[X86][SSE] createVariablePermute - PSHUFB requires SSSE3 not just SSE3

llvm-svn: 327259

1 files changed, 455 insertions, 14 deletions

diff --git a/llvm/test/CodeGen/X86/var-permute-128.ll b/llvm/test/CodeGen/X86/var-permute-128.ll
index 462db996f0c..c81892e4b1d 100644
--- a/llvm/test/CodeGen/X86/var-permute-128.ll
+++ b/llvm/test/CodeGen/X86/var-permute-128.ll
@@ -1,4 +1,5 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse3 | FileCheck %s --check-prefixes=SSE,SSE3
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+ssse3 | FileCheck %s --check-prefixes=SSE,SSSE3
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=SSE,SSE41
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop | FileCheck %s --check-prefixes=AVX,AVXNOVLBW,XOP
@@ -11,6 +12,19 @@
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw,+avx512vl,+avx512vbmi | FileCheck %s --check-prefixes=AVX,AVX512VL,VLVBMI
 
 define <2 x i64> @var_shuffle_v2i64(<2 x i64> %v, <2 x i64> %indices) nounwind {
+; SSE3-LABEL: var_shuffle_v2i64:
+; SSE3:       # %bb.0:
+; SSE3-NEXT:    movq %xmm1, %rax
+; SSE3-NEXT:    andl $1, %eax
+; SSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
+; SSE3-NEXT:    movq %xmm1, %rcx
+; SSE3-NEXT:    andl $1, %ecx
+; SSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE3-NEXT:    movsd {{.*#+}} xmm1 = mem[0],zero
+; SSE3-NEXT:    movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE3-NEXT:    retq
+;
 ; SSSE3-LABEL: var_shuffle_v2i64:
 ; SSSE3:       # %bb.0:
 ; SSSE3-NEXT:    movq %xmm1, %rax
@@ -51,6 +65,29 @@ define <2 x i64> @var_shuffle_v2i64(<2 x i64> %v, <2 x i64> %indices) nounwind {
 }
 
 define <4 x i32> @var_shuffle_v4i32(<4 x i32> %v, <4 x i32> %indices) nounwind {
+; SSE3-LABEL: var_shuffle_v4i32:
+; SSE3:       # %bb.0:
+; SSE3-NEXT:    movd %xmm1, %eax
+; SSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,2,3]
+; SSE3-NEXT:    movd %xmm2, %ecx
+; SSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
+; SSE3-NEXT:    movd %xmm2, %edx
+; SSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[3,1,2,3]
+; SSE3-NEXT:    movd %xmm1, %esi
+; SSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    andl $3, %eax
+; SSE3-NEXT:    andl $3, %ecx
+; SSE3-NEXT:    andl $3, %edx
+; SSE3-NEXT:    andl $3, %esi
+; SSE3-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE3-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE3-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; SSE3-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE3-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSE3-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE3-NEXT:    movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE3-NEXT:    retq
+;
 ; SSSE3-LABEL: var_shuffle_v4i32:
 ; SSSE3:       # %bb.0:
 ; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [67372036,67372036,67372036,67372036]
@@ -91,12 +128,63 @@ define <4 x i32> @var_shuffle_v4i32(<4 x i32> %v, <4 x i32> %indices) nounwind {
 }
 
 define <8 x i16> @var_shuffle_v8i16(<8 x i16> %v, <8 x i16> %indices) nounwind {
-; SSE-LABEL: var_shuffle_v8i16:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pmullw {{.*}}(%rip), %xmm1
-; SSE-NEXT:    paddw {{.*}}(%rip), %xmm1
-; SSE-NEXT:    pshufb %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE3-LABEL: var_shuffle_v8i16:
+; SSE3:       # %bb.0:
+; SSE3-NEXT:    movd %xmm1, %r8d
+; SSE3-NEXT:    pextrw $1, %xmm1, %r9d
+; SSE3-NEXT:    pextrw $2, %xmm1, %r10d
+; SSE3-NEXT:    pextrw $3, %xmm1, %esi
+; SSE3-NEXT:    pextrw $4, %xmm1, %edi
+; SSE3-NEXT:    pextrw $5, %xmm1, %eax
+; SSE3-NEXT:    pextrw $6, %xmm1, %ecx
+; SSE3-NEXT:    pextrw $7, %xmm1, %edx
+; SSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    andl $7, %r8d
+; SSE3-NEXT:    andl $7, %r9d
+; SSE3-NEXT:    andl $7, %r10d
+; SSE3-NEXT:    andl $7, %esi
+; SSE3-NEXT:    andl $7, %edi
+; SSE3-NEXT:    andl $7, %eax
+; SSE3-NEXT:    andl $7, %ecx
+; SSE3-NEXT:    andl $7, %edx
+; SSE3-NEXT:    movzwl -24(%rsp,%rdx,2), %edx
+; SSE3-NEXT:    movd %edx, %xmm0
+; SSE3-NEXT:    movzwl -24(%rsp,%rcx,2), %ecx
+; SSE3-NEXT:    movd %ecx, %xmm1
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; SSE3-NEXT:    movzwl -24(%rsp,%rax,2), %eax
+; SSE3-NEXT:    movd %eax, %xmm0
+; SSE3-NEXT:    movzwl -24(%rsp,%rdi,2), %eax
+; SSE3-NEXT:    movd %eax, %xmm2
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
+; SSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; SSE3-NEXT:    movzwl -24(%rsp,%rsi,2), %eax
+; SSE3-NEXT:    movd %eax, %xmm0
+; SSE3-NEXT:    movzwl -24(%rsp,%r10,2), %eax
+; SSE3-NEXT:    movd %eax, %xmm1
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; SSE3-NEXT:    movzwl -24(%rsp,%r9,2), %eax
+; SSE3-NEXT:    movd %eax, %xmm3
+; SSE3-NEXT:    movzwl -24(%rsp,%r8,2), %eax
+; SSE3-NEXT:    movd %eax, %xmm0
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
+; SSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: var_shuffle_v8i16:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    pmullw {{.*}}(%rip), %xmm1
+; SSSE3-NEXT:    paddw {{.*}}(%rip), %xmm1
+; SSSE3-NEXT:    pshufb %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: var_shuffle_v8i16:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pmullw {{.*}}(%rip), %xmm1
+; SSE41-NEXT:    paddw {{.*}}(%rip), %xmm1
+; SSE41-NEXT:    pshufb %xmm1, %xmm0
+; SSE41-NEXT:    retq
 ;
 ; AVXNOVLBW-LABEL: var_shuffle_v8i16:
 ; AVXNOVLBW:       # %bb.0:
@@ -137,10 +225,100 @@ define <8 x i16> @var_shuffle_v8i16(<8 x i16> %v, <8 x i16> %indices) nounwind {
 }
 
 define <16 x i8> @var_shuffle_v16i8(<16 x i8> %v, <16 x i8> %indices) nounwind {
-; SSE-LABEL: var_shuffle_v16i8:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pshufb %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE3-LABEL: var_shuffle_v16i8:
+; SSE3:       # %bb.0:
+; SSE3-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm8
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm15
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm9
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm3
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm10
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm7
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm11
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm6
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm12
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm5
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm13
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm4
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm14
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm1
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm2
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm0
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm8[0],xmm15[1],xmm8[1],xmm15[2],xmm8[2],xmm15[3],xmm8[3],xmm15[4],xmm8[4],xmm15[5],xmm8[5],xmm15[6],xmm8[6],xmm15[7],xmm8[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3],xmm3[4],xmm9[4],xmm3[5],xmm9[5],xmm3[6],xmm9[6],xmm3[7],xmm9[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm15[0],xmm3[1],xmm15[1],xmm3[2],xmm15[2],xmm3[3],xmm15[3]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3],xmm7[4],xmm10[4],xmm7[5],xmm10[5],xmm7[6],xmm10[6],xmm7[7],xmm10[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1],xmm6[2],xmm7[2],xmm6[3],xmm7[3]
+; SSE3-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm12[0],xmm5[1],xmm12[1],xmm5[2],xmm12[2],xmm5[3],xmm12[3],xmm5[4],xmm12[4],xmm5[5],xmm12[5],xmm5[6],xmm12[6],xmm5[7],xmm12[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm13[0],xmm4[1],xmm13[1],xmm4[2],xmm13[2],xmm4[3],xmm13[3],xmm4[4],xmm13[4],xmm4[5],xmm13[5],xmm4[6],xmm13[6],xmm4[7],xmm13[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm14[0],xmm1[1],xmm14[1],xmm1[2],xmm14[2],xmm1[3],xmm14[3],xmm1[4],xmm14[4],xmm1[5],xmm14[5],xmm1[6],xmm14[6],xmm1[7],xmm14[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm6[0]
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: var_shuffle_v16i8:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    pshufb %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: var_shuffle_v16i8:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pshufb %xmm1, %xmm0
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: var_shuffle_v16i8:
 ; AVX:       # %bb.0:
@@ -198,6 +376,18 @@ define <16 x i8> @var_shuffle_v16i8(<16 x i8> %v, <16 x i8> %indices) nounwind {
 }
 
 define <2 x double> @var_shuffle_v2f64(<2 x double> %v, <2 x i64> %indices) nounwind {
+; SSE3-LABEL: var_shuffle_v2f64:
+; SSE3:       # %bb.0:
+; SSE3-NEXT:    movq %xmm1, %rax
+; SSE3-NEXT:    andl $1, %eax
+; SSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
+; SSE3-NEXT:    movq %xmm1, %rcx
+; SSE3-NEXT:    andl $1, %ecx
+; SSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE3-NEXT:    movhpd {{.*#+}} xmm0 = xmm0[0],mem[0]
+; SSE3-NEXT:    retq
+;
 ; SSSE3-LABEL: var_shuffle_v2f64:
 ; SSSE3:       # %bb.0:
 ; SSSE3-NEXT:    movq %xmm1, %rax
@@ -236,6 +426,29 @@ define <2 x double> @var_shuffle_v2f64(<2 x double> %v, <2 x i64> %indices) noun
 }
 
 define <4 x float> @var_shuffle_v4f32(<4 x float> %v, <4 x i32> %indices) nounwind {
+; SSE3-LABEL: var_shuffle_v4f32:
+; SSE3:       # %bb.0:
+; SSE3-NEXT:    movd %xmm1, %eax
+; SSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,2,3]
+; SSE3-NEXT:    movd %xmm2, %ecx
+; SSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
+; SSE3-NEXT:    movd %xmm2, %edx
+; SSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[3,1,2,3]
+; SSE3-NEXT:    movd %xmm1, %esi
+; SSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    andl $3, %eax
+; SSE3-NEXT:    andl $3, %ecx
+; SSE3-NEXT:    andl $3, %edx
+; SSE3-NEXT:    andl $3, %esi
+; SSE3-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE3-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE3-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; SSE3-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE3-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSE3-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE3-NEXT:    movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE3-NEXT:    retq
+;
 ; SSSE3-LABEL: var_shuffle_v4f32:
 ; SSSE3:       # %bb.0:
 ; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [67372036,67372036,67372036,67372036]
@@ -276,10 +489,100 @@ define <4 x float> @var_shuffle_v4f32(<4 x float> %v, <4 x i32> %indices) nounwi
 }
 
 define <16 x i8> @var_shuffle_v16i8_from_v16i8_v32i8(<16 x i8> %v, <32 x i8> %indices) nounwind {
-; SSE-LABEL: var_shuffle_v16i8_from_v16i8_v32i8:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pshufb %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE3-LABEL: var_shuffle_v16i8_from_v16i8_v32i8:
+; SSE3:       # %bb.0:
+; SSE3-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm8
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm15
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm9
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm3
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm10
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm7
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm11
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm6
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm12
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm5
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm13
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm4
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm14
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm1
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm2
+; SSE3-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    andl $15, %eax
+; SSE3-NEXT:    movzbl -24(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm0
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm8[0],xmm15[1],xmm8[1],xmm15[2],xmm8[2],xmm15[3],xmm8[3],xmm15[4],xmm8[4],xmm15[5],xmm8[5],xmm15[6],xmm8[6],xmm15[7],xmm8[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3],xmm3[4],xmm9[4],xmm3[5],xmm9[5],xmm3[6],xmm9[6],xmm3[7],xmm9[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm15[0],xmm3[1],xmm15[1],xmm3[2],xmm15[2],xmm3[3],xmm15[3]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3],xmm7[4],xmm10[4],xmm7[5],xmm10[5],xmm7[6],xmm10[6],xmm7[7],xmm10[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1],xmm6[2],xmm7[2],xmm6[3],xmm7[3]
+; SSE3-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm12[0],xmm5[1],xmm12[1],xmm5[2],xmm12[2],xmm5[3],xmm12[3],xmm5[4],xmm12[4],xmm5[5],xmm12[5],xmm5[6],xmm12[6],xmm5[7],xmm12[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm13[0],xmm4[1],xmm13[1],xmm4[2],xmm13[2],xmm4[3],xmm13[3],xmm4[4],xmm13[4],xmm4[5],xmm13[5],xmm4[6],xmm13[6],xmm4[7],xmm13[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm14[0],xmm1[1],xmm14[1],xmm1[2],xmm14[2],xmm1[3],xmm14[3],xmm1[4],xmm14[4],xmm1[5],xmm14[5],xmm1[6],xmm14[6],xmm1[7],xmm14[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm6[0]
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: var_shuffle_v16i8_from_v16i8_v32i8:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    pshufb %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: var_shuffle_v16i8_from_v16i8_v32i8:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pshufb %xmm1, %xmm0
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: var_shuffle_v16i8_from_v16i8_v32i8:
 ; AVX:       # %bb.0:
@@ -338,6 +641,144 @@ define <16 x i8> @var_shuffle_v16i8_from_v16i8_v32i8(<16 x i8> %v, <32 x i8> %in
 }
 
 define <16 x i8> @var_shuffle_v16i8_from_v32i8_v16i8(<32 x i8> %v, <16 x i8> %indices) nounwind {
+; SSE3-LABEL: var_shuffle_v16i8_from_v32i8_v16i8:
+; SSE3:       # %bb.0:
+; SSE3-NEXT:    pushq %rbp
+; SSE3-NEXT:    movq %rsp, %rbp
+; SSE3-NEXT:    pushq %r15
+; SSE3-NEXT:    pushq %r14
+; SSE3-NEXT:    pushq %r13
+; SSE3-NEXT:    pushq %r12
+; SSE3-NEXT:    pushq %rbx
+; SSE3-NEXT:    andq $-32, %rsp
+; SSE3-NEXT:    subq $608, %rsp # imm = 0x260
+; SSE3-NEXT:    movaps %xmm2, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    movq %rax, {{[0-9]+}}(%rsp) # 8-byte Spill
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    movq %rax, {{[0-9]+}}(%rsp) # 8-byte Spill
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    movq %rax, {{[0-9]+}}(%rsp) # 8-byte Spill
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r11d
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r14d
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r15d
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r12d
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r13d
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r10d
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r8d
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %edi
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %esi
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %edx
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
+; SSE3-NEXT:    movaps %xmm1, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movaps %xmm0, {{[0-9]+}}(%rsp)
+; SSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r9d
+; SSE3-NEXT:    andl $31, %r9d
+; SSE3-NEXT:    movzbl 64(%rsp,%r9), %ebx
+; SSE3-NEXT:    movd %ebx, %xmm8
+; SSE3-NEXT:    andl $31, %eax
+; SSE3-NEXT:    movzbl 96(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm15
+; SSE3-NEXT:    andl $31, %edx
+; SSE3-NEXT:    movzbl 128(%rsp,%rdx), %eax
+; SSE3-NEXT:    movd %eax, %xmm9
+; SSE3-NEXT:    andl $31, %ecx
+; SSE3-NEXT:    movzbl 160(%rsp,%rcx), %eax
+; SSE3-NEXT:    movd %eax, %xmm3
+; SSE3-NEXT:    andl $31, %esi
+; SSE3-NEXT:    movzbl 192(%rsp,%rsi), %eax
+; SSE3-NEXT:    movd %eax, %xmm10
+; SSE3-NEXT:    andl $31, %edi
+; SSE3-NEXT:    movzbl 224(%rsp,%rdi), %eax
+; SSE3-NEXT:    movd %eax, %xmm7
+; SSE3-NEXT:    andl $31, %r8d
+; SSE3-NEXT:    movzbl 256(%rsp,%r8), %eax
+; SSE3-NEXT:    movd %eax, %xmm11
+; SSE3-NEXT:    andl $31, %r10d
+; SSE3-NEXT:    movzbl 288(%rsp,%r10), %eax
+; SSE3-NEXT:    movd %eax, %xmm6
+; SSE3-NEXT:    andl $31, %r13d
+; SSE3-NEXT:    movzbl 320(%rsp,%r13), %eax
+; SSE3-NEXT:    movd %eax, %xmm12
+; SSE3-NEXT:    andl $31, %r12d
+; SSE3-NEXT:    movzbl 352(%rsp,%r12), %eax
+; SSE3-NEXT:    movd %eax, %xmm5
+; SSE3-NEXT:    andl $31, %r15d
+; SSE3-NEXT:    movzbl 384(%rsp,%r15), %eax
+; SSE3-NEXT:    movd %eax, %xmm13
+; SSE3-NEXT:    andl $31, %r14d
+; SSE3-NEXT:    movzbl 416(%rsp,%r14), %eax
+; SSE3-NEXT:    movd %eax, %xmm4
+; SSE3-NEXT:    andl $31, %r11d
+; SSE3-NEXT:    movzbl 448(%rsp,%r11), %eax
+; SSE3-NEXT:    movd %eax, %xmm14
+; SSE3-NEXT:    movq {{[0-9]+}}(%rsp), %rax # 8-byte Reload
+; SSE3-NEXT:    andl $31, %eax
+; SSE3-NEXT:    movzbl 480(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm1
+; SSE3-NEXT:    movq {{[0-9]+}}(%rsp), %rax # 8-byte Reload
+; SSE3-NEXT:    andl $31, %eax
+; SSE3-NEXT:    movzbl 512(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm2
+; SSE3-NEXT:    movq {{[0-9]+}}(%rsp), %rax # 8-byte Reload
+; SSE3-NEXT:    andl $31, %eax
+; SSE3-NEXT:    movzbl 544(%rsp,%rax), %eax
+; SSE3-NEXT:    movd %eax, %xmm0
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm8[0],xmm15[1],xmm8[1],xmm15[2],xmm8[2],xmm15[3],xmm8[3],xmm15[4],xmm8[4],xmm15[5],xmm8[5],xmm15[6],xmm8[6],xmm15[7],xmm8[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3],xmm3[4],xmm9[4],xmm3[5],xmm9[5],xmm3[6],xmm9[6],xmm3[7],xmm9[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm15[0],xmm3[1],xmm15[1],xmm3[2],xmm15[2],xmm3[3],xmm15[3]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3],xmm7[4],xmm10[4],xmm7[5],xmm10[5],xmm7[6],xmm10[6],xmm7[7],xmm10[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1],xmm6[2],xmm7[2],xmm6[3],xmm7[3]
+; SSE3-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm12[0],xmm5[1],xmm12[1],xmm5[2],xmm12[2],xmm5[3],xmm12[3],xmm5[4],xmm12[4],xmm5[5],xmm12[5],xmm5[6],xmm12[6],xmm5[7],xmm12[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm13[0],xmm4[1],xmm13[1],xmm4[2],xmm13[2],xmm4[3],xmm13[3],xmm4[4],xmm13[4],xmm4[5],xmm13[5],xmm4[6],xmm13[6],xmm4[7],xmm13[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm14[0],xmm1[1],xmm14[1],xmm1[2],xmm14[2],xmm1[3],xmm14[3],xmm1[4],xmm14[4],xmm1[5],xmm14[5],xmm1[6],xmm14[6],xmm1[7],xmm14[7]
+; SSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; SSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm6[0]
+; SSE3-NEXT:    leaq -40(%rbp), %rsp
+; SSE3-NEXT:    popq %rbx
+; SSE3-NEXT:    popq %r12
+; SSE3-NEXT:    popq %r13
+; SSE3-NEXT:    popq %r14
+; SSE3-NEXT:    popq %r15
+; SSE3-NEXT:    popq %rbp
+; SSE3-NEXT:    retq
+;
 ; SSSE3-LABEL: var_shuffle_v16i8_from_v32i8_v16i8:
 ; SSSE3:       # %bb.0:
 ; SSSE3-NEXT:    pushq %rbp