[X86] Fix a bad early out in combineExtInVec that prevented recursive shuffle combini...

llvm-svn: 367798

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 447265c32ea368c0493e5dea3c2d32c69e676467..104623774cc46ee7cf2baeee6be19e4ecb7a16ac 100644 (file)
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -44400,14 +44400,12 @@ static SDValue combineExtInVec(SDNode *N, SelectionDAG &DAG,
     }
   }
 
+  // Combine (ext_invec (ext_invec X)) -> (ext_invec X)
   // Disabling for widening legalization for now. We can enable if we find a
   // case that needs it. Otherwise it can be deleted when we switch to
   // widening legalization.
-  if (ExperimentalVectorWideningLegalization)
-    return SDValue();
-
-  // Combine (ext_invec (ext_invec X)) -> (ext_invec X)
-  if (In.getOpcode() == N->getOpcode() &&
+  if (!ExperimentalVectorWideningLegalization &&
+      In.getOpcode() == N->getOpcode() &&
       TLI.isTypeLegal(VT) && TLI.isTypeLegal(In.getOperand(0).getValueType()))
     return DAG.getNode(N->getOpcode(), SDLoc(N), VT, In.getOperand(0));
 

diff --git a/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll b/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
index a5c060e4e1e0c12cb6e99b0a8e5e0310a08f6046..023393574d0a2ec1bda8a99071b0d3d69ae0e72e 100644 (file)
--- a/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
+++ b/llvm/test/CodeGen/X86/vector-reduce-mul-widen.ll
@@ -1665,33 +1665,55 @@ define i8 @test_v8i8(<8 x i8> %a0) {
 ; AVX-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; AVX-NEXT:    vpmullw %xmm0, %xmm1, %xmm0
-; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
-; AVX-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
-; AVX-NEXT:    vpmullw %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; AVX-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[2],zero,xmm0[6],zero,xmm0[10],zero,xmm0[14],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero
 ; AVX-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX-NEXT:    # kill: def $al killed $al killed $eax
 ; AVX-NEXT:    retq
 ;
-; AVX512-LABEL: test_v8i8:
-; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX512-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
-; AVX512-NEXT:    vpmullw %xmm0, %xmm1, %xmm0
-; AVX512-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
-; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX512-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
-; AVX512-NEXT:    vpmullw %xmm0, %xmm1, %xmm0
-; AVX512-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[2],zero,xmm0[6],zero,xmm0[10],zero,xmm0[14],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero
-; AVX512-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX512-NEXT:    vpextrb $0, %xmm0, %eax
-; AVX512-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512-NEXT:    retq
+; AVX512BW-LABEL: test_v8i8:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX512BW-NEXT:    vpmullw %xmm0, %xmm1, %xmm0
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[2],zero,xmm0[6],zero,xmm0[10],zero,xmm0[14],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero
+; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX512BW-NEXT:    vpextrb $0, %xmm0, %eax
+; AVX512BW-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512BW-NEXT:    retq
+;
+; AVX512VL-LABEL: test_v8i8:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX512VL-NEXT:    vpmullw %xmm0, %xmm1, %xmm0
+; AVX512VL-NEXT:    vpalignr {{.*#+}} xmm1 = xmm0[4,5,6,7,8,9,10,11,12,13,14,15,0,1,2,3]
+; AVX512VL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[2],zero,xmm0[6],zero,xmm0[10],zero,xmm0[14],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero
+; AVX512VL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vpextrb $0, %xmm0, %eax
+; AVX512VL-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512VL-NEXT:    retq
+;
+; AVX512DQ-LABEL: test_v8i8:
+; AVX512DQ:       # %bb.0:
+; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX512DQ-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX512DQ-NEXT:    vpmullw %xmm0, %xmm1, %xmm0
+; AVX512DQ-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; AVX512DQ-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX512DQ-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[2],zero,xmm0[6],zero,xmm0[10],zero,xmm0[14],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero,xmm0[u],zero
+; AVX512DQ-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX512DQ-NEXT:    vpextrb $0, %xmm0, %eax
+; AVX512DQ-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512DQ-NEXT:    retq
   %1 = call i8 @llvm.experimental.vector.reduce.mul.v8i8(<8 x i8> %a0)
