[DAGCombiner] fold extract_subvector of extract_subvector

[spatel]

Summary

This is the sibling fold for insert-of-insert that was added with D56604.

Now that we have x86 shuffle narrowing (D57156), this change shows improvements for lots of AVX512 reduction code (not sure that we would ever expect extract-of-extract otherwise).

There's a small regression in some of the partial-permute tests (extracting followed by splat). I'll try to reduce that and file a bug.

[spatel]

test/CodeGen/X86/avx512-shuffles/partial_permute.ll

3806–3815 ↗

(On Diff #183882)

See:
https://bugs.llvm.org/show_bug.cgi?id=40500

[RKSimon]

LGTM

llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

SDValue DAGCombiner::visitEXTRACT_SUBVECTOR(SDNode* N) {
   // Extract from UNDEF is UNDEF.
   if (V.isUndef())
     return DAG.getUNDEF(NVT);
 
   if (TLI.isOperationLegalOrCustomOrPromote(ISD::LOAD, NVT))
     if (SDValue NarrowLoad = narrowExtractedVectorLoad(N, DAG))
       return NarrowLoad;
 
+  // Combine an extract of an extract into a single extract_subvector.
+  // ext (ext X, C), 0 --> ext X, C
+  if (isNullConstant(N->getOperand(1)) &&
+      V.getOpcode() == ISD::EXTRACT_SUBVECTOR && V.hasOneUse() &&
+      isa<ConstantSDNode>(V.getOperand(1))) {
+    if (TLI.isExtractSubvectorCheap(NVT, V.getOperand(0).getValueType(),
+                                    V.getConstantOperandVal(1)) &&
+        TLI.isOperationLegalOrCustom(ISD::EXTRACT_SUBVECTOR, NVT)) {
+      return DAG.getNode(ISD::EXTRACT_SUBVECTOR, SDLoc(N), NVT, V.getOperand(0),
+                         V.getOperand(1));
+    }
+  }
+
   // Combine:
   //    (extract_subvec (concat V1, V2, ...), i)
   // Into:
   //    Vi if possible
   // Only operand 0 is checked as 'concat' assumes all inputs of the same
   // type.
   if (V.getOpcode() == ISD::CONCAT_VECTORS &&
       isa<ConstantSDNode>(N->getOperand(1)) &&

llvm/trunk/test/CodeGen/X86/avx512-hadd-hsub.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ;RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl | FileCheck %s --check-prefix=KNL
 ;RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=skx | FileCheck %s --check-prefix=SKX
 
 define i32 @hadd_16(<16 x i32> %x225) {
 ; KNL-LABEL: hadd_16:
 ; KNL:       # %bb.0:
 ; KNL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; KNL-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; KNL-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; KNL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; KNL-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; KNL-NEXT:    vmovd %xmm0, %eax
 ; KNL-NEXT:    retq
 ;
 ; SKX-LABEL: hadd_16:
 ; SKX:       # %bb.0:
 ; SKX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; SKX-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; SKX-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; SKX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; SKX-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; SKX-NEXT:    vmovd %xmm0, %eax
 ; SKX-NEXT:    vzeroupper
 ; SKX-NEXT:    retq
   %x226 = shufflevector <16 x i32> %x225, <16 x i32> undef, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   %x227 = add <16 x i32> %x225, %x226
   %x228 = shufflevector <16 x i32> %x227, <16 x i32> undef, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   %x229 = add <16 x i32> %x227, %x228
   %x230 = extractelement <16 x i32> %x229, i32 0
   ret i32 %x230
 }
 
 define i32 @hsub_16(<16 x i32> %x225) {
 ; KNL-LABEL: hsub_16:
 ; KNL:       # %bb.0:
 ; KNL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; KNL-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; KNL-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; KNL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; KNL-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
 ; KNL-NEXT:    vmovd %xmm0, %eax
 ; KNL-NEXT:    retq
 ;
 ; SKX-LABEL: hsub_16:
 ; SKX:       # %bb.0:
 ; SKX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; SKX-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; SKX-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; SKX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; SKX-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
 ; SKX-NEXT:    vmovd %xmm0, %eax
 ; SKX-NEXT:    vzeroupper
 ; SKX-NEXT:    retq
   %x226 = shufflevector <16 x i32> %x225, <16 x i32> undef, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   %x227 = add <16 x i32> %x225, %x226
   %x228 = shufflevector <16 x i32> %x227, <16 x i32> undef, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   %x229 = sub <16 x i32> %x227, %x228
   %x230 = extractelement <16 x i32> %x229, i32 0
   ret i32 %x230
 }
 
 define float @fhadd_16(<16 x float> %x225) {
 ; KNL-LABEL: fhadd_16:
 ; KNL:       # %bb.0:
 ; KNL-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; KNL-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; KNL-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; KNL-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; KNL-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; KNL-NEXT:    retq
 ;
 ; SKX-LABEL: fhadd_16:
 ; SKX:       # %bb.0:
 ; SKX-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; SKX-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; SKX-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; SKX-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; SKX-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; SKX-NEXT:    vzeroupper
 ; SKX-NEXT:    retq
   %x226 = shufflevector <16 x float> %x225, <16 x float> undef, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   %x227 = fadd <16 x float> %x225, %x226
   %x228 = shufflevector <16 x float> %x227, <16 x float> undef, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   %x229 = fadd <16 x float> %x227, %x228
   %x230 = extractelement <16 x float> %x229, i32 0
   ret float %x230
 }
 
