From 603f94aa2a1b87b3e1c5722a3ba270147111e70d Mon Sep 17 00:00:00 2001
From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Sat, 27 Jul 2019 14:11:59 +0000
Subject: [PATCH] [TargetLowering] SimplifyMultipleUseDemandedBits - add
 BITCAST pass through support (Reapplied)

This allows us to peek through BITCASTs, attempt to simplify the source operand, and then bitcast back.

This reapplies rL367091 which was reverted at rL367118 - we were inconsistently peeking through the bitcasts to the source value.

Fixes PR42777

llvm-svn: 367174
---
 .../CodeGen/SelectionDAG/TargetLowering.cpp   |  61 ++-
 .../CodeGen/X86/urem-seteq-vec-nonsplat.ll    | 396 ++++++++----------
 .../CodeGen/X86/vector-reduce-mul-widen.ll    | 103 ++---
 llvm/test/CodeGen/X86/vector-reduce-mul.ll    | 115 ++---
 llvm/test/CodeGen/X86/vector-reduce-umax.ll   | 115 +++--
 llvm/test/CodeGen/X86/vector-reduce-umin.ll   |  98 ++---
 6 files changed, 431 insertions(+), 457 deletions(-)

diff --git a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
index 92706e1f1169..9633a8c873a9 100644
--- a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -565,15 +565,72 @@ bool TargetLowering::SimplifyDemandedBits(SDValue Op, const APInt &DemandedBits,
 }
 
 // TODO: Can we merge SelectionDAG::GetDemandedBits into this?
-// TODO: Under what circumstances can we create nodes? BITCAST? Constant?
+// TODO: Under what circumstances can we create nodes? Constant folding?
 SDValue TargetLowering::SimplifyMultipleUseDemandedBits(
     SDValue Op, const APInt &DemandedBits, const APInt &DemandedElts,
     SelectionDAG &DAG, unsigned Depth) const {
   if (Depth >= 6) // Limit search depth.
     return SDValue();
 
+  unsigned NumElts = DemandedElts.getBitWidth();
   KnownBits LHSKnown, RHSKnown;
   switch (Op.getOpcode()) {
+  case ISD::BITCAST: {
+    SDValue Src = peekThroughBitcasts(Op.getOperand(0));
+    EVT SrcVT = Src.getValueType();
+    EVT DstVT = Op.getValueType();
+    unsigned NumSrcEltBits = SrcVT.getScalarSizeInBits();
+    unsigned NumDstEltBits = DstVT.getScalarSizeInBits();
+
+    if (NumSrcEltBits == NumDstEltBits)
+      if (SDValue V = SimplifyMultipleUseDemandedBits(
+              Src, DemandedBits, DemandedElts, DAG, Depth + 1))
+        return DAG.getBitcast(DstVT, V);
+
+    // TODO - bigendian once we have test coverage.
+    if (SrcVT.isVector() && (NumDstEltBits % NumSrcEltBits) == 0 &&
+        DAG.getDataLayout().isLittleEndian()) {
+      unsigned Scale = NumDstEltBits / NumSrcEltBits;
+      unsigned NumSrcElts = SrcVT.getVectorNumElements();
+      APInt DemandedSrcBits = APInt::getNullValue(NumSrcEltBits);
+      APInt DemandedSrcElts = APInt::getNullValue(NumSrcElts);
+      for (unsigned i = 0; i != Scale; ++i) {
+        unsigned Offset = i * NumSrcEltBits;
+        APInt Sub = DemandedBits.extractBits(NumSrcEltBits, Offset);
+        if (!Sub.isNullValue()) {
+          DemandedSrcBits |= Sub;
+          for (unsigned j = 0; j != NumElts; ++j)
+            if (DemandedElts[j])
+              DemandedSrcElts.setBit((j * Scale) + i);
+        }
+      }
+
+      if (SDValue V = SimplifyMultipleUseDemandedBits(
+              Src, DemandedSrcBits, DemandedSrcElts, DAG, Depth + 1))
+        return DAG.getBitcast(DstVT, V);
+    }
+
+    // TODO - bigendian once we have test coverage.
+    if ((NumSrcEltBits % NumDstEltBits) == 0 &&
+        DAG.getDataLayout().isLittleEndian()) {
+      unsigned Scale = NumSrcEltBits / NumDstEltBits;
+      unsigned NumSrcElts = SrcVT.isVector() ? SrcVT.getVectorNumElements() : 1;
+      APInt DemandedSrcBits = APInt::getNullValue(NumSrcEltBits);
+      APInt DemandedSrcElts = APInt::getNullValue(NumSrcElts);
+      for (unsigned i = 0; i != NumElts; ++i)
+        if (DemandedElts[i]) {
+          unsigned Offset = (i % Scale) * NumDstEltBits;
+          DemandedSrcBits.insertBits(DemandedBits, Offset);
+          DemandedSrcElts.setBit(i / Scale);
+        }
+
+      if (SDValue V = SimplifyMultipleUseDemandedBits(
+              Src, DemandedSrcBits, DemandedSrcElts, DAG, Depth + 1))
+        return DAG.getBitcast(DstVT, V);
+    }
+
+    break;
+  }
   case ISD::AND: {
     LHSKnown = DAG.computeKnownBits(Op.getOperand(0), DemandedElts, Depth + 1);
     RHSKnown = DAG.computeKnownBits(Op.getOperand(1), DemandedElts, Depth + 1);
@@ -625,7 +682,7 @@ SDValue TargetLowering::SimplifyMultipleUseDemandedBits(
     // If all the demanded elts are from one operand and are inline,
     // then we can use the operand directly.
     bool AllUndef = true, IdentityLHS = true, IdentityRHS = true;
-    for (unsigned i = 0, NumElts = ShuffleMask.size(); i != NumElts; ++i) {
+    for (unsigned i = 0; i != NumElts; ++i) {
       int M = ShuffleMask[i];
       if (M < 0 || !DemandedElts[i])
         continue;
diff --git a/llvm/test/CodeGen/X86/urem-seteq-vec-nonsplat.ll b/llvm/test/CodeGen/X86/urem-seteq-vec-nonsplat.ll
index c1669cf4adb2..8717944e83e8 100644
--- a/llvm/test/CodeGen/X86/urem-seteq-vec-nonsplat.ll
+++ b/llvm/test/CodeGen/X86/urem-seteq-vec-nonsplat.ll
@@ -11,32 +11,30 @@
 define <4 x i32> @test_urem_odd_even(<4 x i32> %X) nounwind {
 ; CHECK-SSE2-LABEL: test_urem_odd_even:
 ; CHECK-SSE2:       # %bb.0:
-; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; CHECK-SSE2-NEXT:    psrld $1, %xmm1
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,1374389535,1374389535]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,3,2,3]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [3435973837,2454267027,1374389535,1374389535]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm3
+; CHECK-SSE2-NEXT:    psrld $1, %xmm3
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm0[3,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    psrld $5, %xmm2
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm3
+; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psrld $5, %xmm1
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm3
 ; CHECK-SSE2-NEXT:    psrld $2, %xmm3
-; CHECK-SSE2-NEXT:    psrld $3, %xmm1
-; CHECK-SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm3[0],xmm1[1]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm2[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [5,14,25,100]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
+; CHECK-SSE2-NEXT:    psrld $3, %xmm2
+; CHECK-SSE2-NEXT:    movsd {{.*#+}} xmm2 = xmm3[0],xmm2[1]
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm1[3,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [5,14,25,100]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm4
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[0,2,2,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
@@ -48,14 +46,13 @@ define <4 x i32> @test_urem_odd_even(<4 x i32> %X) nounwind {
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    movdqa %xmm0, %xmm1
 ; CHECK-SSE41-NEXT:    psrld $1, %xmm1
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,1374389535,1374389535]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE41-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE41-NEXT:    pmuludq %xmm3, %xmm1
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3],xmm1[4,5],xmm4[6,7]
+; CHECK-SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [3435973837,2454267027,1374389535,1374389535]
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pmuludq %xmm1, %xmm3
+; CHECK-SSE41-NEXT:    pmuludq %xmm0, %xmm2
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm2
 ; CHECK-SSE41-NEXT:    psrld $5, %xmm2
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm3
@@ -74,14 +71,13 @@ define <4 x i32> @test_urem_odd_even(<4 x i32> %X) nounwind {
 ; CHECK-AVX1-LABEL: test_urem_odd_even:
 ; CHECK-AVX1:       # %bb.0:
 ; CHECK-AVX1-NEXT:    vpsrld $1, %xmm0, %xmm1
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [3435973837,2454267027,1374389535,1374389535]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpmuludq %xmm4, %xmm2, %xmm2
+; CHECK-AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [3435973837,2454267027,1374389535,1374389535]
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
 ; CHECK-AVX1-NEXT:    vpmuludq %xmm3, %xmm1, %xmm1
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3],xmm1[4,5],xmm2[6,7]
+; CHECK-AVX1-NEXT:    vpmuludq %xmm2, %xmm0, %xmm2
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3],xmm2[4,5],xmm1[6,7]
 ; CHECK-AVX1-NEXT:    vpsrld $5, %xmm1, %xmm2
 ; CHECK-AVX1-NEXT:    vpsrld $2, %xmm1, %xmm3
 ; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3],xmm2[4,5,6,7]
@@ -410,32 +406,30 @@ define <4 x i32> @test_urem_even_allones_ne(<4 x i32> %X) nounwind {
 define <4 x i32> @test_urem_odd_even_allones_eq(<4 x i32> %X) nounwind {
 ; CHECK-SSE2-LABEL: test_urem_odd_even_allones_eq:
 ; CHECK-SSE2:       # %bb.0:
-; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; CHECK-SSE2-NEXT:    psrld $1, %xmm1
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,2147483649,1374389535]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,3,2,3]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [3435973837,2454267027,2147483649,1374389535]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm3
+; CHECK-SSE2-NEXT:    psrld $1, %xmm3
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm0[3,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    psrld $5, %xmm2
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm3
+; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psrld $5, %xmm1
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm3
 ; CHECK-SSE2-NEXT:    psrld $2, %xmm3
-; CHECK-SSE2-NEXT:    psrld $31, %xmm1
-; CHECK-SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm3[0],xmm1[1]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm2[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [5,14,4294967295,100]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
+; CHECK-SSE2-NEXT:    psrld $31, %xmm2
+; CHECK-SSE2-NEXT:    movsd {{.*#+}} xmm2 = xmm3[0],xmm2[1]
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm1[3,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [5,14,4294967295,100]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm4
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[0,2,2,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
@@ -447,14 +441,13 @@ define <4 x i32> @test_urem_odd_even_allones_eq(<4 x i32> %X) nounwind {
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    movdqa %xmm0, %xmm1
 ; CHECK-SSE41-NEXT:    psrld $1, %xmm1
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,2147483649,1374389535]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE41-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE41-NEXT:    pmuludq %xmm3, %xmm1
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3],xmm1[4,5],xmm4[6,7]
+; CHECK-SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [3435973837,2454267027,2147483649,1374389535]
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pmuludq %xmm1, %xmm3
+; CHECK-SSE41-NEXT:    pmuludq %xmm0, %xmm2
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm2
 ; CHECK-SSE41-NEXT:    psrld $5, %xmm2
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm3
@@ -473,14 +466,13 @@ define <4 x i32> @test_urem_odd_even_allones_eq(<4 x i32> %X) nounwind {
 ; CHECK-AVX1-LABEL: test_urem_odd_even_allones_eq:
 ; CHECK-AVX1:       # %bb.0:
 ; CHECK-AVX1-NEXT:    vpsrld $1, %xmm0, %xmm1
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [3435973837,2454267027,2147483649,1374389535]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpmuludq %xmm4, %xmm2, %xmm2
+; CHECK-AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [3435973837,2454267027,2147483649,1374389535]
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
 ; CHECK-AVX1-NEXT:    vpmuludq %xmm3, %xmm1, %xmm1
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3],xmm1[4,5],xmm2[6,7]
+; CHECK-AVX1-NEXT:    vpmuludq %xmm2, %xmm0, %xmm2
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3],xmm2[4,5],xmm1[6,7]
 ; CHECK-AVX1-NEXT:    vpsrld $5, %xmm1, %xmm2
 ; CHECK-AVX1-NEXT:    vpsrld $2, %xmm1, %xmm3
 ; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3],xmm2[4,5,6,7]
@@ -528,32 +520,30 @@ define <4 x i32> @test_urem_odd_even_allones_eq(<4 x i32> %X) nounwind {
 define <4 x i32> @test_urem_odd_even_allones_ne(<4 x i32> %X) nounwind {
 ; CHECK-SSE2-LABEL: test_urem_odd_even_allones_ne:
 ; CHECK-SSE2:       # %bb.0:
-; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; CHECK-SSE2-NEXT:    psrld $1, %xmm1
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,2147483649,1374389535]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,3,2,3]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [3435973837,2454267027,2147483649,1374389535]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm3
+; CHECK-SSE2-NEXT:    psrld $1, %xmm3
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm0[3,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    psrld $5, %xmm2
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm3
+; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psrld $5, %xmm1
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm3
 ; CHECK-SSE2-NEXT:    psrld $2, %xmm3
-; CHECK-SSE2-NEXT:    psrld $31, %xmm1
-; CHECK-SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm3[0],xmm1[1]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm2[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [5,14,4294967295,100]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
+; CHECK-SSE2-NEXT:    psrld $31, %xmm2
+; CHECK-SSE2-NEXT:    movsd {{.*#+}} xmm2 = xmm3[0],xmm2[1]
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm1[3,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [5,14,4294967295,100]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm4
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[0,2,2,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
@@ -565,14 +555,13 @@ define <4 x i32> @test_urem_odd_even_allones_ne(<4 x i32> %X) nounwind {
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    movdqa %xmm0, %xmm1
 ; CHECK-SSE41-NEXT:    psrld $1, %xmm1
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,2147483649,1374389535]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE41-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE41-NEXT:    pmuludq %xmm3, %xmm1
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3],xmm1[4,5],xmm4[6,7]
+; CHECK-SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [3435973837,2454267027,2147483649,1374389535]
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pmuludq %xmm1, %xmm3
+; CHECK-SSE41-NEXT:    pmuludq %xmm0, %xmm2
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm2
 ; CHECK-SSE41-NEXT:    psrld $5, %xmm2
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm3
@@ -591,14 +580,13 @@ define <4 x i32> @test_urem_odd_even_allones_ne(<4 x i32> %X) nounwind {
 ; CHECK-AVX1-LABEL: test_urem_odd_even_allones_ne:
 ; CHECK-AVX1:       # %bb.0:
 ; CHECK-AVX1-NEXT:    vpsrld $1, %xmm0, %xmm1
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [3435973837,2454267027,2147483649,1374389535]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpmuludq %xmm4, %xmm2, %xmm2
+; CHECK-AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [3435973837,2454267027,2147483649,1374389535]
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
 ; CHECK-AVX1-NEXT:    vpmuludq %xmm3, %xmm1, %xmm1
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3],xmm1[4,5],xmm2[6,7]
+; CHECK-AVX1-NEXT:    vpmuludq %xmm2, %xmm0, %xmm2