   ret i8 %1
 }
@@ -1785,28 +1807,18 @@ define i8 @test_v16i8(<16 x i8> %a0) {
 ;
 ; AVX2-LABEL: test_v16i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
-; AVX2-NEXT:    vpmullw %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpand {{.*}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm1
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,2,3]
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
-; AVX2-NEXT:    vpmullw %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX2-NEXT:    vpackuswb %xmm0, %xmm2, %xmm2
-; AVX2-NEXT:    vpsrld $16, %xmm2, %xmm2
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
-; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm1
-; AVX2-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX2-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
+; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[4],zero,xmm0[6],zero,xmm0[4],zero,xmm0[6],zero,xmm0[8],zero,xmm0[10],zero,xmm0[12],zero,xmm0[14],zero
+; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm0[1,1,2,3,4,5,6,7]
 ; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpextrb $0, %xmm0, %eax
 ; AVX2-NEXT:    # kill: def $al killed $al killed $eax
-; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_v16i8:
@@ -1992,14 +2004,12 @@ define i8 @test_v32i8(<32 x i8> %a0) {
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX1-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX1-NEXT:    vpmullw %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
 ; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
@@ -2028,32 +2038,25 @@ define i8 @test_v32i8(<32 x i8> %a0) {
 ; AVX2-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX2-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %xmm3, %xmm2, %xmm4
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX2-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm4, %xmm0, %xmm0
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX2-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpsrld $16, %xmm0, %xmm2
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX2-NEXT:    vpsrld $16, %xmm2, %xmm2
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpsrlw $8, %xmm0, %xmm1
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm1
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX2-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpextrb $0, %xmm0, %eax
@@ -2123,32 +2126,25 @@ define i8 @test_v32i8(<32 x i8> %a0) {
 ; AVX512DQ-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX512DQ-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
-; AVX512DQ-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm2, %xmm4
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm4, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512DQ-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512DQ-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpsrld $16, %xmm0, %xmm2
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512DQ-NEXT:    vpsrld $16, %xmm2, %xmm2
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpsrlw $8, %xmm0, %xmm1
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm1
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX512DQ-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512DQ-NEXT:    vpextrb $0, %xmm0, %eax
@@ -2295,23 +2291,17 @@ define i8 @test_v64i8(<64 x i8> %a0) {
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX1-NEXT:    vpmullw %xmm3, %xmm5, %xmm3
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm5, %xmm3
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
 ; AVX1-NEXT:    vpmullw %xmm2, %xmm4, %xmm2
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm2
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX1-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX1-NEXT:    vpmullw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm5, %xmm0
+; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
+; AVX1-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpmullw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
@@ -2350,32 +2340,25 @@ define i8 @test_v64i8(<64 x i8> %a0) {
 ; AVX2-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX2-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %xmm3, %xmm2, %xmm4
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX2-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm4, %xmm0, %xmm0
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX2-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpsrld $16, %xmm0, %xmm2
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX2-NEXT:    vpsrld $16, %xmm2, %xmm2
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpsrlw $8, %xmm0, %xmm1
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm1
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX2-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpextrb $0, %xmm0, %eax
@@ -2400,32 +2383,25 @@ define i8 @test_v64i8(<64 x i8> %a0) {
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX512BW-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
-; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX512BW-NEXT:    vpand %xmm3, %xmm2, %xmm4
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX512BW-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpackuswb %xmm4, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512BW-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512BW-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; AVX512BW-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512BW-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX512BW-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512BW-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm2
-; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512BW-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512BW-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512BW-NEXT:    vpsrld $16, %xmm2, %xmm2
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512BW-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpsrlw $8, %xmm0, %xmm1
-; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512BW-NEXT:    vpand %xmm1, %xmm0, %xmm1
+; AVX512BW-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpextrb $0, %xmm0, %eax
@@ -2492,32 +2468,25 @@ define i8 @test_v64i8(<64 x i8> %a0) {
 ; AVX512DQ-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX512DQ-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
-; AVX512DQ-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm2, %xmm4
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm4, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512DQ-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512DQ-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpsrld $16, %xmm0, %xmm2
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512DQ-NEXT:    vpsrld $16, %xmm2, %xmm2
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpsrlw $8, %xmm0, %xmm1
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm1
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX512DQ-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512DQ-NEXT:    vpextrb $0, %xmm0, %eax
@@ -2714,8 +2683,8 @@ define i8 @test_v128i8(<128 x i8> %a0) {
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm4
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm5 = xmm2[8],xmm0[8],xmm2[9],xmm0[9],xmm2[10],xmm0[10],xmm2[11],xmm0[11],xmm2[12],xmm0[12],xmm2[13],xmm0[13],xmm2[14],xmm0[14],xmm2[15],xmm0[15]
-; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm9
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm8 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm8
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm9 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm11
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm7 = xmm11[8],xmm0[8],xmm11[9],xmm0[9],xmm11[10],xmm0[10],xmm11[11],xmm0[11],xmm11[12],xmm0[12],xmm11[13],xmm0[13],xmm11[14],xmm0[14],xmm11[15],xmm0[15]
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
@@ -2727,40 +2696,30 @@ define i8 @test_v128i8(<128 x i8> %a0) {
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm5
 ; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm6 = xmm5[8],xmm0[8],xmm5[9],xmm0[9],xmm5[10],xmm0[10],xmm5[11],xmm0[11],xmm5[12],xmm0[12],xmm5[13],xmm0[13],xmm5[14],xmm0[14],xmm5[15],xmm0[15]
 ; AVX1-NEXT:    vpmullw %xmm10, %xmm6, %xmm6
-; AVX1-NEXT:    vpmullw %xmm6, %xmm9, %xmm9
+; AVX1-NEXT:    vpmullw %xmm6, %xmm8, %xmm6
+; AVX1-NEXT:    vpmullw %xmm6, %xmm9, %xmm6
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm7 = xmm7[0],zero,xmm7[1],zero,xmm7[2],zero,xmm7[3],zero,xmm7[4],zero,xmm7[5],zero,xmm7[6],zero,xmm7[7],zero
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero,xmm5[4],zero,xmm5[5],zero,xmm5[6],zero,xmm5[7],zero
 ; AVX1-NEXT:    vpmullw %xmm7, %xmm5, %xmm5
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm7, %xmm5, %xmm5
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm6 = xmm11[0],zero,xmm11[1],zero,xmm11[2],zero,xmm11[3],zero,xmm11[4],zero,xmm11[5],zero,xmm11[6],zero,xmm11[7],zero
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm7 = xmm11[0],zero,xmm11[1],zero,xmm11[2],zero,xmm11[3],zero,xmm11[4],zero,xmm11[5],zero,xmm11[6],zero,xmm11[7],zero
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