 define float @fhsub_16(<16 x float> %x225) {
 ; KNL-LABEL: fhsub_16:
 ; KNL:       # %bb.0:
 ; KNL-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; KNL-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; KNL-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; KNL-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; KNL-NEXT:    vsubps %xmm1, %xmm0, %xmm0
 ; KNL-NEXT:    retq
 ;
 ; SKX-LABEL: fhsub_16:
 ; SKX:       # %bb.0:
 ; SKX-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; SKX-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; SKX-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; SKX-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; SKX-NEXT:    vsubps %xmm1, %xmm0, %xmm0
 ; SKX-NEXT:    vzeroupper
 ; SKX-NEXT:    retq
   %x226 = shufflevector <16 x float> %x225, <16 x float> undef, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   %x227 = fadd <16 x float> %x225, %x226
   %x228 = shufflevector <16 x float> %x227, <16 x float> undef, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   %x229 = fsub <16 x float> %x227, %x228

llvm/trunk/test/CodeGen/X86/avx512-shuffles/partial_permute.ll

; CHECK-NEXT:    retq
   %shuf = shufflevector <8 x i64> %vec, <8 x i64> undef, <4 x i32> <i32 7, i32 6, i32 5, i32 3>
   %cmp = icmp eq <4 x i64> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x i64> %shuf, <4 x i64> zeroinitializer
   ret <4 x i64> %res
 }
 define <4 x i64> @test_masked_8xi64_to_4xi64_perm_mask7(<8 x i64> %vec, <4 x i64> %vec2, <4 x i64> %mask) {
 ; CHECK-LABEL: test_masked_8xi64_to_4xi64_perm_mask7:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextracti64x4 $1, %zmm0, %ymm3
-; CHECK-NEXT:    vmovdqa {{.*#+}} ymm4 = [2,0,3,4]
+; CHECK-NEXT:    vextracti32x4 $2, %zmm0, %xmm3
+; CHECK-NEXT:    vpbroadcastq %xmm3, %ymm3
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm4 = [2,0,3,7]
 ; CHECK-NEXT:    vpermi2q %ymm3, %ymm0, %ymm4
 ; CHECK-NEXT:    vptestnmq %ymm2, %ymm2, %k1
 ; CHECK-NEXT:    vpblendmq %ymm4, %ymm1, %ymm0 {%k1}
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <8 x i64> %vec, <8 x i64> undef, <4 x i32> <i32 2, i32 0, i32 3, i32 4>
   %cmp = icmp eq <4 x i64> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x i64> %shuf, <4 x i64> %vec2
   ret <4 x i64> %res
 }
 
 define <4 x i64> @test_masked_z_8xi64_to_4xi64_perm_mask7(<8 x i64> %vec, <4 x i64> %mask) {
 ; CHECK-LABEL: test_masked_z_8xi64_to_4xi64_perm_mask7:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextracti64x4 $1, %zmm0, %ymm3
-; CHECK-NEXT:    vmovdqa {{.*#+}} ymm2 = [2,0,3,4]
+; CHECK-NEXT:    vextracti32x4 $2, %zmm0, %xmm2
+; CHECK-NEXT:    vpbroadcastq %xmm2, %ymm3
+; CHECK-NEXT:    vmovdqa {{.*#+}} ymm2 = [2,0,3,7]
 ; CHECK-NEXT:    vptestnmq %ymm1, %ymm1, %k1
 ; CHECK-NEXT:    vpermi2q %ymm3, %ymm0, %ymm2 {%k1} {z}
 ; CHECK-NEXT:    vmovdqa %ymm2, %ymm0
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <8 x i64> %vec, <8 x i64> undef, <4 x i32> <i32 2, i32 0, i32 3, i32 4>
   %cmp = icmp eq <4 x i64> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x i64> %shuf, <4 x i64> zeroinitializer
   ret <4 x i64> %res
; CHECK-NEXT:    retq
   %shuf = shufflevector <16 x float> %vec, <16 x float> undef, <4 x i32> <i32 4, i32 8, i32 9, i32 10>
   %cmp = fcmp oeq <4 x float> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x float> %shuf, <4 x float> zeroinitializer
   ret <4 x float> %res
 }
 define <4 x float> @test_masked_16xfloat_to_4xfloat_perm_mask1(<16 x float> %vec, <4 x float> %vec2, <4 x float> %mask) {
 ; CHECK-LABEL: test_masked_16xfloat_to_4xfloat_perm_mask1:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm3
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm3
 ; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
 ; CHECK-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,2]
 ; CHECK-NEXT:    vblendps {{.*#+}} xmm0 = xmm3[0],xmm0[1],xmm3[2],xmm0[3]
 ; CHECK-NEXT:    vxorps %xmm3, %xmm3, %xmm3
 ; CHECK-NEXT:    vcmpeqps %xmm3, %xmm2, %k1
 ; CHECK-NEXT:    vblendmps %xmm0, %xmm1, %xmm0 {%k1}
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <16 x float> %vec, <16 x float> undef, <4 x i32> <i32 8, i32 6, i32 10, i32 6>
   %cmp = fcmp oeq <4 x float> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x float> %shuf, <4 x float> %vec2
   ret <4 x float> %res
 }
 
