Index: lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- lib/Target/X86/X86ISelLowering.cpp
+++ lib/Target/X86/X86ISelLowering.cpp
@@ -23318,7 +23318,8 @@
                                        SelectionDAG &DAG) {
   MVT VT = Amt.getSimpleValueType();
   if (!(VT == MVT::v8i16 || VT == MVT::v4i32 ||
-        (Subtarget.hasInt256() && VT == MVT::v16i16)))
+        (Subtarget.hasInt256() && VT == MVT::v16i16) ||
+        (!Subtarget.hasAVX512() && VT == MVT::v16i8)))
     return SDValue();
 
   if (ISD::isBuildVectorOfConstantSDNodes(Amt.getNode())) {
Index: test/CodeGen/X86/vector-mul.ll
===================================================================
--- test/CodeGen/X86/vector-mul.ll
+++ test/CodeGen/X86/vector-mul.ll
@@ -177,41 +177,31 @@
 define <16 x i8> @mul_v16i8_1_2_4_8_1_2_4_8_1_2_4_8_1_2_4_8(<16 x i8> %a0) nounwind {
 ; X86-LABEL: mul_v16i8_1_2_4_8_1_2_4_8_1_2_4_8_1_2_4_8:
 ; X86:       # %bb.0:
-; X86-NEXT:    movdqa %xmm0, %xmm1
-; X86-NEXT:    movdqa %xmm0, %xmm2
-; X86-NEXT:    psllw $4, %xmm2
-; X86-NEXT:    pand {{\.LCPI.*}}, %xmm2
-; X86-NEXT:    movdqa {{.*#+}} xmm0 = [8192,24640,8192,24640,8192,24640,8192,24640]
-; X86-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
-; X86-NEXT:    movdqa %xmm1, %xmm2
-; X86-NEXT:    psllw $2, %xmm2
-; X86-NEXT:    pand {{\.LCPI.*}}, %xmm2
-; X86-NEXT:    paddb %xmm0, %xmm0
-; X86-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
-; X86-NEXT:    movdqa %xmm1, %xmm2
-; X86-NEXT:    paddb %xmm1, %xmm2
-; X86-NEXT:    paddb %xmm0, %xmm0
-; X86-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
+; X86-NEXT:    movdqa {{.*#+}} xmm2 = [1,2,4,8,1,2,4,8,1,2,4,8,1,2,4,8]
+; X86-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; X86-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; X86-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; X86-NEXT:    pmullw %xmm2, %xmm0
+; X86-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; X86-NEXT:    pand %xmm2, %xmm0
+; X86-NEXT:    pmullw {{\.LCPI.*}}, %xmm1
+; X86-NEXT:    pand %xmm2, %xmm1
+; X86-NEXT:    packuswb %xmm0, %xmm1
 ; X86-NEXT:    movdqa %xmm1, %xmm0
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: mul_v16i8_1_2_4_8_1_2_4_8_1_2_4_8_1_2_4_8:
 ; X64:       # %bb.0:
-; X64-NEXT:    movdqa %xmm0, %xmm1
-; X64-NEXT:    movdqa %xmm0, %xmm2
-; X64-NEXT:    psllw $4, %xmm2
-; X64-NEXT:    pand {{.*}}(%rip), %xmm2
-; X64-NEXT:    movdqa {{.*#+}} xmm0 = [8192,24640,8192,24640,8192,24640,8192,24640]
-; X64-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
-; X64-NEXT:    movdqa %xmm1, %xmm2
-; X64-NEXT:    psllw $2, %xmm2
-; X64-NEXT:    pand {{.*}}(%rip), %xmm2
-; X64-NEXT:    paddb %xmm0, %xmm0
-; X64-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
-; X64-NEXT:    movdqa %xmm1, %xmm2
-; X64-NEXT:    paddb %xmm1, %xmm2
-; X64-NEXT:    paddb %xmm0, %xmm0
-; X64-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
+; X64-NEXT:    movdqa {{.*#+}} xmm2 = [1,2,4,8,1,2,4,8,1,2,4,8,1,2,4,8]
+; X64-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; X64-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; X64-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; X64-NEXT:    pmullw %xmm2, %xmm0
+; X64-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; X64-NEXT:    pand %xmm2, %xmm0
+; X64-NEXT:    pmullw {{.*}}(%rip), %xmm1
+; X64-NEXT:    pand %xmm2, %xmm1
+; X64-NEXT:    packuswb %xmm0, %xmm1
 ; X64-NEXT:    movdqa %xmm1, %xmm0
 ; X64-NEXT:    retq
 ;
@@ -222,17 +212,14 @@
 ;
 ; X64-AVX2-LABEL: mul_v16i8_1_2_4_8_1_2_4_8_1_2_4_8_1_2_4_8:
 ; X64-AVX2:       # %bb.0:
-; X64-AVX2-NEXT:    vpsllw $4, %xmm0, %xmm1
-; X64-AVX2-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
-; X64-AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [8192,24640,8192,24640,8192,24640,8192,24640]
-; X64-AVX2-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
-; X64-AVX2-NEXT:    vpsllw $2, %xmm0, %xmm1
-; X64-AVX2-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
-; X64-AVX2-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
-; X64-AVX2-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