-; AVX1-NEXT:    vpmullw %xmm6, %xmm4, %xmm4
-; AVX1-NEXT:    vpshufb %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpmullw %xmm7, %xmm4, %xmm4
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX1-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm7, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX1-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpshufb %xmm7, %xmm1, %xmm1
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm5[0],zero,xmm5[1],zero,xmm5[2],zero,xmm5[3],zero,xmm5[4],zero,xmm5[5],zero,xmm5[6],zero,xmm5[7],zero
-; AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpshufb %xmm7, %xmm1, %xmm1
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm3
+; AVX1-NEXT:    vpand %xmm2, %xmm5, %xmm4
+; AVX1-NEXT:    vpmullw %xmm3, %xmm4, %xmm3
+; AVX1-NEXT:    vpand %xmm2, %xmm3, %xmm3
+; AVX1-NEXT:    vpmullw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpmullw %xmm6, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX1-NEXT:    vpmullw %xmm0, %xmm9, %xmm0
-; AVX1-NEXT:    vpmullw %xmm0, %xmm8, %xmm0
 ; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
@@ -2807,32 +2766,25 @@ define i8 @test_v128i8(<128 x i8> %a0) {
 ; AVX2-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX2-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %xmm3, %xmm2, %xmm4
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX2-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm4, %xmm0, %xmm0
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX2-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpsrld $16, %xmm0, %xmm2
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX2-NEXT:    vpsrld $16, %xmm2, %xmm2
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX2-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpsrlw $8, %xmm0, %xmm1
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm1
+; AVX2-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX2-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX2-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpextrb $0, %xmm0, %eax
@@ -2864,32 +2816,25 @@ define i8 @test_v128i8(<128 x i8> %a0) {
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX512BW-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
-; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX512BW-NEXT:    vpand %xmm3, %xmm2, %xmm4
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX512BW-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpackuswb %xmm4, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512BW-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512BW-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; AVX512BW-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512BW-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX512BW-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512BW-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpsrld $16, %xmm0, %xmm2
-; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512BW-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512BW-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512BW-NEXT:    vpsrld $16, %xmm2, %xmm2
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512BW-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpsrlw $8, %xmm0, %xmm1
-; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512BW-NEXT:    vpand %xmm1, %xmm0, %xmm1
+; AVX512BW-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX512BW-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vpextrb $0, %xmm0, %eax
@@ -2971,32 +2916,25 @@ define i8 @test_v128i8(<128 x i8> %a0) {
 ; AVX512DQ-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; AVX512DQ-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
-; AVX512DQ-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm2, %xmm4
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm4, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512DQ-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512DQ-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpsrld $16, %xmm0, %xmm2
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm2
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX512DQ-NEXT:    vpsrld $16, %xmm2, %xmm2
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpsrlw $8, %xmm0, %xmm1
-; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX512DQ-NEXT:    vpand %xmm1, %xmm0, %xmm1
+; AVX512DQ-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX512DQ-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX512DQ-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
 ; AVX512DQ-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
 ; AVX512DQ-NEXT:    vpextrb $0, %xmm0, %eax