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3],xmm2[4,5],xmm1[6,7]
 ; CHECK-AVX1-NEXT:    vpsrld $5, %xmm1, %xmm2
 ; CHECK-AVX1-NEXT:    vpsrld $2, %xmm1, %xmm3
 ; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3],xmm2[4,5,6,7]
@@ -840,31 +828,29 @@ define <4 x i32> @test_urem_even_poweroftwo(<4 x i32> %X) nounwind {
 define <4 x i32> @test_urem_odd_even_poweroftwo(<4 x i32> %X) nounwind {
 ; CHECK-SSE2-LABEL: test_urem_odd_even_poweroftwo:
 ; CHECK-SSE2:       # %bb.0:
-; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; CHECK-SSE2-NEXT:    psrld $1, %xmm1
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,268435456,1374389535]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,3,2,3]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [3435973837,2454267027,268435456,1374389535]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm3
+; CHECK-SSE2-NEXT:    psrld $1, %xmm3
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm0[3,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    psrld $5, %xmm2
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm3
+; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psrld $5, %xmm1
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm3
 ; CHECK-SSE2-NEXT:    psrld $2, %xmm3
-; CHECK-SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm3[0],xmm1[1]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm2[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [5,14,16,100]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
+; CHECK-SSE2-NEXT:    movsd {{.*#+}} xmm2 = xmm3[0],xmm2[1]
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm1[3,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [5,14,16,100]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm4
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[0,2,2,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
@@ -876,14 +862,13 @@ define <4 x i32> @test_urem_odd_even_poweroftwo(<4 x i32> %X) nounwind {
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    movdqa %xmm0, %xmm1
 ; CHECK-SSE41-NEXT:    psrld $1, %xmm1
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,268435456,1374389535]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE41-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE41-NEXT:    pmuludq %xmm3, %xmm1
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3],xmm1[4,5],xmm4[6,7]
+; CHECK-SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [3435973837,2454267027,268435456,1374389535]
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pmuludq %xmm1, %xmm3
+; CHECK-SSE41-NEXT:    pmuludq %xmm0, %xmm2
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm2
 ; CHECK-SSE41-NEXT:    psrld $5, %xmm2
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm3
@@ -901,14 +886,13 @@ define <4 x i32> @test_urem_odd_even_poweroftwo(<4 x i32> %X) nounwind {
 ; CHECK-AVX1-LABEL: test_urem_odd_even_poweroftwo:
 ; CHECK-AVX1:       # %bb.0:
 ; CHECK-AVX1-NEXT:    vpsrld $1, %xmm0, %xmm1
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [3435973837,2454267027,268435456,1374389535]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpmuludq %xmm4, %xmm2, %xmm2
+; CHECK-AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [3435973837,2454267027,268435456,1374389535]
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
 ; CHECK-AVX1-NEXT:    vpmuludq %xmm3, %xmm1, %xmm1
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3],xmm1[4,5],xmm2[6,7]
+; CHECK-AVX1-NEXT:    vpmuludq %xmm2, %xmm0, %xmm2
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3],xmm2[4,5],xmm1[6,7]
 ; CHECK-AVX1-NEXT:    vpsrld $5, %xmm1, %xmm2
 ; CHECK-AVX1-NEXT:    vpsrld $2, %xmm1, %xmm3
 ; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3],xmm2[4,5,6,7]
@@ -1122,33 +1106,31 @@ define <4 x i32> @test_urem_even_one(<4 x i32> %X) nounwind {
 define <4 x i32> @test_urem_odd_even_one(<4 x i32> %X) nounwind {
 ; CHECK-SSE2-LABEL: test_urem_odd_even_one:
 ; CHECK-SSE2:       # %bb.0:
-; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; CHECK-SSE2-NEXT:    psrld $1, %xmm1
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,0,1374389535]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,3,2,3]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [3435973837,2454267027,0,1374389535]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm3
+; CHECK-SSE2-NEXT:    psrld $1, %xmm3
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm0[3,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    psrld $5, %xmm2
-; CHECK-SSE2-NEXT:    psrld $2, %xmm1
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm3
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm2[3,3]
+; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psrld $5, %xmm1
+; CHECK-SSE2-NEXT:    psrld $2, %xmm2
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm3
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,1],xmm1[3,3]
 ; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [5,14,1,100]
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm5
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm5[0,2,2,3]
 ; CHECK-SSE2-NEXT:    movaps %xmm0, %xmm5
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm5 = xmm5[2,0],xmm2[3,0]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm5[0,2]
-; CHECK-SSE2-NEXT:    pmuludq %xmm4, %xmm1
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm5 = xmm5[2,0],xmm1[3,0]
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,1],xmm5[0,2]
+; CHECK-SSE2-NEXT:    pmuludq %xmm4, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
@@ -1160,14 +1142,13 @@ define <4 x i32> @test_urem_odd_even_one(<4 x i32> %X) nounwind {
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    movdqa %xmm0, %xmm1
 ; CHECK-SSE41-NEXT:    psrld $1, %xmm1
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [3435973837,2454267027,0,1374389535]
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-SSE41-NEXT:    pmuludq %xmm2, %xmm4
-; CHECK-SSE41-NEXT:    pmuludq %xmm3, %xmm1
-; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3],xmm1[4,5],xmm4[6,7]
+; CHECK-SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [3435973837,2454267027,0,1374389535]
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pmuludq %xmm1, %xmm3
+; CHECK-SSE41-NEXT:    pmuludq %xmm0, %xmm2
+; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; CHECK-SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
 ; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm2
 ; CHECK-SSE41-NEXT:    psrld $5, %xmm2
 ; CHECK-SSE41-NEXT:    psrld $2, %xmm1
@@ -1183,14 +1164,13 @@ define <4 x i32> @test_urem_odd_even_one(<4 x i32> %X) nounwind {
 ; CHECK-AVX1-LABEL: test_urem_odd_even_one:
 ; CHECK-AVX1:       # %bb.0:
 ; CHECK-AVX1-NEXT:    vpsrld $1, %xmm0, %xmm1
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [3435973837,2454267027,0,1374389535]
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpmuludq %xmm4, %xmm2, %xmm2
+; CHECK-AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [3435973837,2454267027,0,1374389535]
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
 ; CHECK-AVX1-NEXT:    vpmuludq %xmm3, %xmm1, %xmm1
-; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3],xmm1[4,5],xmm2[6,7]
+; CHECK-AVX1-NEXT:    vpmuludq %xmm2, %xmm0, %xmm2
+; CHECK-AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3],xmm2[4,5],xmm1[6,7]
 ; CHECK-AVX1-NEXT:    vpsrld $5, %xmm1, %xmm2
 ; CHECK-AVX1-NEXT:    vpsrld $2, %xmm1, %xmm1
 ; CHECK-AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm2[4,5,6,7]
@@ -2180,27 +2160,25 @@ define <4 x i32> @test_urem_odd_allones_and_poweroftwo_and_one(<4 x i32> %X) nou
 ; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [3435973837,2147483649,268435456,0]
 ; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,3,2,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm4
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm4[1,3,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
-; CHECK-SSE2-NEXT:    movdqa %xmm3, %xmm1
-; CHECK-SSE2-NEXT:    psrld $31, %xmm1
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [5,4294967295,16,1]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm5
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm5[0,2,2,3]
-; CHECK-SSE2-NEXT:    psrld $2, %xmm3
-; CHECK-SSE2-NEXT:    movaps %xmm0, %xmm5
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm5 = xmm5[3,0],xmm2[3,3]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,3],xmm5[2,0]
-; CHECK-SSE2-NEXT:    pmuludq %xmm4, %xmm3
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,3,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; CHECK-SSE2-NEXT:    psubd %xmm2, %xmm0
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psrld $2, %xmm1
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,3],xmm2[2,1]
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [5,4294967295,16,1]
+; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm1
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; CHECK-SSE2-NEXT:    psrld $31, %xmm2
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[3,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm3
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
+; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
 ; CHECK-SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    psrld $31, %xmm0
@@ -2297,20 +2275,18 @@ define <4 x i32> @test_urem_even_allones_and_poweroftwo_and_one(<4 x i32> %X) no
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm4[1,3,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    psrld $31, %xmm2
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[3,3]
+; CHECK-SSE2-NEXT:    psrld $2, %xmm2
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,3],xmm1[2,1]
 ; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [14,4294967295,16,1]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm2, %xmm5
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm5[0,2,2,3]
-; CHECK-SSE2-NEXT:    psrld $2, %xmm1
-; CHECK-SSE2-NEXT:    movaps %xmm0, %xmm5
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm5 = xmm5[3,0],xmm4[3,3]
-; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,3],xmm5[2,0]
-; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm1
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
+; CHECK-SSE2-NEXT:    pmuludq %xmm3, %xmm2
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
+; CHECK-SSE2-NEXT:    psrld $31, %xmm1
+; CHECK-SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3]
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
+; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[0,2,2,3]
+; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; CHECK-SSE2-NEXT:    psubd %xmm2, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
 ; CHECK-SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    psrld $31, %xmm0
diff --git a/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll b/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
index 1b81d04c6e2b..ce73ad545d1a 100644
--- a/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
@@ -814,12 +814,8 @@ define i32 @test_v4i32(<4 x i32> %a0) {
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[3,3,1,1]
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm2, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
+; SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    retq
 ;
@@ -856,23 +852,16 @@ define i32 @test_v4i32(<4 x i32> %a0) {
 define i32 @test_v8i32(<8 x i32> %a0) {
 ; SSE2-LABEL: test_v8i32:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm2, %xmm3
 ; SSE2-NEXT:    pmuludq %xmm1, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,2,0,0]
-; SSE2-NEXT:    pmuludq %xmm1, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    movd %xmm1, %eax
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[2,2,0,0]
+; SSE2-NEXT:    pmuludq %xmm3, %xmm0
+; SSE2-NEXT:    pmuludq %xmm1, %xmm0
+; SSE2-NEXT:    movd %xmm0, %eax
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_v8i32:
@@ -927,29 +916,22 @@ define i32 @test_v8i32(<8 x i32> %a0) {
 define i32 @test_v16i32(<16 x i32> %a0) {
 ; SSE2-LABEL: test_v16i32:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm4, %xmm5
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm0[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm4, %xmm6
-; SSE2-NEXT:    pmuludq %xmm5, %xmm6
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm2, %xmm0
 ; SSE2-NEXT:    pmuludq %xmm1, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm5, %xmm2
+; SSE2-NEXT:    pmuludq %xmm1, %xmm2
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[2,2,0,0]
-; SSE2-NEXT:    pmuludq %xmm6, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    movd %xmm1, %eax
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[2,2,0,0]
+; SSE2-NEXT:    pmuludq %xmm2, %xmm0
+; SSE2-NEXT:    pmuludq %xmm1, %xmm0
+; SSE2-NEXT:    movd %xmm0, %eax
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_v16i32:
@@ -1012,39 +994,32 @@ define i32 @test_v16i32(<16 x i32> %a0) {
 define i32 @test_v32i32(<32 x i32> %a0) {
 ; SSE2-LABEL: test_v32i32:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm6[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm8, %xmm9
+; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm4[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm8, %xmm10
+; SSE2-NEXT:    pmuludq %xmm9, %xmm10
+; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm7[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm3[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm8, %xmm9
+; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm5[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm8, %xmm11
+; SSE2-NEXT:    pmuludq %xmm9, %xmm11
+; SSE2-NEXT:    pmuludq %xmm10, %xmm11
 ; SSE2-NEXT:    pmuludq %xmm6, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm0[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm4, %xmm0
 ; SSE2-NEXT:    pmuludq %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm7, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm1[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm5, %xmm1
 ; SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm8, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm9, %xmm3
-; SSE2-NEXT:    pmuludq %xmm0, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm7[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm5[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm10, %xmm2
-; SSE2-NEXT:    pmuludq %xmm0, %xmm2
-; SSE2-NEXT:    pmuludq %xmm3, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[2,2,0,0]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
+; SSE2-NEXT:    pmuludq %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[2,2,0,0]
+; SSE2-NEXT:    pmuludq %xmm11, %xmm1
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
 ; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    retq
diff --git a/llvm/test/CodeGen/X86/vector-reduce-mul.ll b/llvm/test/CodeGen/X86/vector-reduce-mul.ll
index 5f1455daab95..b474cc0b75fc 100644
--- a/llvm/test/CodeGen/X86/vector-reduce-mul.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-mul.ll
@@ -807,12 +807,8 @@ define i32 @test_v4i32(<4 x i32> %a0) {
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[3,3,1,1]
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm2, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
+; SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    retq
 ;
@@ -849,23 +845,16 @@ define i32 @test_v4i32(<4 x i32> %a0) {
 define i32 @test_v8i32(<8 x i32> %a0) {
 ; SSE2-LABEL: test_v8i32:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm2, %xmm3
 ; SSE2-NEXT:    pmuludq %xmm1, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,2,0,0]
-; SSE2-NEXT:    pmuludq %xmm1, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    movd %xmm1, %eax
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[2,2,0,0]
+; SSE2-NEXT:    pmuludq %xmm3, %xmm0
+; SSE2-NEXT:    pmuludq %xmm1, %xmm0
+; SSE2-NEXT:    movd %xmm0, %eax
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_v8i32:
@@ -920,29 +909,22 @@ define i32 @test_v8i32(<8 x i32> %a0) {
 define i32 @test_v16i32(<16 x i32> %a0) {
 ; SSE2-LABEL: test_v16i32:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm4, %xmm5
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm0[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm4, %xmm6
-; SSE2-NEXT:    pmuludq %xmm5, %xmm6
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm2, %xmm0
 ; SSE2-NEXT:    pmuludq %xmm1, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm5, %xmm2
+; SSE2-NEXT:    pmuludq %xmm1, %xmm2
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[2,2,0,0]
-; SSE2-NEXT:    pmuludq %xmm6, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    movd %xmm1, %eax
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[2,2,0,0]
+; SSE2-NEXT:    pmuludq %xmm2, %xmm0
+; SSE2-NEXT:    pmuludq %xmm1, %xmm0
+; SSE2-NEXT:    movd %xmm0, %eax
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_v16i32:
@@ -1005,39 +987,32 @@ define i32 @test_v16i32(<16 x i32> %a0) {
 define i32 @test_v32i32(<32 x i32> %a0) {
 ; SSE2-LABEL: test_v32i32:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm6[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm8, %xmm9
+; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm4[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm8, %xmm10
+; SSE2-NEXT:    pmuludq %xmm9, %xmm10
+; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm7[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm3[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm8, %xmm9
+; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm5[1,1,3,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pmuludq %xmm8, %xmm11
+; SSE2-NEXT:    pmuludq %xmm9, %xmm11
+; SSE2-NEXT:    pmuludq %xmm10, %xmm11
 ; SSE2-NEXT:    pmuludq %xmm6, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm0[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm4, %xmm0
 ; SSE2-NEXT:    pmuludq %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm7, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm1[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm5, %xmm1
 ; SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm8, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm9, %xmm3
-; SSE2-NEXT:    pmuludq %xmm0, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm7[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm5[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm10, %xmm2
-; SSE2-NEXT:    pmuludq %xmm0, %xmm2
-; SSE2-NEXT:    pmuludq %xmm3, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[2,2,0,0]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
+; SSE2-NEXT:    pmuludq %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[2,2,0,0]
+; SSE2-NEXT:    pmuludq %xmm11, %xmm1
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
 ; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    retq
@@ -1151,12 +1126,8 @@ define i16 @test_v4i16(<4 x i16> %a0) {
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[3,3,1,1]
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm2, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
+; SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; SSE2-NEXT:    pextrw $0, %xmm1, %eax
 ; SSE2-NEXT:    # kill: def $ax killed $ax killed $eax
 ; SSE2-NEXT:    retq
@@ -1603,12 +1574,8 @@ define i8 @test_v4i8(<4 x i8> %a0) {
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[3,3,1,1]
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; SSE2-NEXT:    pmuludq %xmm2, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
-; SSE2-NEXT:    pmuludq %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; SSE2-NEXT:    pmuludq %xmm0, %xmm1
+; SSE2-NEXT:    pmuludq %xmm3, %xmm1
 ; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq
diff --git a/llvm/test/CodeGen/X86/vector-reduce-umax.ll b/llvm/test/CodeGen/X86/vector-reduce-umax.ll
index 075d219fd771..6efd3724d710 100644
--- a/llvm/test/CodeGen/X86/vector-reduce-umax.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-umax.ll
@@ -730,23 +730,24 @@ define i32 @test_v2i32(<2 x i32> %a0) {
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [4294967295,0,4294967295,0]
 ; SSE2-NEXT:    pand %xmm0, %xmm1
 ; SSE2-NEXT:    pxor %xmm2, %xmm2
-; SSE2-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
 ; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pxor %xmm2, %xmm3
-; SSE2-NEXT:    pxor %xmm1, %xmm2
-; SSE2-NEXT:    movdqa %xmm2, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    punpckhdq {{.*#+}} xmm3 = xmm3[2],xmm2[2],xmm3[3],xmm2[3]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pxor %xmm2, %xmm4
+; SSE2-NEXT:    pxor %xmm2, %xmm1
+; SSE2-NEXT:    movdqa %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm4, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm1
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pandn %xmm0, %xmm3
-; SSE2-NEXT:    por %xmm1, %xmm3
-; SSE2-NEXT:    movd %xmm3, %eax
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn %xmm3, %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    movd %xmm2, %eax
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_v2i32:
@@ -1212,24 +1213,24 @@ define i16 @test_v2i16(<2 x i16> %a0) {
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,0,0,65535,0,0,0]
 ; SSE2-NEXT:    pand %xmm0, %xmm1
 ; SSE2-NEXT:    pxor %xmm2, %xmm2
-; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE2-NEXT:    movdqa %xmm0, %xmm3
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    por %xmm0, %xmm2
-; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    por %xmm2, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm2
+; SSE2-NEXT:    movdqa %xmm1, %xmm4
 ; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pandn %xmm0, %xmm3
-; SSE2-NEXT:    por %xmm1, %xmm3
-; SSE2-NEXT:    movd %xmm3, %eax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn %xmm3, %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    movd %xmm2, %eax
 ; SSE2-NEXT:    # kill: def $ax killed $ax killed $eax
 ; SSE2-NEXT:    retq
 ;
