdi), %xmm1
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; AVX1-NEXT:    vmovaps %ymm0, (%rax)
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: aggregate_sitofp_8i16_to_8f32:
 ; AVX2:       # %bb.0:

test/CodeGen/X86/vector-sext.ll

 ; AVX1-LABEL: sext_16i8_to_16i32:
 ; AVX1:       # %bb.0: # %entry
 ; AVX1-NEXT:    vpmovsxbd %xmm0, %xmm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmovsxbd %xmm2, %xmm2
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovsxbd %xmm1, %xmm1
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,0,1]
 ; AVX1-NEXT:    vpmovsxbd %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
 ; AVX1-NEXT:    vmovaps %ymm2, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: sext_16i8_to_16i32:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovsxbd %xmm0, %ymm2

 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: sext_16i8_to_8i64:
 ; AVX1:       # %bb.0: # %entry
 ; AVX1-NEXT:    vpmovsxbq %xmm0, %xmm1
 ; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm2
 ; AVX1-NEXT:    vpmovsxbq %xmm2, %xmm2
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm2
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX1-NEXT:    vpmovsxbq %xmm1, %xmm1
-; AVX1-NEXT:    vpsrlq $48, %xmm0, %xmm0
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
+; AVX1-NEXT:    vpmovsxbq %xmm0, %xmm1
+; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovsxbq %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
 ; AVX1-NEXT:    vmovaps %ymm2, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: sext_16i8_to_8i64:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovsxbq %xmm0, %ymm2

 ; AVX1-LABEL: sext_8i16_to_8i64:
 ; AVX1:       # %bb.0: # %entry
 ; AVX1-NEXT:    vpmovsxwq %xmm0, %xmm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmovsxwq %xmm2, %xmm2
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovsxwq %xmm1, %xmm1
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,0,1]
 ; AVX1-NEXT:    vpmovsxwq %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
 ; AVX1-NEXT:    vmovaps %ymm2, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: sext_8i16_to_8i64:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovsxwq %xmm0, %ymm2

 ; SSE41-NEXT:    psrad $31, %xmm1
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: load_sext_4i1_to_4i64:
 ; AVX1:       # %bb.0: # %entry
 ; AVX1-NEXT:    movzbl (%rdi), %eax
 ; AVX1-NEXT:    movq %rax, %rcx
-; AVX1-NEXT:    shlq $62, %rcx
+; AVX1-NEXT:    shlq $60, %rcx
 ; AVX1-NEXT:    sarq $63, %rcx
-; AVX1-NEXT:    movq %rax, %rdx
-; AVX1-NEXT:    shlq $63, %rdx
-; AVX1-NEXT:    sarq $63, %rdx
-; AVX1-NEXT:    vmovd %edx, %xmm0
-; AVX1-NEXT:    vpinsrd $1, %ecx, %xmm0, %xmm0
+; AVX1-NEXT:    vmovq %rcx, %xmm0
 ; AVX1-NEXT:    movq %rax, %rcx
 ; AVX1-NEXT:    shlq $61, %rcx
 ; AVX1-NEXT:    sarq $63, %rcx
-; AVX1-NEXT:    vpinsrd $2, %ecx, %xmm0, %xmm0
-; AVX1-NEXT:    shlq $60, %rax
+; AVX1-NEXT:    vmovq %rcx, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX1-NEXT:    movq %rax, %rcx
+; AVX1-NEXT:    shlq $62, %rcx
+; AVX1-NEXT:    sarq $63, %rcx
+; AVX1-NEXT:    vmovq %rcx, %xmm1
+; AVX1-NEXT:    shlq $63, %rax
 ; AVX1-NEXT:    sarq $63, %rax
-; AVX1-NEXT:    vpinsrd $3, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovsxdq %xmm0, %xmm1
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovsxdq %xmm0, %xmm0
+; AVX1-NEXT:    vmovq %rax, %xmm2
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: load_sext_4i1_to_4i64:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    movzbl (%rdi), %eax
 ; AVX2-NEXT:    movq %rax, %rcx
 ; AVX2-NEXT:    shlq $60, %rcx

test/CodeGen/X86/vector-zext.ll

 ; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; SSE41-NEXT:    movdqa %xmm2, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: zext_16i8_to_16i16:
 ; AVX1:       # %bb.0: # %entry
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15]
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext_16i8_to_16i16:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: zext_16i8_to_16i16:

 ; AVX1-LABEL: zext_16i8_to_16i32:
 ; AVX1:       # %bb.0: # %entry
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
 ; AVX1-NEXT:    vmovaps %ymm2, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext_16i8_to_16i32:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero

 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: zext_16i8_to_8i64:
 ; AVX1:       # %bb.0: # %entry
 ; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
 ; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm2
 ; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,zero,zero,zero,zero,xmm2[1],zero,zero,zero,zero,zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm2
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
-; AVX1-NEXT:    vpsrlq $48, %xmm0, %xmm0
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
+; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
 ; AVX1-NEXT:    vmovaps %ymm2, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext_16i8_to_8i64:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovzxbq {{.*#+}} ymm2 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero,xmm0[2],zero,zero,zero,zero,zero,zero,zero,xmm0[3],zero,zero,zero,zero,zero,zero,zero

 ; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; SSE41-NEXT:    movdqa %xmm2, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: zext_8i16_to_8i32:
 ; AVX1:       # %bb.0: # %entry
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext_8i16_to_8i32:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: zext_8i16_to_8i32:

 ; AVX1-LABEL: zext_8i16_to_8i64:
 ; AVX1:       # %bb.0: # %entry
 ; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
 ; AVX1-NEXT:    vmovaps %ymm2, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext_8i16_to_8i64:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovzxwq {{.*#+}} ymm2 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero

 ; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero
 ; SSE41-NEXT:    movdqa %xmm2, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: zext_4i32_to_4i64:
 ; AVX1:       # %bb.0: # %entry
-; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm1 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
 ; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext_4i32_to_4i64:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpmovzxdq {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: zext_4i32_to_4i64:

 ; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; SSE41-NEXT:    movdqa %xmm2, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: zext_8i8_to_8i32:
 ; AVX1:       # %bb.0: # %entry
 ; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext_8i8_to_8i32:
 ; AVX2:       # %bb.0: # %entry
 ; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    retq
 ;

 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm4
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm3[1,1,2,3]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm3[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[3,1,2,3]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[3,3,0,1]
 ; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm3
 ; AVX1-NEXT:    vmovaps %ymm4, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: zext_32i8_to_32i32:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm3 = xmm1[3,1,2,3]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
-; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
-; AVX2-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm3
 ; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm4 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
 ; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[3,1,2,3]
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm3 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
 ; AVX2-NEXT:    vmovdqa %ymm4, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: zext_32i8_to_32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpmovzxbd {{.*#+}} zmm2 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm0
 ; AVX512-NEXT:    vpmovzxbd {{.*#+}} zmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero

test/CodeGen/X86/vselect-avx.ll

 ; When shrinking the condition used into the select to match a blend, this
 ; test case exercises the path where the modified node is not the root
 ; of the condition.
 
 define void @test2(double** %call1559, i64 %indvars.iv4198, <4 x i1> %tmp1895) {
 ; AVX1-LABEL: test2:
 ; AVX1:       ## %bb.0: ## %bb
 ; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX1-NEXT:    vpsrad $31, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovsxdq %xmm0, %xmm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovsxdq %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    movq (%rdi,%rsi,8), %rax
 ; AVX1-NEXT:    vmovapd {{.*#+}} ymm1 = [0.5,0.5,0.5,0.5]
 ; AVX1-NEXT:    vblendvpd %ymm0, {{.*}}(%rip), %ymm1, %ymm0
 ; AVX1-NEXT:    vmovupd %ymm0, (%rax)