diff --git a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
index b55d370..dcd8478 100644
--- a/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp
@@ -1381,11 +1381,18 @@
     bool IsVecInReg = Op.getOpcode() == ISD::ZERO_EXTEND_VECTOR_INREG;
 
     // If none of the top bits are demanded, convert this into an any_extend.
-    if (DemandedBits.getActiveBits() <= InBits)
-      return TLO.CombineTo(
-          Op, TLO.DAG.getNode(IsVecInReg ? ISD::ANY_EXTEND_VECTOR_INREG
-                                         : ISD::ANY_EXTEND,
-                              dl, VT, Src));
+    // TODO: Add ZERO_EXTEND_VECTOR_INREG - ANY_EXTEND_VECTOR_INREG fold.
+    if (DemandedBits.getActiveBits() <= InBits) {
+      // If we only need the non-extended bits of the bottom element
+      // then we can just bitcast to the result.
+      if (IsVecInReg && DemandedElts == 1 &&
+          VT.getSizeInBits() == SrcVT.getSizeInBits() &&
+          TLO.DAG.getDataLayout().isLittleEndian())
+        return TLO.CombineTo(Op, TLO.DAG.getBitcast(VT, Src));
+
+      if (!IsVecInReg)
+        return TLO.CombineTo(Op, TLO.DAG.getNode(ISD::ANY_EXTEND, dl, VT, Src));
+    }
 
     APInt InDemandedBits = DemandedBits.trunc(InBits);
     APInt InDemandedElts = DemandedElts.zextOrSelf(InElts);
@@ -1406,11 +1413,9 @@
     bool IsVecInReg = Op.getOpcode() == ISD::SIGN_EXTEND_VECTOR_INREG;
 
     // If none of the top bits are demanded, convert this into an any_extend.
-    if (DemandedBits.getActiveBits() <= InBits)
-      return TLO.CombineTo(
-          Op, TLO.DAG.getNode(IsVecInReg ? ISD::ANY_EXTEND_VECTOR_INREG
-                                         : ISD::ANY_EXTEND,
-                              dl, VT, Src));
+    // TODO: Add SIGN_EXTEND_VECTOR_INREG - ANY_EXTEND_VECTOR_INREG fold.
+    if (DemandedBits.getActiveBits() <= InBits && !IsVecInReg)
+      return TLO.CombineTo(Op, TLO.DAG.getNode(ISD::ANY_EXTEND, dl, VT, Src));
 
     APInt InDemandedBits = DemandedBits.trunc(InBits);
     APInt InDemandedElts = DemandedElts.zextOrSelf(InElts);
@@ -1429,28 +1434,17 @@
     Known = Known.sext(BitWidth);
 
     // If the sign bit is known zero, convert this to a zero extend.
-    if (Known.isNonNegative())
-      return TLO.CombineTo(
-          Op, TLO.DAG.getNode(IsVecInReg ? ISD::ZERO_EXTEND_VECTOR_INREG
-                                         : ISD::ZERO_EXTEND,
-                              dl, VT, Src));
+    // TODO: Add SIGN_EXTEND_VECTOR_INREG - ZERO_EXTEND_VECTOR_INREG fold.
+    if (Known.isNonNegative() && !IsVecInReg)
+      return TLO.CombineTo(Op, TLO.DAG.getNode(ISD::ZERO_EXTEND, dl, VT, Src));
     break;
   }
-  case ISD::ANY_EXTEND:
-  case ISD::ANY_EXTEND_VECTOR_INREG: {
+  case ISD::ANY_EXTEND: {
+    // TODO: Add ISD::ANY_EXTEND_VECTOR_INREG support.
     SDValue Src = Op.getOperand(0);
     EVT SrcVT = Src.getValueType();
     unsigned InBits = SrcVT.getScalarSizeInBits();
     unsigned InElts = SrcVT.isVector() ? SrcVT.getVectorNumElements() : 1;
-    bool IsVecInReg = Op.getOpcode() == ISD::ANY_EXTEND_VECTOR_INREG;
-
-    // If we only need the bottom element then we can just bitcast.
-    // TODO: Handle ANY_EXTEND?