 define <4 x float> @test_masked_z_16xfloat_to_4xfloat_perm_mask1(<16 x float> %vec, <4 x float> %mask) {
 ; CHECK-LABEL: test_masked_z_16xfloat_to_4xfloat_perm_mask1:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm2
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm2
 ; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
 ; CHECK-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,2]
 ; CHECK-NEXT:    vblendps {{.*#+}} xmm0 = xmm2[0],xmm0[1],xmm2[2],xmm0[3]
 ; CHECK-NEXT:    vxorps %xmm2, %xmm2, %xmm2
 ; CHECK-NEXT:    vcmpeqps %xmm2, %xmm1, %k1
 ; CHECK-NEXT:    vmovaps %xmm0, %xmm0 {%k1} {z}
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
; CHECK-NEXT:    retq
   %res = select <4 x i1> %cmp, <4 x float> %shuf, <4 x float> zeroinitializer
   ret <4 x float> %res
 }
 define <4 x float> @test_16xfloat_to_4xfloat_perm_mask3(<16 x float> %vec) {
 ; CHECK-LABEL: test_16xfloat_to_4xfloat_perm_mask3:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vmovaps {{.*#+}} ymm1 = [0,2,4,6,4,6,6,7]
 ; CHECK-NEXT:    vpermps %ymm0, %ymm1, %ymm1
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm0
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm0
 ; CHECK-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,1,3,3]
 ; CHECK-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
   %res = shufflevector <16 x float> %vec, <16 x float> undef, <4 x i32> <i32 10, i32 2, i32 11, i32 6>
   ret <4 x float> %res
 }
 define <4 x float> @test_masked_16xfloat_to_4xfloat_perm_mask3(<16 x float> %vec, <4 x float> %vec2, <4 x float> %mask) {
 ; CHECK-LABEL: test_masked_16xfloat_to_4xfloat_perm_mask3:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vmovaps {{.*#+}} ymm3 = [0,2,4,6,4,6,6,7]
 ; CHECK-NEXT:    vpermps %ymm0, %ymm3, %ymm3
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm0
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm0
 ; CHECK-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,1,3,3]
 ; CHECK-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm3[1],xmm0[2],xmm3[3]
 ; CHECK-NEXT:    vxorps %xmm3, %xmm3, %xmm3
 ; CHECK-NEXT:    vcmpeqps %xmm3, %xmm2, %k1
 ; CHECK-NEXT:    vblendmps %xmm0, %xmm1, %xmm0 {%k1}
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <16 x float> %vec, <16 x float> undef, <4 x i32> <i32 10, i32 2, i32 11, i32 6>
   %cmp = fcmp oeq <4 x float> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x float> %shuf, <4 x float> %vec2
   ret <4 x float> %res
 }
 
 define <4 x float> @test_masked_z_16xfloat_to_4xfloat_perm_mask3(<16 x float> %vec, <4 x float> %mask) {
 ; CHECK-LABEL: test_masked_z_16xfloat_to_4xfloat_perm_mask3:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vmovaps {{.*#+}} ymm2 = [0,2,4,6,4,6,6,7]
 ; CHECK-NEXT:    vpermps %ymm0, %ymm2, %ymm2
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm0
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm0
 ; CHECK-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,1,3,3]
 ; CHECK-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
 ; CHECK-NEXT:    vxorps %xmm2, %xmm2, %xmm2
 ; CHECK-NEXT:    vcmpeqps %xmm2, %xmm1, %k1
 ; CHECK-NEXT:    vmovaps %xmm0, %xmm0 {%k1} {z}
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <16 x float> %vec, <16 x float> undef, <4 x i32> <i32 10, i32 2, i32 11, i32 6>
; CHECK-NEXT:    retq
   %shuf = shufflevector <8 x double> %vec, <8 x double> undef, <4 x i32> <i32 2, i32 3, i32 2, i32 0>
   %cmp = fcmp oeq <4 x double> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x double> %shuf, <4 x double> zeroinitializer
   ret <4 x double> %res
 }
 define <4 x double> @test_8xdouble_to_4xdouble_perm_mask3(<8 x double> %vec) {
 ; CHECK-LABEL: test_8xdouble_to_4xdouble_perm_mask3:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm2
-; CHECK-NEXT:    vmovapd {{.*#+}} ymm1 = [0,2,1,4]
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm1
+; CHECK-NEXT:    vbroadcastsd %xmm1, %ymm2
+; CHECK-NEXT:    vmovapd {{.*#+}} ymm1 = [0,2,1,7]
 ; CHECK-NEXT:    vpermi2pd %ymm2, %ymm0, %ymm1
 ; CHECK-NEXT:    vmovapd %ymm1, %ymm0
 ; CHECK-NEXT:    retq
   %res = shufflevector <8 x double> %vec, <8 x double> undef, <4 x i32> <i32 0, i32 2, i32 1, i32 4>
   ret <4 x double> %res
 }
 define <4 x double> @test_masked_8xdouble_to_4xdouble_perm_mask3(<8 x double> %vec, <4 x double> %vec2, <4 x double> %mask) {
 ; CHECK-LABEL: test_masked_8xdouble_to_4xdouble_perm_mask3:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm3
-; CHECK-NEXT:    vmovapd {{.*#+}} ymm4 = [0,2,1,4]
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm3
+; CHECK-NEXT:    vbroadcastsd %xmm3, %ymm3
+; CHECK-NEXT:    vmovapd {{.*#+}} ymm4 = [0,2,1,7]
 ; CHECK-NEXT:    vpermi2pd %ymm3, %ymm0, %ymm4
 ; CHECK-NEXT:    vxorpd %xmm0, %xmm0, %xmm0
 ; CHECK-NEXT:    vcmpeqpd %ymm0, %ymm2, %k1
 ; CHECK-NEXT:    vblendmpd %ymm4, %ymm1, %ymm0 {%k1}
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <8 x double> %vec, <8 x double> undef, <4 x i32> <i32 0, i32 2, i32 1, i32 4>
   %cmp = fcmp oeq <4 x double> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x double> %shuf, <4 x double> %vec2
   ret <4 x double> %res
 }
 