@@ -1309,12 +1310,11 @@ define i16 @test_v4i16(<4 x i16> %a0) {
 ; SSE2-NEXT:    pxor %xmm2, %xmm2
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    pandn %xmm0, %xmm2
-; SSE2-NEXT:    por %xmm1, %xmm2
-; SSE2-NEXT:    movd %xmm2, %eax
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pandn %xmm0, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    # kill: def $ax killed $ax killed $eax
 ; SSE2-NEXT:    retq
 ;
@@ -1765,25 +1765,25 @@ define i8 @test_v2i8(<2 x i8> %a0) {
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0]
 ; SSE2-NEXT:    pand %xmm0, %xmm1
 ; SSE2-NEXT:    pxor %xmm2, %xmm2
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE2-NEXT:    movdqa %xmm0, %xmm3
+; SSE2-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    por %xmm0, %xmm2
-; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    por %xmm2, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm2
+; SSE2-NEXT:    movdqa %xmm1, %xmm4
 ; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pandn %xmm0, %xmm3
-; SSE2-NEXT:    por %xmm1, %xmm3
-; SSE2-NEXT:    movd %xmm3, %eax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn %xmm3, %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    movd %xmm2, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq
 ;
@@ -1861,12 +1861,11 @@ define i8 @test_v4i8(<4 x i8> %a0) {
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    pandn %xmm0, %xmm2
-; SSE2-NEXT:    por %xmm1, %xmm2
-; SSE2-NEXT:    movd %xmm2, %eax
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pandn %xmm0, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq
 ;
diff --git a/llvm/test/CodeGen/X86/vector-reduce-umin.ll b/llvm/test/CodeGen/X86/vector-reduce-umin.ll
index b75523562d75..184d54528630 100644
--- a/llvm/test/CodeGen/X86/vector-reduce-umin.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-umin.ll
@@ -728,23 +728,23 @@ define i32 @test_v2i32(<2 x i32> %a0) {
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [4294967295,0,4294967295,0]
 ; SSE2-NEXT:    pand %xmm0, %xmm1
 ; SSE2-NEXT:    pxor %xmm2, %xmm2
-; SSE2-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; SSE2-NEXT:    movdqa %xmm0, %xmm3
+; SSE2-NEXT:    punpckhdq {{.*#+}} xmm3 = xmm3[2],xmm2[2],xmm3[3],xmm2[3]
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    pxor %xmm2, %xmm3
-; SSE2-NEXT:    pxor %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm2, %xmm1
+; SSE2-NEXT:    pxor %xmm3, %xmm2
 ; SSE2-NEXT:    movdqa %xmm2, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm4
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pandn %xmm0, %xmm3
-; SSE2-NEXT:    por %xmm1, %xmm3
-; SSE2-NEXT:    movd %xmm3, %eax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn %xmm3, %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    movd %xmm2, %eax
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_v2i32:
@@ -1210,24 +1210,24 @@ define i16 @test_v2i16(<2 x i16> %a0) {
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,0,0,65535,0,0,0]
 ; SSE2-NEXT:    pand %xmm0, %xmm1
 ; SSE2-NEXT:    pxor %xmm2, %xmm2
-; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE2-NEXT:    movdqa %xmm0, %xmm3
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    por %xmm2, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm2
 ; SSE2-NEXT:    movdqa %xmm2, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm4
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pandn %xmm0, %xmm3
-; SSE2-NEXT:    por %xmm1, %xmm3
-; SSE2-NEXT:    movd %xmm3, %eax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn %xmm3, %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    movd %xmm2, %eax
 ; SSE2-NEXT:    # kill: def $ax killed $ax killed $eax
 ; SSE2-NEXT:    retq
 ;
@@ -1309,9 +1309,9 @@ define i16 @test_v4i16(<4 x i16> %a0) {
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
 ; SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
-; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm3
 ; SSE2-NEXT:    pandn %xmm0, %xmm2
-; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    por %xmm3, %xmm2
 ; SSE2-NEXT:    movd %xmm2, %eax
 ; SSE2-NEXT:    # kill: def $ax killed $ax killed $eax
 ; SSE2-NEXT:    retq
@@ -1670,25 +1670,25 @@ define i8 @test_v2i8(<2 x i8> %a0) {
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0]
 ; SSE2-NEXT:    pand %xmm0, %xmm1
 ; SSE2-NEXT:    pxor %xmm2, %xmm2
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE2-NEXT:    movdqa %xmm0, %xmm3
+; SSE2-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [2147483648,2147483648]
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    por %xmm2, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm2
 ; SSE2-NEXT:    movdqa %xmm2, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm4
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pandn %xmm0, %xmm3
-; SSE2-NEXT:    por %xmm1, %xmm3
-; SSE2-NEXT:    movd %xmm3, %eax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn %xmm3, %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    movd %xmm2, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq
 ;
@@ -1768,9 +1768,9 @@ define i8 @test_v4i8(<4 x i8> %a0) {
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
 ; SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
-; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm3
 ; SSE2-NEXT:    pandn %xmm0, %xmm2
-; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    por %xmm3, %xmm2
 ; SSE2-NEXT:    movd %xmm2, %eax
 ; SSE2-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE2-NEXT:    retq