-    if (IsVecInReg && DemandedElts == 1 &&
-        VT.getSizeInBits() == SrcVT.getSizeInBits() &&
-        TLO.DAG.getDataLayout().isLittleEndian())
-      return TLO.CombineTo(Op, TLO.DAG.getBitcast(VT, Src));
-
     APInt InDemandedBits = DemandedBits.trunc(InBits);
     APInt InDemandedElts = DemandedElts.zextOrSelf(InElts);
     if (SimplifyDemandedBits(Src, InDemandedBits, InDemandedElts, Known, TLO,
diff --git a/llvm/test/CodeGen/X86/combine-pmuldq.ll b/llvm/test/CodeGen/X86/combine-pmuldq.ll
index 5b5ae38..39b644a 100644
--- a/llvm/test/CodeGen/X86/combine-pmuldq.ll
+++ b/llvm/test/CodeGen/X86/combine-pmuldq.ll
@@ -91,15 +91,18 @@
 define <8 x i64> @combine_zext_pmuludq_256(<8 x i32> %a) {
 ; SSE-LABEL: combine_zext_pmuludq_256:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,2,3,3]
+; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
+; SSE-NEXT:    pmovzxdq {{.*#+}} xmm3 = xmm2[0],zero,xmm2[1],zero
+; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
+; SSE-NEXT:    pmovzxdq {{.*#+}} xmm4 = xmm2[0],zero,xmm2[1],zero
 ; SSE-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm1[0],zero,xmm1[1],zero
-; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,2,3,3]
 ; SSE-NEXT:    pmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [715827883,715827883]
-; SSE-NEXT:    pmuludq %xmm4, %xmm0
-; SSE-NEXT:    pmuludq %xmm4, %xmm1
-; SSE-NEXT:    pmuludq %xmm4, %xmm2
-; SSE-NEXT:    pmuludq %xmm4, %xmm3
+; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [715827883,715827883]
+; SSE-NEXT:    pmuludq %xmm1, %xmm0
+; SSE-NEXT:    pmuludq %xmm1, %xmm2
+; SSE-NEXT:    pmuludq %xmm1, %xmm4
+; SSE-NEXT:    pmuludq %xmm1, %xmm3
+; SSE-NEXT:    movdqa %xmm4, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX2-LABEL: combine_zext_pmuludq_256:
diff --git a/llvm/test/CodeGen/X86/pmul.ll b/llvm/test/CodeGen/X86/pmul.ll
index 8e6b75f..1960e79 100644
--- a/llvm/test/CodeGen/X86/pmul.ll
+++ b/llvm/test/CodeGen/X86/pmul.ll
@@ -1016,24 +1016,30 @@
 define <4 x i32> @mul_v4i64_zero_upper(<4 x i32> %val1, <4 x i32> %val2) {
 ; SSE2-LABEL: mul_v4i64_zero_upper:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[0,1,1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,1,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,1,1,3]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm3, %xmm1
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,3],xmm1[1,3]
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; SSE2-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm3[2],xmm0[3],xmm3[3]
+; SSE2-NEXT:    movdqa %xmm1, %xmm4
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
+; SSE2-NEXT:    pmuludq %xmm4, %xmm2
+; SSE2-NEXT:    punpckhdq {{.*#+}} xmm1 = xmm1[2],xmm3[2],xmm1[3],xmm3[3]
+; SSE2-NEXT:    pmuludq %xmm0, %xmm1
+; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,3],xmm1[1,3]
+; SSE2-NEXT:    movaps %xmm2, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: mul_v4i64_zero_upper:
 ; SSE41:       # %bb.0: # %entry
-; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,2,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm3 = xmm0[0],zero,xmm0[1],zero
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm4 = xmm0[0],zero,xmm0[1],zero
+; SSE41-NEXT:    pmuludq %xmm2, %xmm4
 ; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm0 = xmm1[0],zero,xmm1[1],zero
-; SSE41-NEXT:    pmuludq %xmm2, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,2,3,3]
-; SSE41-NEXT:    pmuludq %xmm3, %xmm1
-; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,3],xmm1[1,3]
+; SSE41-NEXT:    pmuludq %xmm3, %xmm0
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,3],xmm4[1,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: mul_v4i64_zero_upper:
@@ -1165,37 +1171,48 @@
 define <8 x i32> @mul_v8i64_zero_upper(<8 x i32> %val1, <8 x i32> %val2) {
 ; SSE2-LABEL: mul_v8i64_zero_upper:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[0,1,1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[2,1,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm1[0,1,1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm1[2,1,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,1,1,3]
-; SSE2-NEXT:    pmuludq %xmm4, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[2,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm5, %xmm1
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,3],xmm1[1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[0,1,1,3]