 define <4 x double> @test_masked_z_8xdouble_to_4xdouble_perm_mask3(<8 x double> %vec, <4 x double> %mask) {
 ; CHECK-LABEL: test_masked_z_8xdouble_to_4xdouble_perm_mask3:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm3
-; CHECK-NEXT:    vmovapd {{.*#+}} ymm2 = [0,2,1,4]
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm2
+; CHECK-NEXT:    vbroadcastsd %xmm2, %ymm3
+; CHECK-NEXT:    vmovapd {{.*#+}} ymm2 = [0,2,1,7]
 ; CHECK-NEXT:    vxorpd %xmm4, %xmm4, %xmm4
 ; CHECK-NEXT:    vcmpeqpd %ymm4, %ymm1, %k1
 ; CHECK-NEXT:    vpermi2pd %ymm3, %ymm0, %ymm2 {%k1} {z}
 ; CHECK-NEXT:    vmovapd %ymm2, %ymm0
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <8 x double> %vec, <8 x double> undef, <4 x i32> <i32 0, i32 2, i32 1, i32 4>
   %cmp = fcmp oeq <4 x double> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x double> %shuf, <4 x double> zeroinitializer
   ret <4 x double> %res
 }
 define <4 x double> @test_masked_8xdouble_to_4xdouble_perm_mask4(<8 x double> %vec, <4 x double> %vec2, <4 x double> %mask) {
 ; CHECK-LABEL: test_masked_8xdouble_to_4xdouble_perm_mask4:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm3
-; CHECK-NEXT:    vmovapd {{.*#+}} ymm4 = [1,1,5,5]
-; CHECK-NEXT:    vpermi2pd %ymm3, %ymm0, %ymm4
-; CHECK-NEXT:    vxorpd %xmm0, %xmm0, %xmm0
-; CHECK-NEXT:    vcmpeqpd %ymm0, %ymm2, %k1
-; CHECK-NEXT:    vblendmpd %ymm4, %ymm1, %ymm0 {%k1}
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm3
+; CHECK-NEXT:    vunpckhpd {{.*#+}} xmm0 = xmm0[1],xmm3[1]
+; CHECK-NEXT:    vxorpd %xmm3, %xmm3, %xmm3
+; CHECK-NEXT:    vcmpeqpd %ymm3, %ymm2, %k1
+; CHECK-NEXT:    vpermpd {{.*#+}} ymm1 {%k1} = ymm0[0,0,1,1]
+; CHECK-NEXT:    vmovapd %ymm1, %ymm0
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <8 x double> %vec, <8 x double> undef, <4 x i32> <i32 1, i32 1, i32 5, i32 5>
   %cmp = fcmp oeq <4 x double> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x double> %shuf, <4 x double> %vec2
   ret <4 x double> %res
 }
 
 define <4 x double> @test_masked_z_8xdouble_to_4xdouble_perm_mask4(<8 x double> %vec, <4 x double> %mask) {
 ; CHECK-LABEL: test_masked_z_8xdouble_to_4xdouble_perm_mask4:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vextractf64x4 $1, %zmm0, %ymm3
-; CHECK-NEXT:    vmovapd {{.*#+}} ymm2 = [1,1,5,5]
-; CHECK-NEXT:    vxorpd %xmm4, %xmm4, %xmm4
-; CHECK-NEXT:    vcmpeqpd %ymm4, %ymm1, %k1
-; CHECK-NEXT:    vpermi2pd %ymm3, %ymm0, %ymm2 {%k1} {z}
-; CHECK-NEXT:    vmovapd %ymm2, %ymm0
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm2
+; CHECK-NEXT:    vunpckhpd {{.*#+}} xmm0 = xmm0[1],xmm2[1]
+; CHECK-NEXT:    vxorpd %xmm2, %xmm2, %xmm2
+; CHECK-NEXT:    vcmpeqpd %ymm2, %ymm1, %k1
+; CHECK-NEXT:    vpermpd {{.*#+}} ymm0 {%k1} {z} = ymm0[0,0,1,1]
 ; CHECK-NEXT:    retq
   %shuf = shufflevector <8 x double> %vec, <8 x double> undef, <4 x i32> <i32 1, i32 1, i32 5, i32 5>
   %cmp = fcmp oeq <4 x double> %mask, zeroinitializer
   %res = select <4 x i1> %cmp, <4 x double> %shuf, <4 x double> zeroinitializer
   ret <4 x double> %res
 }
 define <4 x double> @test_masked_8xdouble_to_4xdouble_perm_mask5(<8 x double> %vec, <4 x double> %vec2, <4 x double> %mask) {
 ; CHECK-LABEL: test_masked_8xdouble_to_4xdouble_perm_mask5:

llvm/trunk/test/CodeGen/X86/madd.ll

 ; AVX512-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    addq $16, %rcx
 ; AVX512-NEXT:    cmpq %rcx, %rax
 ; AVX512-NEXT:    jne .LBB2_1
 ; AVX512-NEXT:  # %bb.2: # %middle.block
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 entry:
   %3 = zext i32 %2 to i64
   br label %vector.body

 ; AVX512F-NEXT:    addq $16, %rcx
 ; AVX512F-NEXT:    cmpq %rcx, %rax
 ; AVX512F-NEXT:    jne .LBB3_1
 ; AVX512F-NEXT:  # %bb.2: # %middle.block
 ; AVX512F-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vmovd %xmm0, %eax
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: _Z10test_shortPsS_i_1024:
 ; AVX512BW:       # %bb.0: # %entry

 ; AVX512BW-NEXT:    addq $16, %rcx
 ; AVX512BW-NEXT:    cmpq %rcx, %rax
 ; AVX512BW-NEXT:    jne .LBB3_1
 ; AVX512BW-NEXT:  # %bb.2: # %middle.block
 ; AVX512BW-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vmovd %xmm0, %eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 entry:
   %3 = zext i32 %2 to i64
   br label %vector.body

 ; AVX512-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    addq $16, %rcx
 ; AVX512-NEXT:    cmpq %rcx, %rax
 ; AVX512-NEXT:    jne .LBB6_1
 ; AVX512-NEXT:  # %bb.2: # %middle.block
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 entry:
   %3 = zext i32 %2 to i64
   br label %vector.body

 ; AVX512F-NEXT:    addq $32, %rcx
 ; AVX512F-NEXT:    cmpq %rcx, %rax
 ; AVX512F-NEXT:    jne .LBB7_1
 ; AVX512F-NEXT:  # %bb.2: # %middle.block
 ; AVX512F-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vmovd %xmm0, %eax
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: _Z9test_charPcS_i_1024:
 ; AVX512BW:       # %bb.0: # %entry

 ; AVX512BW-NEXT:    addq $32, %rcx
 ; AVX512BW-NEXT:    cmpq %rcx, %rax
 ; AVX512BW-NEXT:    jne .LBB7_1
 ; AVX512BW-NEXT:  # %bb.2: # %middle.block
 ; AVX512BW-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vmovd %xmm0, %eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 entry:
   %3 = zext i32 %2 to i64
   br label %vector.body

 ; AVX512-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    addq $16, %rcx
 ; AVX512-NEXT:    cmpq %rcx, %rax
 ; AVX512-NEXT:    jne .LBB10_1
 ; AVX512-NEXT:  # %bb.2: # %middle.block
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 entry:
   %3 = zext i32 %2 to i64
   br label %vector.body

 ; AVX512-NEXT:    addq $16, %rcx
 ; AVX512-NEXT:    cmpq %rcx, %rax
 ; AVX512-NEXT:    jne .LBB11_1
 ; AVX512-NEXT:  # %bb.2: # %middle.block
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 entry:
   %3 = zext i32 %2 to i64
   br label %vector.body

llvm/trunk/test/CodeGen/X86/min-legal-vector-width.ll

 ; CHECK-NEXT:    addq $32, %rcx
 ; CHECK-NEXT:    cmpq %rcx, %rax
 ; CHECK-NEXT:    jne .LBB9_1
 ; CHECK-NEXT:  # %bb.2: # %middle.block
 ; CHECK-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; CHECK-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; CHECK-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; CHECK-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; CHECK-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; CHECK-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vmovd %xmm0, %eax
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
 entry:
   %3 = zext i32 %2 to i64
   br label %vector.body

 ; CHECK-NEXT:    vpsadbw b+1024(%rax), %xmm1, %xmm1
 ; CHECK-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; CHECK-NEXT:    addq $4, %rax
 ; CHECK-NEXT:    jne .LBB11_1
 ; CHECK-NEXT:  # %bb.2: # %middle.block
 ; CHECK-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; CHECK-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; CHECK-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; CHECK-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; CHECK-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vmovd %xmm0, %eax
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
 entry:
   br label %vector.body
 

llvm/trunk/test/CodeGen/X86/sad.ll

 ; AVX512-NEXT:    vpsadbw b+1024(%rax), %xmm1, %xmm1
 ; AVX512-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    addq $4, %rax
 ; AVX512-NEXT:    jne .LBB0_1
 ; AVX512-NEXT:  # %bb.2: # %middle.block
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 entry:
   br label %vector.body
 

 ; AVX512-NEXT:    vpaddd %zmm1, %zmm2, %zmm1
 ; AVX512-NEXT:    addq $4, %rax
 ; AVX512-NEXT:    jne .LBB1_1
 ; AVX512-NEXT:  # %bb.2: # %middle.block
 ; AVX512-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 entry:
   br label %vector.body
 

 ; AVX512F-NEXT:    jne .LBB2_1
 ; AVX512F-NEXT:  # %bb.2: # %middle.block
 ; AVX512F-NEXT:    vpaddd %zmm2, %zmm0, %zmm0
 ; AVX512F-NEXT:    vpaddd %zmm3, %zmm1, %zmm1
 ; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512F-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512F-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512F-NEXT:    vmovd %xmm0, %eax
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: sad_avx64i8:
 ; AVX512BW:       # %bb.0: # %entry

 ; AVX512BW-NEXT:    jne .LBB2_1
 ; AVX512BW-NEXT:  # %bb.2: # %middle.block
 ; AVX512BW-NEXT:    vpaddd %zmm0, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpaddd %zmm0, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BW-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512BW-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vmovd %xmm0, %eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 entry:
   br label %vector.body
 

 ; AVX512-NEXT:    vmovd %eax, %xmm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vmovdqu (%rdx), %xmm1
 ; AVX512-NEXT:    vpsadbw (%rcx), %xmm1, %xmm1
 ; AVX512-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 bb:
   %tmp = load <16 x i8>, <16 x i8>* %arg, align 1
   %tmp4 = load <16 x i8>, <16 x i8>* %arg1, align 1

 ; AVX512-NEXT:    vmovdqu (%rdi), %xmm0
 ; AVX512-NEXT:    vmovdqu (%rdx), %xmm1
 ; AVX512-NEXT:    vpsadbw (%rsi), %xmm0, %xmm0
 ; AVX512-NEXT:    vpsadbw (%rcx), %xmm1, %xmm1
 ; AVX512-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 bb:
   %tmp = load <16 x i8>, <16 x i8>* %arg, align 1
   %tmp4 = load <16 x i8>, <16 x i8>* %arg1, align 1

llvm/trunk/test/CodeGen/X86/vector-reduce-add-widen.ll

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddq %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpaddq %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.add.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddq %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpaddq %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.add.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.add.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.add.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> %a0)
   ret i8 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v128i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> %a0)
   ret i8 %1

llvm/trunk/test/CodeGen/X86/vector-reduce-add.ll

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddq %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpaddq %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.add.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddq %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpaddq %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.add.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.add.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.add.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpaddw %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpaddw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> %a0)
   ret i8 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v128i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> %a0)
   ret i8 %1

llvm/trunk/test/CodeGen/X86/vector-reduce-and-widen.ll

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> %a0)
   ret i8 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v128i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> %a0)
   ret i8 %1

llvm/trunk/test/CodeGen/X86/vector-reduce-and.ll

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> %a0)
   ret i8 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v128i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> %a0)
   ret i8 %1

llvm/trunk/test/CodeGen/X86/vector-reduce-fadd-fast.ll

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm1, %ymm0
 ; AVX512-NEXT:    vaddps %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast float @llvm.experimental.vector.reduce.fadd.f32.f32.v16f32(float %a0, <16 x float> %a1)
   ret float %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f32_zero:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast float @llvm.experimental.vector.reduce.fadd.f32.f32.v16f32(float 0.0, <16 x float> %a0)
   ret float %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f32_undef:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast float @llvm.experimental.vector.reduce.fadd.f32.f32.v16f32(float undef, <16 x float> %a0)
   ret float %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8f64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm1, %ymm0
 ; AVX512-NEXT:    vaddpd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fadd.f64.f64.v8f64(double %a0, <8 x double> %a1)
   ret double %1
 }
 

 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vaddpd %zmm2, %zmm1, %zmm0
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fadd.f64.f64.v16f64(double %a0, <16 x double> %a1)
   ret double %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8f64_zero:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fadd.f64.f64.v8f64(double 0.0, <8 x double> %a0)
   ret double %1
 }
 

 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f64_zero:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fadd.f64.f64.v16f64(double 0.0, <16 x double> %a0)
   ret double %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8f64_undef:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fadd.f64.f64.v8f64(double undef, <8 x double> %a0)
   ret double %1
 }
 

 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f64_undef:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fadd.f64.f64.v16f64(double undef, <16 x double> %a0)
   ret double %1
 }
 

llvm/trunk/test/CodeGen/X86/vector-reduce-fmul-fast.ll

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm1, %ymm0
 ; AVX512-NEXT:    vmulps %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; AVX512-NEXT:    vmulps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast float @llvm.experimental.vector.reduce.fmul.f32.f32.v16f32(float %a0, <16 x float> %a1)
   ret float %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f32_zero:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vmulps %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; AVX512-NEXT:    vmulps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast float @llvm.experimental.vector.reduce.fmul.f32.f32.v16f32(float 1.0, <16 x float> %a0)
   ret float %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f32_undef:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vmulps %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
-; AVX512-NEXT:    vmulps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast float @llvm.experimental.vector.reduce.fmul.f32.f32.v16f32(float undef, <16 x float> %a0)
   ret float %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8f64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm1, %ymm0
 ; AVX512-NEXT:    vmulpd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fmul.f64.f64.v8f64(double %a0, <8 x double> %a1)
   ret double %1
 }
 

 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vmulpd %zmm2, %zmm1, %zmm0
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fmul.f64.f64.v16f64(double %a0, <16 x double> %a1)
   ret double %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8f64_zero:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fmul.f64.f64.v8f64(double 1.0, <8 x double> %a0)
   ret double %1
 }
 

 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f64_zero:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fmul.f64.f64.v16f64(double 1.0, <16 x double> %a0)
   ret double %1
 }
 

 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8f64_undef:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fmul.f64.f64.v8f64(double undef, <8 x double> %a0)
   ret double %1
 }
 

 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16f64_undef:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm0[1,0]
 ; AVX512-NEXT:    vmulpd %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call fast double @llvm.experimental.vector.reduce.fmul.f64.f64.v16f64(double undef, <16 x double> %a0)
   ret double %1
 }
 

llvm/trunk/test/CodeGen/X86/vector-reduce-mul-widen.ll

 ; AVX512DQ-NEXT:    vzeroupper
 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512DQVL-LABEL: test_v8i64:
 ; AVX512DQVL:       # %bb.0:
 ; AVX512DQVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
 ; AVX512DQVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
+; AVX512DQVL-NEXT:    vpmullq %xmm1, %xmm0, %xmm0
 ; AVX512DQVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512DQVL-NEXT:    vpmullq %xmm1, %xmm0, %xmm0
 ; AVX512DQVL-NEXT:    vmovq %xmm0, %rax
 ; AVX512DQVL-NEXT:    vzeroupper
 ; AVX512DQVL-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512DQVL-LABEL: test_v16i64:
 ; AVX512DQVL:       # %bb.0:
 ; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
 ; AVX512DQVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
 ; AVX512DQVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
+; AVX512DQVL-NEXT:    vpmullq %xmm1, %xmm0, %xmm0
 ; AVX512DQVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512DQVL-NEXT:    vpmullq %xmm1, %xmm0, %xmm0
 ; AVX512DQVL-NEXT:    vmovq %xmm0, %rax
 ; AVX512DQVL-NEXT:    vzeroupper
 ; AVX512DQVL-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_v32i16:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vmovd %xmm0, %eax
 ; AVX512BW-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512BWVL-LABEL: test_v32i16:
 ; AVX512BWVL:       # %bb.0:
 ; AVX512BWVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vmovd %xmm0, %eax
 ; AVX512BWVL-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: test_v32i16:

 ; AVX2-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_v64i16:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vmovd %xmm0, %eax
 ; AVX512BW-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512BWVL-LABEL: test_v64i16:
 ; AVX512BWVL:       # %bb.0:
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vmovd %xmm0, %eax
 ; AVX512BWVL-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: test_v64i16:

 ; AVX2-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_v128i8:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512BW-NEXT:    vpmullw %zmm2, %zmm3, %zmm3
-; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BW-NEXT:    vpmullw %zmm2, %zmm3, %zmm2
+; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
+; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm4 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
-; AVX512BW-NEXT:    vpmullw %zmm4, %zmm3, %zmm3
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BW-NEXT:    vpmullw %zmm4, %zmm2, %zmm2
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
+; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm4 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
-; AVX512BW-NEXT:    vpmullw %zmm4, %zmm3, %zmm3
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BW-NEXT:    vpmullw %zmm4, %zmm2, %zmm2
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,2,3]
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BW-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm3
+; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BW-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpsrlw $8, %xmm0, %xmm3
+; AVX512BW-NEXT:    vpsrlw $8, %xmm0, %xmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BW-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512BW-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512BWVL-LABEL: test_v128i8:
 ; AVX512BWVL:       # %bb.0:
 ; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
 ; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm3, %zmm3
-; AVX512BWVL-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm3, %zmm2
+; AVX512BWVL-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
+; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm4 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
-; AVX512BWVL-NEXT:    vpmullw %zmm4, %zmm3, %zmm3
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BWVL-NEXT:    vpmullw %zmm4, %zmm2, %zmm2
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
+; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm4 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
-; AVX512BWVL-NEXT:    vpmullw %zmm4, %zmm3, %zmm3
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BWVL-NEXT:    vpmullw %zmm4, %zmm2, %zmm2
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512BWVL-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,2,3]
+; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BWVL-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpsrld $16, %xmm0, %xmm3
+; AVX512BWVL-NEXT:    vpsrld $16, %xmm0, %xmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BWVL-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpsrlw $8, %xmm0, %xmm3
+; AVX512BWVL-NEXT:    vpsrlw $8, %xmm0, %xmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BWVL-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512BWVL-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: test_v128i8:
 ; AVX512DQ:       # %bb.0:

llvm/trunk/test/CodeGen/X86/vector-reduce-mul.ll

 ; AVX512DQ-NEXT:    vzeroupper
 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512DQVL-LABEL: test_v8i64:
 ; AVX512DQVL:       # %bb.0:
 ; AVX512DQVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
 ; AVX512DQVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
+; AVX512DQVL-NEXT:    vpmullq %xmm1, %xmm0, %xmm0
 ; AVX512DQVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512DQVL-NEXT:    vpmullq %xmm1, %xmm0, %xmm0
 ; AVX512DQVL-NEXT:    vmovq %xmm0, %rax
 ; AVX512DQVL-NEXT:    vzeroupper
 ; AVX512DQVL-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512DQVL-LABEL: test_v16i64:
 ; AVX512DQVL:       # %bb.0:
 ; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
 ; AVX512DQVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
 ; AVX512DQVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512DQVL-NEXT:    vpmullq %zmm1, %zmm0, %zmm0
+; AVX512DQVL-NEXT:    vpmullq %xmm1, %xmm0, %xmm0
 ; AVX512DQVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512DQVL-NEXT:    vpmullq %xmm1, %xmm0, %xmm0
 ; AVX512DQVL-NEXT:    vmovq %xmm0, %rax
 ; AVX512DQVL-NEXT:    vzeroupper
 ; AVX512DQVL-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpmulld %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_v32i16:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vmovd %xmm0, %eax
 ; AVX512BW-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512BWVL-LABEL: test_v32i16:
 ; AVX512BWVL:       # %bb.0:
 ; AVX512BWVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vmovd %xmm0, %eax
 ; AVX512BWVL-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: test_v32i16:

 ; AVX2-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_v64i16:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vmovd %xmm0, %eax
 ; AVX512BW-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512BWVL-LABEL: test_v64i16:
 ; AVX512BWVL:       # %bb.0:
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512BWVL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BWVL-NEXT:    vmovd %xmm0, %eax
 ; AVX512BWVL-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: test_v64i16:

 ; AVX2-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_v128i8:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512BW-NEXT:    vpmullw %zmm2, %zmm3, %zmm3
-; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BW-NEXT:    vpmullw %zmm2, %zmm3, %zmm2
+; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
+; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm4 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
-; AVX512BW-NEXT:    vpmullw %zmm4, %zmm3, %zmm3
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BW-NEXT:    vpmullw %zmm4, %zmm2, %zmm2
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
+; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm4 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
-; AVX512BW-NEXT:    vpmullw %zmm4, %zmm3, %zmm3
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BW-NEXT:    vpmullw %zmm4, %zmm2, %zmm2
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,2,3]
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BW-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm3
+; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BW-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpsrlw $8, %xmm0, %xmm3
+; AVX512BW-NEXT:    vpsrlw $8, %xmm0, %xmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BW-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512BW-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512BWVL-LABEL: test_v128i8:
 ; AVX512BWVL:       # %bb.0:
 ; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
 ; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm3, %zmm3
-; AVX512BWVL-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm3, %zmm2
+; AVX512BWVL-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
+; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm4 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
-; AVX512BWVL-NEXT:    vpmullw %zmm4, %zmm3, %zmm3
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BWVL-NEXT:    vpmullw %zmm4, %zmm2, %zmm2
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm3 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
+; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BWVL-NEXT:    vpunpckhbw {{.*#+}} zmm4 = zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[10],zmm0[10],zmm1[11],zmm0[11],zmm1[12],zmm0[12],zmm1[13],zmm0[13],zmm1[14],zmm0[14],zmm1[15],zmm0[15],zmm1[24],zmm0[24],zmm1[25],zmm0[25],zmm1[26],zmm0[26],zmm1[27],zmm0[27],zmm1[28],zmm0[28],zmm1[29],zmm0[29],zmm1[30],zmm0[30],zmm1[31],zmm0[31],zmm1[40],zmm0[40],zmm1[41],zmm0[41],zmm1[42],zmm0[42],zmm1[43],zmm0[43],zmm1[44],zmm0[44],zmm1[45],zmm0[45],zmm1[46],zmm0[46],zmm1[47],zmm0[47],zmm1[56],zmm0[56],zmm1[57],zmm0[57],zmm1[58],zmm0[58],zmm1[59],zmm0[59],zmm1[60],zmm0[60],zmm1[61],zmm0[61],zmm1[62],zmm0[62],zmm1[63],zmm0[63]
-; AVX512BWVL-NEXT:    vpmullw %zmm4, %zmm3, %zmm3
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm3, %zmm3
+; AVX512BWVL-NEXT:    vpmullw %zmm4, %zmm2, %zmm2
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpackuswb %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[2],zmm0[2],zmm1[3],zmm0[3],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[6],zmm0[6],zmm1[7],zmm0[7],zmm1[16],zmm0[16],zmm1[17],zmm0[17],zmm1[18],zmm0[18],zmm1[19],zmm0[19],zmm1[20],zmm0[20],zmm1[21],zmm0[21],zmm1[22],zmm0[22],zmm1[23],zmm0[23],zmm1[32],zmm0[32],zmm1[33],zmm0[33],zmm1[34],zmm0[34],zmm1[35],zmm0[35],zmm1[36],zmm0[36],zmm1[37],zmm0[37],zmm1[38],zmm0[38],zmm1[39],zmm0[39],zmm1[48],zmm0[48],zmm1[49],zmm0[49],zmm1[50],zmm0[50],zmm1[51],zmm0[51],zmm1[52],zmm0[52],zmm1[53],zmm0[53],zmm1[54],zmm0[54],zmm1[55],zmm0[55]
 ; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512BWVL-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,2,3]
+; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BWVL-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpsrld $16, %xmm0, %xmm3
+; AVX512BWVL-NEXT:    vpsrld $16, %xmm0, %xmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BWVL-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpsrlw $8, %xmm0, %xmm3
+; AVX512BWVL-NEXT:    vpsrlw $8, %xmm0, %xmm2
 ; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
-; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm3 = zmm3[0],zmm0[0],zmm3[1],zmm0[1],zmm3[2],zmm0[2],zmm3[3],zmm0[3],zmm3[4],zmm0[4],zmm3[5],zmm0[5],zmm3[6],zmm0[6],zmm3[7],zmm0[7],zmm3[16],zmm0[16],zmm3[17],zmm0[17],zmm3[18],zmm0[18],zmm3[19],zmm0[19],zmm3[20],zmm0[20],zmm3[21],zmm0[21],zmm3[22],zmm0[22],zmm3[23],zmm0[23],zmm3[32],zmm0[32],zmm3[33],zmm0[33],zmm3[34],zmm0[34],zmm3[35],zmm0[35],zmm3[36],zmm0[36],zmm3[37],zmm0[37],zmm3[38],zmm0[38],zmm3[39],zmm0[39],zmm3[48],zmm0[48],zmm3[49],zmm0[49],zmm3[50],zmm0[50],zmm3[51],zmm0[51],zmm3[52],zmm0[52],zmm3[53],zmm0[53],zmm3[54],zmm0[54],zmm3[55],zmm0[55]
-; AVX512BWVL-NEXT:    vpmullw %zmm3, %zmm0, %zmm0
-; AVX512BWVL-NEXT:    vpandq %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpunpcklbw {{.*#+}} zmm2 = zmm2[0],zmm0[0],zmm2[1],zmm0[1],zmm2[2],zmm0[2],zmm2[3],zmm0[3],zmm2[4],zmm0[4],zmm2[5],zmm0[5],zmm2[6],zmm0[6],zmm2[7],zmm0[7],zmm2[16],zmm0[16],zmm2[17],zmm0[17],zmm2[18],zmm0[18],zmm2[19],zmm0[19],zmm2[20],zmm0[20],zmm2[21],zmm0[21],zmm2[22],zmm0[22],zmm2[23],zmm0[23],zmm2[32],zmm0[32],zmm2[33],zmm0[33],zmm2[34],zmm0[34],zmm2[35],zmm0[35],zmm2[36],zmm0[36],zmm2[37],zmm0[37],zmm2[38],zmm0[38],zmm2[39],zmm0[39],zmm2[48],zmm0[48],zmm2[49],zmm0[49],zmm2[50],zmm0[50],zmm2[51],zmm0[51],zmm2[52],zmm0[52],zmm2[53],zmm0[53],zmm2[54],zmm0[54],zmm2[55],zmm0[55]
+; AVX512BWVL-NEXT:    vpmullw %zmm2, %zmm0, %zmm0
+; AVX512BWVL-NEXT:    vpandq %zmm3, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512BWVL-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512BWVL-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: test_v128i8:
 ; AVX512DQ:       # %bb.0:

llvm/trunk/test/CodeGen/X86/vector-reduce-or-widen.ll

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> %a0)
   ret i8 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v128i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> %a0)
   ret i8 %1

llvm/trunk/test/CodeGen/X86/vector-reduce-or.ll

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> %a0)
   ret i8 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v128i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> %a0)
   ret i8 %1

llvm/trunk/test/CodeGen/X86/vector-reduce-xor-widen.ll

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> %a0)
   ret i8 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v128i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> %a0)
   ret i8 %1

llvm/trunk/test/CodeGen/X86/vector-reduce-xor.ll

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i64:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovq %xmm0, %rax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> %a0)
   ret i64 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v16i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i32:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxord %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> %a0)
   ret i32 %1
 }

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v32i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vmovd %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> %a0)
   ret i16 %1

 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v64i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> %a0)
   ret i8 %1

 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v128i8:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm1
-; AVX512-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm1
 ; AVX512-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX512-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> %a0)
   ret i8 %1

llvm/trunk/test/CodeGen/X86/vector-shuffle-512-v16.ll

; ALL-NEXT:    retq
   %v2 = shufflevector <16 x float> %v_a, <16 x float> undef, <8 x i32>  <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
   ret <8 x float> %v2
 }
 
 ;FIXME: can do better with vcompressp
 define <8 x float> @test_v16f32_0_1_2_3_4_6_7_10 (<16 x float> %v) {
 ; ALL-LABEL: test_v16f32_0_1_2_3_4_6_7_10:
 ; ALL:       # %bb.0:
-; ALL-NEXT:    vextractf64x4 $1, %zmm0, %ymm1
+; ALL-NEXT:    vextractf32x4 $2, %zmm0, %xmm1
 ; ALL-NEXT:    vmovsldup {{.*#+}} xmm1 = xmm1[0,0,2,2]
 ; ALL-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1
 ; ALL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,6,7,u]
 ; ALL-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6],ymm1[7]
 ; ALL-NEXT:    retq
   %res = shufflevector <16 x float> %v, <16 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 6, i32 7, i32 10>
   ret <8 x float> %res
 }