diff --git a/llvm/test/CodeGen/X86/vector-zext-widen.ll b/llvm/test/CodeGen/X86/vector-zext-widen.ll
index 90dece81b94c86e45207b6688cfeca7182b3eab6..78c36d6fc8fdac5aa0fcdd9c10c381446cb30aa0 100644 (file)
--- a/llvm/test/CodeGen/X86/vector-zext-widen.ll
+++ b/llvm/test/CodeGen/X86/vector-zext-widen.ll
@@ -1717,11 +1717,33 @@ define <2 x i64> @shuf_zext_16i8_to_2i64_offset6(<16 x i8> %A) nounwind uwtable
 ; SSE41-NEXT:    pmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: shuf_zext_16i8_to_2i64_offset6:
-; AVX:       # %bb.0: # %entry
-; AVX-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
-; AVX-NEXT:    retq
+; AVX1-LABEL: shuf_zext_16i8_to_2i64_offset6:
+; AVX1:       # %bb.0: # %entry
+; AVX1-NEXT:    vpsrlq $48, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT:    retq
+;
+; AVX2-SLOW-LABEL: shuf_zext_16i8_to_2i64_offset6:
+; AVX2-SLOW:       # %bb.0: # %entry
+; AVX2-SLOW-NEXT:    vpsrlq $48, %xmm0, %xmm0
+; AVX2-SLOW-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; AVX2-SLOW-NEXT:    retq
+;
+; AVX2-FAST-LABEL: shuf_zext_16i8_to_2i64_offset6:
+; AVX2-FAST:       # %bb.0: # %entry
+; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6],zero,zero,zero,zero,zero,zero,zero,xmm0[7],zero,zero,zero,zero,zero,zero,zero
+; AVX2-FAST-NEXT:    retq
+;
+; AVX512F-LABEL: shuf_zext_16i8_to_2i64_offset6:
+; AVX512F:       # %bb.0: # %entry
+; AVX512F-NEXT:    vpsrlq $48, %xmm0, %xmm0
+; AVX512F-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; AVX512F-NEXT:    retq
+;
+; AVX512BW-LABEL: shuf_zext_16i8_to_2i64_offset6:
+; AVX512BW:       # %bb.0: # %entry
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6],zero,zero,zero,zero,zero,zero,zero,xmm0[7],zero,zero,zero,zero,zero,zero,zero
+; AVX512BW-NEXT:    retq
 entry:
   %B = shufflevector <16 x i8> %A, <16 x i8> zeroinitializer, <16 x i32> <i32 6, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 7, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16>
   %Z = bitcast <16 x i8> %B to <2 x i64>
@@ -1805,11 +1827,33 @@ define <2 x i64> @shuf_zext_8i16_to_2i64_offset6(<8 x i16> %A) nounwind uwtable
 ; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: shuf_zext_8i16_to_2i64_offset6:
-; AVX:       # %bb.0: # %entry
-; AVX-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
-; AVX-NEXT:    vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
-; AVX-NEXT:    retq
+; AVX1-LABEL: shuf_zext_8i16_to_2i64_offset6:
+; AVX1:       # %bb.0: # %entry
+; AVX1-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
+; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; AVX1-NEXT:    retq
+;
+; AVX2-SLOW-LABEL: shuf_zext_8i16_to_2i64_offset6:
+; AVX2-SLOW:       # %bb.0: # %entry
+; AVX2-SLOW-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
+; AVX2-SLOW-NEXT:    vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; AVX2-SLOW-NEXT:    retq
+;
+; AVX2-FAST-LABEL: shuf_zext_8i16_to_2i64_offset6:
+; AVX2-FAST:       # %bb.0: # %entry
+; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6,7],zero,zero,zero,zero,zero,zero,xmm0[8,9],zero,zero,zero,zero,zero,zero
+; AVX2-FAST-NEXT:    retq
+;
+; AVX512F-LABEL: shuf_zext_8i16_to_2i64_offset6:
+; AVX512F:       # %bb.0: # %entry
+; AVX512F-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
+; AVX512F-NEXT:    vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; AVX512F-NEXT:    retq
+;
+; AVX512BW-LABEL: shuf_zext_8i16_to_2i64_offset6:
+; AVX512BW:       # %bb.0: # %entry
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6,7],zero,zero,zero,zero,zero,zero,xmm0[8,9],zero,zero,zero,zero,zero,zero
+; AVX512BW-NEXT:    retq
 entry:
   %B = shufflevector <8 x i16> %A, <8 x i16> zeroinitializer, <8 x i32> <i32 3, i32 8, i32 8, i32 8, i32 4, i32 8, i32 8, i32 8>
   %Z = bitcast <8 x i16> %B to <2 x i64>
@@ -2580,8 +2624,7 @@ define <8 x i32> @splatshuf_zext_v8i32_matching_undefs(<8 x i16> %x) {
 ;
 ; AVX1-LABEL: splatshuf_zext_v8i32_matching_undefs:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,6,7,6,7,14,15,0,1,6,7,6,7,14,15]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1],zero,zero,xmm0[6,7],zero,zero,xmm0[6,7],zero,zero,xmm0[14,15],zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
@@ -2677,8 +2720,7 @@ define <16 x i16> @splatshuf_zext_v16i16(<16 x i8> %x) {
 ;
 ; AVX1-LABEL: splatshuf_zext_v16i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[14,14,14,14,14,14,14,14,14,14,14,14,14,14,14,14]
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[14],zero,xmm0[14],zero,xmm0[14],zero,xmm0[14],zero,xmm0[14],zero,xmm0[14],zero,xmm0[14],zero,xmm0[14],zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;