-; SSE2-NEXT:    pmuludq %xmm6, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[2,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm7, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,3],xmm2[1,3]
+; SSE2-NEXT:    pxor %xmm6, %xmm6
+; SSE2-NEXT:    movdqa %xmm0, %xmm4
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm6[0],xmm4[1],xmm6[1]
+; SSE2-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm6[2],xmm0[3],xmm6[3]
+; SSE2-NEXT:    movdqa %xmm1, %xmm5
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1]
+; SSE2-NEXT:    punpckhdq {{.*#+}} xmm1 = xmm1[2],xmm6[2],xmm1[3],xmm6[3]
+; SSE2-NEXT:    movdqa %xmm2, %xmm7
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm6[0],xmm7[1],xmm6[1]
+; SSE2-NEXT:    pmuludq %xmm7, %xmm4
+; SSE2-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm6[2],xmm2[3],xmm6[3]
+; SSE2-NEXT:    pmuludq %xmm0, %xmm2
+; SSE2-NEXT:    shufps {{.*#+}} xmm4 = xmm4[1,3],xmm2[1,3]
+; SSE2-NEXT:    movdqa %xmm3, %xmm0
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1]
+; SSE2-NEXT:    pmuludq %xmm0, %xmm5
+; SSE2-NEXT:    punpckhdq {{.*#+}} xmm3 = xmm3[2],xmm6[2],xmm3[3],xmm6[3]
+; SSE2-NEXT:    pmuludq %xmm1, %xmm3
+; SSE2-NEXT:    shufps {{.*#+}} xmm5 = xmm5[1,3],xmm3[1,3]
+; SSE2-NEXT:    movaps %xmm4, %xmm0
+; SSE2-NEXT:    movaps %xmm5, %xmm1
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: mul_v8i64_zero_upper:
 ; SSE41:       # %bb.0: # %entry
-; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm4 = xmm0[0],zero,xmm0[1],zero
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[2,2,3,3]
-; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm6 = xmm1[0],zero,xmm1[1],zero
-; SSE41-NEXT:    pshufd {{.*#+}} xmm7 = xmm1[2,2,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[2,3,0,1]
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm5 = xmm0[0],zero,xmm0[1],zero
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm6 = xmm0[0],zero,xmm0[1],zero
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm7 = xmm1[0],zero,xmm1[1],zero
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero
+; SSE41-NEXT:    pmuludq %xmm4, %xmm1
 ; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm0 = xmm2[0],zero,xmm2[1],zero
-; SSE41-NEXT:    pmuludq %xmm4, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[2,2,3,3]
-; SSE41-NEXT:    pmuludq %xmm5, %xmm1
+; SSE41-NEXT:    pmuludq %xmm5, %xmm0
 ; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,3],xmm1[1,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[2,3,0,1]
+; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm1[0],zero,xmm1[1],zero
+; SSE41-NEXT:    pmuludq %xmm6, %xmm2
 ; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm1 = xmm3[0],zero,xmm3[1],zero
-; SSE41-NEXT:    pmuludq %xmm6, %xmm1
-; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[2,2,3,3]
-; SSE41-NEXT:    pmuludq %xmm7, %xmm2
+; SSE41-NEXT:    pmuludq %xmm7, %xmm1
 ; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,3],xmm2[1,3]
 ; SSE41-NEXT:    retq
 ;
@@ -1309,13 +1326,15 @@
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,2,3]
 ; SSE41-NEXT:    pmovsxwq %xmm3, %xmm6
 ; SSE41-NEXT:    pmovsxwq %xmm0, %xmm7
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[2,2,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
+; SSE41-NEXT:    pmovsxdq %xmm0, %xmm3
 ; SSE41-NEXT:    pmuldq %xmm4, %xmm3
-; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero
+; SSE41-NEXT:    pmovsxdq %xmm2, %xmm2
 ; SSE41-NEXT:    pmuldq %xmm5, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[2,2,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,3,0,1]
+; SSE41-NEXT:    pmovsxdq %xmm0, %xmm4
 ; SSE41-NEXT:    pmuldq %xmm6, %xmm4
-; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm0 = xmm1[0],zero,xmm1[1],zero
+; SSE41-NEXT:    pmovsxdq %xmm1, %xmm0
 ; SSE41-NEXT:    pmuldq %xmm7, %xmm0
 ; SSE41-NEXT:    movdqa %xmm4, %xmm1
 ; SSE41-NEXT:    retq
diff --git a/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll b/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
index 135d232..573f9c8 100644
--- a/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
@@ -2320,47 +2320,45 @@
 ;
 ; AVX1-LABEL: test_v64i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm3[8],xmm0[8],xmm3[9],xmm0[9],xmm3[10],xmm0[10],xmm3[11],xmm0[11],xmm3[12],xmm0[12],xmm3[13],xmm0[13],xmm3[14],xmm0[14],xmm3[15],xmm0[15]
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm2[8],xmm0[8],xmm2[9],xmm0[9],xmm2[10],xmm0[10],xmm2[11],xmm0[11],xmm2[12],xmm0[12],xmm2[13],xmm0[13],xmm2[14],xmm0[14],xmm2[15],xmm0[15]
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm4
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm4[8],xmm0[8],xmm4[9],xmm0[9],xmm4[10],xmm0[10],xmm4[11],xmm0[11],xmm4[12],xmm0[12],xmm4[13],xmm0[13],xmm4[14],xmm0[14],xmm4[15],xmm0[15]
-; AVX1-NEXT:    vpmullw %xmm2, %xmm5, %xmm5
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm2, %xmm5, %xmm6
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
-; AVX1-NEXT:    vpmullw %xmm3, %xmm4, %xmm3
-; AVX1-NEXT:    vpand %xmm2, %xmm3, %xmm3
-; AVX1-NEXT:    vpackuswb %xmm6, %xmm3, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm4 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm6 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm5, %xmm3
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm5, %xmm3
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; 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:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm5, %xmm4, %xmm4
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
+; AVX1-NEXT:    vpmullw %xmm2, %xmm4, %xmm2
+; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; 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:    vpmullw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
 ; 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:    vpmullw %xmm0, %xmm4, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm6, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpmullw %xmm0, %xmm5, %xmm0
+; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpmovzxbw {{.*#+}} 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
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpmovzxbw {{.*#+}} 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
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
-; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
@@ -2751,67 +2749,65 @@
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm2[8],xmm0[8],xmm2[9],xmm0[9],xmm2[10],xmm0[10],xmm2[11],xmm0[11],xmm2[12],xmm0[12],xmm2[13],xmm0[13],xmm2[14],xmm0[14],xmm2[15],xmm0[15]
 ; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm9
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm8 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm10
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm7 = xmm10[8],xmm0[8],xmm10[9],xmm0[9],xmm10[10],xmm0[10],xmm10[11],xmm0[11],xmm10[12],xmm0[12],xmm10[13],xmm0[13],xmm10[14],xmm0[14],xmm10[15],xmm0[15]
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm11
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm11[8],xmm0[8],xmm11[9],xmm0[9],xmm11[10],xmm0[10],xmm11[11],xmm0[11],xmm11[12],xmm0[12],xmm11[13],xmm0[13],xmm11[14],xmm0[14],xmm11[15],xmm0[15]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm5, %xmm5
+; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm11
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm7 = xmm11[8],xmm0[8],xmm11[9],xmm0[9],xmm11[10],xmm0[10],xmm11[11],xmm0[11],xmm11[12],xmm0[12],xmm11[13],xmm0[13],xmm11[14],xmm0[14],xmm11[15],xmm0[15]
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm4[8],xmm0[8],xmm4[9],xmm0[9],xmm4[10],xmm0[10],xmm4[11],xmm0[11],xmm4[12],xmm0[12],xmm4[13],xmm0[13],xmm4[14],xmm0[14],xmm4[15],xmm0[15]
+; AVX1-NEXT:    vpmullw %xmm7, %xmm5, %xmm10
 ; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm7
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm6 = xmm7[8],xmm0[8],xmm7[9],xmm0[9],xmm7[10],xmm0[10],xmm7[11],xmm0[11],xmm7[12],xmm0[12],xmm7[13],xmm0[13],xmm7[14],xmm0[14],xmm7[15],xmm0[15]
-; AVX1-NEXT:    vpmullw %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm6
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm4 = xmm6[8],xmm0[8],xmm6[9],xmm0[9],xmm6[10],xmm0[10],xmm6[11],xmm0[11],xmm6[12],xmm0[12],xmm6[13],xmm0[13],xmm6[14],xmm0[14],xmm6[15],xmm0[15]
-; AVX1-NEXT:    vpmullw %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm7[8],xmm0[8],xmm7[9],xmm0[9],xmm7[10],xmm0[10],xmm7[11],xmm0[11],xmm7[12],xmm0[12],xmm7[13],xmm0[13],xmm7[14],xmm0[14],xmm7[15],xmm0[15]
+; AVX1-NEXT:    vpmullw %xmm10, %xmm5, %xmm10
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm5
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm6 = xmm5[8],xmm0[8],xmm5[9],xmm0[9],xmm5[10],xmm0[10],xmm5[11],xmm0[11],xmm5[12],xmm0[12],xmm5[13],xmm0[13],xmm5[14],xmm0[14],xmm5[15],xmm0[15]
+; AVX1-NEXT:    vpmullw %xmm10, %xmm6, %xmm6
+; AVX1-NEXT:    vpmullw %xmm6, %xmm9, %xmm9
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm7 = xmm7[0],zero,xmm7[1],zero,xmm7[2],zero,xmm7[3],zero,xmm7[4],zero,xmm7[5],zero,xmm7[6],zero,xmm7[7],zero
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero,xmm5[4],zero,xmm5[5],zero,xmm5[6],zero,xmm5[7],zero
+; AVX1-NEXT:    vpmullw %xmm7, %xmm5, %xmm5
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX1-NEXT:    vpshufb %xmm7, %xmm5, %xmm5
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm6 = xmm11[0],zero,xmm11[1],zero,xmm11[2],zero,xmm11[3],zero,xmm11[4],zero,xmm11[5],zero,xmm11[6],zero,xmm11[7],zero
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
+; AVX1-NEXT:    vpmullw %xmm6, %xmm4, %xmm4
+; AVX1-NEXT:    vpshufb %xmm7, %xmm4, %xmm4
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; 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:    vpmullw %xmm2, %xmm0, %xmm0
-; 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, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
+; AVX1-NEXT:    vpshufb %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
-; AVX1-NEXT:    vpmullw %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm7[0],zero,xmm7[1],zero,xmm7[2],zero,xmm7[3],zero,xmm7[4],zero,xmm7[5],zero,xmm7[6],zero,xmm7[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = xmm6[0],zero,xmm6[1],zero,xmm6[2],zero,xmm6[3],zero,xmm6[4],zero,xmm6[5],zero,xmm6[6],zero,xmm6[7],zero
-; AVX1-NEXT:    vpmullw %xmm3, %xmm5, %xmm3
-; AVX1-NEXT:    vpshufb %xmm2, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = xmm10[0],zero,xmm10[1],zero,xmm10[2],zero,xmm10[3],zero,xmm10[4],zero,xmm10[5],zero,xmm10[6],zero,xmm10[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm6 = xmm11[0],zero,xmm11[1],zero,xmm11[2],zero,xmm11[3],zero,xmm11[4],zero,xmm11[5],zero,xmm11[6],zero,xmm11[7],zero
-; AVX1-NEXT:    vpmullw %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vpshufb %xmm2, %xmm5, %xmm5
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm6, %xmm4, %xmm7
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero,xmm5[4],zero,xmm5[5],zero,xmm5[6],zero,xmm5[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
-; AVX1-NEXT:    vpmullw %xmm5, %xmm3, %xmm3
-; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
-; AVX1-NEXT:    vpackuswb %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; 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:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm4, %xmm9, %xmm1
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
+; AVX1-NEXT:    vpshufb %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero,xmm5[4],zero,xmm5[5],zero,xmm5[6],zero,xmm5[7],zero
+; AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm1
+; AVX1-NEXT:    vpshufb %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; 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:    vpmullw %xmm3, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpshufb %xmm7, %xmm0, %xmm0
 ; 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:    vpmullw %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpmullw %xmm0, %xmm9, %xmm0
 ; AVX1-NEXT:    vpmullw %xmm0, %xmm8, %xmm0
-; AVX1-NEXT:    vpand %xmm6, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} 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
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmullw %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpand %xmm6, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} 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
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
-; AVX1-NEXT:    vpmullw %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpand %xmm6, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
diff --git a/llvm/test/CodeGen/X86/vector-reduce-mul.ll b/llvm/test/CodeGen/X86/vector-reduce-mul.ll
index 92873ad..2187c9d 100644
--- a/llvm/test/CodeGen/X86/vector-reduce-mul.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-mul.ll
@@ -1766,18 +1766,22 @@
 ; 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:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[8,8,10,10,12,12,14,14,0,0,2,2,4,4,6,6]
-; AVX1-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX1-NEXT:    # kill: def $al killed $al killed $eax
@@ -1985,25 +1989,28 @@
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX1-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; 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:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[8,8,10,10,12,12,14,14,0,0,2,2,4,4,6,6]
-; AVX1-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX1-NEXT:    # kill: def $al killed $al killed $eax
@@ -2269,31 +2276,33 @@
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX1-NEXT:    vpmullw %xmm3, %xmm5, %xmm3
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm5, %xmm3
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; 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:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
+; AVX1-NEXT:    vpmullw %xmm2, %xmm4, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
+; AVX1-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpmullw %xmm3, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
-; AVX1-NEXT:    vpmullw %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm5, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[8,8,10,10,12,12,14,14,0,0,2,2,4,4,6,6]
-; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpmovzxbw {{.*#+}} 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
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpmullw %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} 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
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
-; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX1-NEXT:    vpmullw %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX1-NEXT:    # kill: def $al killed $al killed $eax
@@ -2659,8 +2668,8 @@
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm4
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm2[8],xmm0[8],xmm2[9],xmm0[9],xmm2[10],xmm0[10],xmm2[11],xmm0[11],xmm2[12],xmm0[12],xmm2[13],xmm0[13],xmm2[14],xmm0[14],xmm2[15],xmm0[15]
-; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm9
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm8 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm8
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm9 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm11
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm7 = xmm11[8],xmm0[8],xmm11[9],xmm0[9],xmm11[10],xmm0[10],xmm11[11],xmm0[11],xmm11[12],xmm0[12],xmm11[13],xmm0[13],xmm11[14],xmm0[14],xmm11[15],xmm0[15]
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
@@ -2672,6 +2681,7 @@
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm5
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm6 = xmm5[8],xmm0[8],xmm5[9],xmm0[9],xmm5[10],xmm0[10],xmm5[11],xmm0[11],xmm5[12],xmm0[12],xmm5[13],xmm0[13],xmm5[14],xmm0[14],xmm5[15],xmm0[15]
 ; AVX1-NEXT:    vpmullw %xmm10, %xmm6, %xmm6
+; AVX1-NEXT:    vpmullw %xmm6, %xmm8, %xmm6
 ; AVX1-NEXT:    vpmullw %xmm6, %xmm9, %xmm6
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm7 = xmm7[0],zero,xmm7[1],zero,xmm7[2],zero,xmm7[3],zero,xmm7[4],zero,xmm7[5],zero,xmm7[6],zero,xmm7[7],zero
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero,xmm5[4],zero,xmm5[5],zero,xmm5[6],zero,xmm5[7],zero
@@ -2688,28 +2698,29 @@
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
 ; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm1
-; AVX1-NEXT:    vpand %xmm2, %xmm5, %xmm3
-; AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm6, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm8, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[8,8,10,10,12,12,14,14,0,0,2,2,4,4,6,6]
+; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm3
+; AVX1-NEXT:    vpand %xmm2, %xmm5, %xmm4
+; AVX1-NEXT:    vpmullw %xmm3, %xmm4, %xmm3
+; AVX1-NEXT:    vpand %xmm2, %xmm3, %xmm3
+; AVX1-NEXT:    vpmullw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpmullw %xmm6, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX1-NEXT:    # kill: def $al killed $al killed $eax
diff --git a/llvm/test/CodeGen/X86/xop-ifma.ll b/llvm/test/CodeGen/X86/xop-ifma.ll
index 4712910..b8ec25e 100644
--- a/llvm/test/CodeGen/X86/xop-ifma.ll
+++ b/llvm/test/CodeGen/X86/xop-ifma.ll
@@ -67,10 +67,12 @@
 define <4 x i64> @test_mulx_v4i32_add_v4i64(<4 x i32> %a0, <4 x i32> %a1, <4 x i64> %a2) {
 ; XOP-AVX1-LABEL: test_mulx_v4i32_add_v4i64:
 ; XOP-AVX1:       # %bb.0:
-; XOP-AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm3 = xmm0[0],zero,xmm0[1],zero
-; XOP-AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,3,3]
-; XOP-AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm4 = xmm1[0],zero,xmm1[1],zero
-; XOP-AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,1,3,3]
+; XOP-AVX1-NEXT:    vpmovsxdq %xmm0, %xmm3
+; XOP-AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; XOP-AVX1-NEXT:    vpmovsxdq %xmm0, %xmm0
+; XOP-AVX1-NEXT:    vpmovsxdq %xmm1, %xmm4
+; XOP-AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
+; XOP-AVX1-NEXT:    vpmovsxdq %xmm1, %xmm1
 ; XOP-AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm5
 ; XOP-AVX1-NEXT:    vpmacsdql %xmm5, %xmm1, %xmm0, %xmm0
 ; XOP-AVX1-NEXT:    vpmacsdql %xmm2, %xmm4, %xmm3, %xmm1