-; X64-AVX2-NEXT:    vpaddb %xmm0, %xmm0, %xmm1
-; X64-AVX2-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
-; X64-AVX2-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
+; X64-AVX2-NEXT:    vpmovsxbw %xmm0, %ymm0
+; X64-AVX2-NEXT:    vpmullw {{.*}}(%rip), %ymm0, %ymm0
+; X64-AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; X64-AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; X64-AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; X64-AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; X64-AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq
   %1 = mul <16 x i8> %a0, <i8 1, i8 2, i8 4, i8 8, i8 1, i8 2, i8 4, i8 8, i8 1, i8 2, i8 4, i8 8, i8 1, i8 2, i8 4, i8 8>
   ret <16 x i8> %1
Index: test/CodeGen/X86/vector-shift-shl-128.ll
===================================================================
--- test/CodeGen/X86/vector-shift-shl-128.ll
+++ test/CodeGen/X86/vector-shift-shl-128.ll
@@ -928,68 +928,61 @@
 define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) nounwind {
 ; SSE2-LABEL: constant_shift_v16i8:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [8192,24640,41088,57536,49376,32928,16480,32]
-; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    pxor %xmm3, %xmm3
-; SSE2-NEXT:    pcmpgtb %xmm2, %xmm3
-; SSE2-NEXT:    movdqa %xmm3, %xmm4
-; SSE2-NEXT:    pandn %xmm0, %xmm4
-; SSE2-NEXT:    psllw $4, %xmm0
-; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSE2-NEXT:    pand %xmm3, %xmm0
-; SSE2-NEXT:    por %xmm4, %xmm0
-; SSE2-NEXT:    paddb %xmm2, %xmm2
-; SSE2-NEXT:    pxor %xmm3, %xmm3
-; SSE2-NEXT:    pcmpgtb %xmm2, %xmm3
-; SSE2-NEXT:    movdqa %xmm3, %xmm4
-; SSE2-NEXT:    pandn %xmm0, %xmm4
-; SSE2-NEXT:    psllw $2, %xmm0
-; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSE2-NEXT:    pand %xmm3, %xmm0
-; SSE2-NEXT:    por %xmm4, %xmm0
-; SSE2-NEXT:    paddb %xmm2, %xmm2
-; SSE2-NEXT:    pcmpgtb %xmm2, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128,128,64,32,16,8,4,2,1]
 ; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pandn %xmm0, %xmm2
-; SSE2-NEXT:    paddb %xmm0, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    por %xmm2, %xmm0
+; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = 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]
+; SSE2-NEXT:    movdqa %xmm0, %xmm3
+; SSE2-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm0[8],xmm3[9],xmm0[9],xmm3[10],xmm0[10],xmm3[11],xmm0[11],xmm3[12],xmm0[12],xmm3[13],xmm0[13],xmm3[14],xmm0[14],xmm3[15],xmm0[15]
+; SSE2-NEXT:    pmullw %xmm2, %xmm3
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    pand %xmm2, %xmm3
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT:    pmullw %xmm1, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    packuswb %xmm3, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: constant_shift_v16i8:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    movdqa %xmm0, %xmm1
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    psllw $4, %xmm2
-; SSE41-NEXT:    pand {{.*}}(%rip), %xmm2
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [8192,24640,41088,57536,49376,32928,16480,32]
-; SSE41-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
-; SSE41-NEXT:    movdqa %xmm1, %xmm2
-; SSE41-NEXT:    psllw $2, %xmm2
-; SSE41-NEXT:    pand {{.*}}(%rip), %xmm2
-; SSE41-NEXT:    paddb %xmm0, %xmm0
-; SSE41-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
-; SSE41-NEXT:    movdqa %xmm1, %xmm2
-; SSE41-NEXT:    paddb %xmm1, %xmm2
-; SSE41-NEXT:    paddb %xmm0, %xmm0
-; SSE41-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [1,2,4,8,16,32,64,128,128,64,32,16,8,4,2,1]
+; SSE41-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; SSE41-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; SSE41-NEXT:    pmullw %xmm2, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE41-NEXT:    pand %xmm2, %xmm0
+; SSE41-NEXT:    pmullw {{.*}}(%rip), %xmm1
+; SSE41-NEXT:    pand %xmm2, %xmm1
+; SSE41-NEXT:    packuswb %xmm0, %xmm1
 ; SSE41-NEXT:    movdqa %xmm1, %xmm0
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: constant_shift_v16i8:
-; AVX:       # %bb.0:
-; AVX-NEXT:    vpsllw $4, %xmm0, %xmm1
-; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm2 = [8192,24640,41088,57536,49376,32928,16480,32]
-; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vpsllw $2, %xmm0, %xmm1
-; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
-; AVX-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
-; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vpaddb %xmm0, %xmm0, %xmm1
-; AVX-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
-; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    retq
+; AVX1-LABEL: constant_shift_v16i8:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128,128,64,32,16,8,4,2,1]
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    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 {{.*}}(%rip), %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: constant_shift_v16i8:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpmovsxbw %xmm0, %ymm0
+; AVX2-NEXT:    vpmullw {{.*}}(%rip), %ymm0, %ymm0
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
 ;
 ; XOP-LABEL: constant_shift_v16i8:
 ; XOP:       # %bb.0:
@@ -1032,32 +1025,19 @@
 ;
 ; X32-SSE-LABEL: constant_shift_v16i8:
 ; X32-SSE:       # %bb.0:
-; X32-SSE-NEXT:    movdqa {{.*#+}} xmm2 = [8192,24640,41088,57536,49376,32928,16480,32]
-; X32-SSE-NEXT:    pxor %xmm1, %xmm1
-; X32-SSE-NEXT:    pxor %xmm3, %xmm3
-; X32-SSE-NEXT:    pcmpgtb %xmm2, %xmm3
-; X32-SSE-NEXT:    movdqa %xmm3, %xmm4
-; X32-SSE-NEXT:    pandn %xmm0, %xmm4
-; X32-SSE-NEXT:    psllw $4, %xmm0
-; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
-; X32-SSE-NEXT:    pand %xmm3, %xmm0
-; X32-SSE-NEXT:    por %xmm4, %xmm0
-; X32-SSE-NEXT:    paddb %xmm2, %xmm2
-; X32-SSE-NEXT:    pxor %xmm3, %xmm3
-; X32-SSE-NEXT:    pcmpgtb %xmm2, %xmm3
-; X32-SSE-NEXT:    movdqa %xmm3, %xmm4
-; X32-SSE-NEXT:    pandn %xmm0, %xmm4
-; X32-SSE-NEXT:    psllw $2, %xmm0
-; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
-; X32-SSE-NEXT:    pand %xmm3, %xmm0
-; X32-SSE-NEXT:    por %xmm4, %xmm0
-; X32-SSE-NEXT:    paddb %xmm2, %xmm2
-; X32-SSE-NEXT:    pcmpgtb %xmm2, %xmm1
+; X32-SSE-NEXT:    movdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128,128,64,32,16,8,4,2,1]
 ; X32-SSE-NEXT:    movdqa %xmm1, %xmm2
-; X32-SSE-NEXT:    pandn %xmm0, %xmm2
-; X32-SSE-NEXT:    paddb %xmm0, %xmm0
-; X32-SSE-NEXT:    pand %xmm1, %xmm0
-; X32-SSE-NEXT:    por %xmm2, %xmm0
+; X32-SSE-NEXT:    punpckhbw {{.*#+}} xmm2 = 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]
+; X32-SSE-NEXT:    movdqa %xmm0, %xmm3
+; X32-SSE-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm0[8],xmm3[9],xmm0[9],xmm3[10],xmm0[10],xmm3[11],xmm0[11],xmm3[12],xmm0[12],xmm3[13],xmm0[13],xmm3[14],xmm0[14],xmm3[15],xmm0[15]
+; X32-SSE-NEXT:    pmullw %xmm2, %xmm3
+; X32-SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; X32-SSE-NEXT:    pand %xmm2, %xmm3
+; X32-SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
+; X32-SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; X32-SSE-NEXT:    pmullw %xmm1, %xmm0
+; X32-SSE-NEXT:    pand %xmm2, %xmm0
+; X32-SSE-NEXT:    packuswb %xmm3, %xmm0
 ; X32-SSE-NEXT:    retl
   %shift = shl <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
   ret <16 x i8> %shift
Index: test/CodeGen/X86/vector-shift-shl-256.ll
===================================================================
--- test/CodeGen/X86/vector-shift-shl-256.ll
+++ test/CodeGen/X86/vector-shift-shl-256.ll
@@ -989,29 +989,26 @@
 define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) nounwind {
 ; AVX1-LABEL: constant_shift_v32i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpsllw $4, %xmm1, %xmm2
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [240,240,240,240,240,240,240,240,240,240,240,240,240,240,240,240]
-; AVX1-NEXT:    vpand %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [8192,24640,41088,57536,49376,32928,16480,32]
-; AVX1-NEXT:    vpblendvb %xmm4, %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpsllw $2, %xmm1, %xmm2
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [252,252,252,252,252,252,252,252,252,252,252,252,252,252,252,252]
-; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
-; AVX1-NEXT:    vpaddb %xmm4, %xmm4, %xmm6
-; AVX1-NEXT:    vpblendvb %xmm6, %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpaddb %xmm1, %xmm1, %xmm2
-; AVX1-NEXT:    vpaddb %xmm6, %xmm6, %xmm7
-; AVX1-NEXT:    vpblendvb %xmm7, %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpsllw $4, %xmm0, %xmm2
-; AVX1-NEXT:    vpand %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vpblendvb %xmm4, %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $2, %xmm0, %xmm2
-; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
-; AVX1-NEXT:    vpblendvb %xmm6, %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
-; AVX1-NEXT:    vpblendvb %xmm7, %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128,128,64,32,16,8,4,2,1]
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm3
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; AVX1-NEXT:    vpand %xmm4, %xmm3, %xmm3
+; 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:    vmovdqa {{.*#+}} xmm5 = [1,2,4,8,16,32,64,128]
+; AVX1-NEXT:    vpmullw %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm1
+; AVX1-NEXT:    vpand %xmm4, %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 %xmm5, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: constant_shift_v32i8:
@@ -1093,29 +1090,26 @@
 ;
 ; X32-AVX1-LABEL: constant_shift_v32i8:
 ; X32-AVX1:       # %bb.0:
-; X32-AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; X32-AVX1-NEXT:    vpsllw $4, %xmm1, %xmm2
-; X32-AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [240,240,240,240,240,240,240,240,240,240,240,240,240,240,240,240]
-; X32-AVX1-NEXT:    vpand %xmm3, %xmm2, %xmm2
-; X32-AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [8192,24640,41088,57536,49376,32928,16480,32]
-; X32-AVX1-NEXT:    vpblendvb %xmm4, %xmm2, %xmm1, %xmm1
-; X32-AVX1-NEXT:    vpsllw $2, %xmm1, %xmm2
-; X32-AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [252,252,252,252,252,252,252,252,252,252,252,252,252,252,252,252]
-; X32-AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
-; X32-AVX1-NEXT:    vpaddb %xmm4, %xmm4, %xmm6
-; X32-AVX1-NEXT:    vpblendvb %xmm6, %xmm2, %xmm1, %xmm1
-; X32-AVX1-NEXT:    vpaddb %xmm1, %xmm1, %xmm2
-; X32-AVX1-NEXT:    vpaddb %xmm6, %xmm6, %xmm7
-; X32-AVX1-NEXT:    vpblendvb %xmm7, %xmm2, %xmm1, %xmm1
-; X32-AVX1-NEXT:    vpsllw $4, %xmm0, %xmm2
-; X32-AVX1-NEXT:    vpand %xmm3, %xmm2, %xmm2
-; X32-AVX1-NEXT:    vpblendvb %xmm4, %xmm2, %xmm0, %xmm0
-; X32-AVX1-NEXT:    vpsllw $2, %xmm0, %xmm2
-; X32-AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
-; X32-AVX1-NEXT:    vpblendvb %xmm6, %xmm2, %xmm0, %xmm0
-; X32-AVX1-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
-; X32-AVX1-NEXT:    vpblendvb %xmm7, %xmm2, %xmm0, %xmm0
-; X32-AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; X32-AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128,128,64,32,16,8,4,2,1]
+; X32-AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; X32-AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; X32-AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; X32-AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm3
+; X32-AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; X32-AVX1-NEXT:    vpand %xmm4, %xmm3, %xmm3
+; X32-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
+; X32-AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [1,2,4,8,16,32,64,128]
+; X32-AVX1-NEXT:    vpmullw %xmm5, %xmm2, %xmm2
+; X32-AVX1-NEXT:    vpand %xmm4, %xmm2, %xmm2
+; X32-AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; X32-AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; X32-AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm1
+; X32-AVX1-NEXT:    vpand %xmm4, %xmm1, %xmm1
+; X32-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
+; X32-AVX1-NEXT:    vpmullw %xmm5, %xmm0, %xmm0
+; X32-AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
+; X32-AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; X32-AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; X32-AVX1-NEXT:    retl
 ;
 ; X32-AVX2-LABEL: constant_shift_v32i8: