; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=SSE --check-prefix=SSE2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.2 | FileCheck %s --check-prefix=SSE --check-prefix=SSE42
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=AVX --check-prefix=AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512F
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512VL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BW
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw,+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BWVL

define void @shuffle_v16i8_to_v8i8_1(<16 x i8>* %L, <8 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v8i8_1:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,3,4,5,6,7]
; SSE2-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,5,6,7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[1,0,3,2,4,5,6,7]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,0,3,2,4,5,6,7]
; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movq %xmm0, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v8i8_1:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    movq %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v8i8_1:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vmovq %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v8i8_1:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vmovq %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v8i8_1:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; AVX512VL-NEXT:    vmovq %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v8i8_1:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vmovq %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v8i8_1:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrlw $8, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovwb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
  store <8 x i8> %strided.vec, <8 x i8>* %S
  ret void
}

define void @shuffle_v8i16_to_v4i16_1(<8 x i16>* %L, <4 x i16>* %S) nounwind {
; SSE2-LABEL: shuffle_v8i16_to_v4i16_1:
; SSE2:       # BB#0:
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = mem[3,1,2,3,4,5,6,7]
; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,0,3,2,4,5,6,7]
; SSE2-NEXT:    movq %xmm0, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v8i16_to_v4i16_1:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[2,3,6,7,10,11,14,15,14,15,10,11,12,13,14,15]
; SSE42-NEXT:    movq %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v8i16_to_v4i16_1:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,3,6,7,10,11,14,15,14,15,10,11,12,13,14,15]
; AVX-NEXT:    vmovq %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v8i16_to_v4i16_1:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,3,6,7,10,11,14,15,14,15,10,11,12,13,14,15]
; AVX512F-NEXT:    vmovq %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v8i16_to_v4i16_1:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrld $16, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovdw %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v8i16_to_v4i16_1:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,3,6,7,10,11,14,15,14,15,10,11,12,13,14,15]
; AVX512BW-NEXT:    vmovq %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v8i16_to_v4i16_1:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrld $16, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovdw %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <8 x i16>, <8 x i16>* %L
  %strided.vec = shufflevector <8 x i16> %vec, <8 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  store <4 x i16> %strided.vec, <4 x i16>* %S
  ret void
}

define void @shuffle_v4i32_to_v2i32_1(<4 x i32>* %L, <2 x i32>* %S) nounwind {
; SSE-LABEL: shuffle_v4i32_to_v2i32_1:
; SSE:       # BB#0:
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = mem[1,3,2,3]
; SSE-NEXT:    movq %xmm0, (%rsi)
; SSE-NEXT:    retq
;
; AVX-LABEL: shuffle_v4i32_to_v2i32_1:
; AVX:       # BB#0:
; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = mem[1,3,2,3]
; AVX-NEXT:    vmovlps %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v4i32_to_v2i32_1:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpermilps {{.*#+}} xmm0 = mem[1,3,2,3]
; AVX512F-NEXT:    vmovlps %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v4i32_to_v2i32_1:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = mem[1,1,3,3]
; AVX512VL-NEXT:    vpmovqd %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v4i32_to_v2i32_1:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vpermilps {{.*#+}} xmm0 = mem[1,3,2,3]
; AVX512BW-NEXT:    vmovlps %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v4i32_to_v2i32_1:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm0 = mem[1,1,3,3]
; AVX512BWVL-NEXT:    vpmovqd %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <4 x i32>, <4 x i32>* %L
  %strided.vec = shufflevector <4 x i32> %vec, <4 x i32> undef, <2 x i32> <i32 1, i32 3>
  store <2 x i32> %strided.vec, <2 x i32>* %S
  ret void
}

define void @shuffle_v16i8_to_v4i8_1(<16 x i8>* %L, <4 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v4i8_1:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[1,3,2,3,4,5,6,7]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,3,2,3,4,5,6,7]
; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v4i8_1:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    movd %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v4i8_1:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vmovd %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v4i8_1:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vmovd %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v4i8_1:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512VL-NEXT:    vpsrlw $8, %xmm0, %xmm0
; AVX512VL-NEXT:    vpmovdb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v4i8_1:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vmovd %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v4i8_1:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrlw $8, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovdb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <4 x i32> <i32 1, i32 5, i32 9, i32 13>
  store <4 x i8> %strided.vec, <4 x i8>* %S
  ret void
}

define void @shuffle_v16i8_to_v4i8_2(<16 x i8>* %L, <4 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v4i8_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,0,3,2,4,5,6,7]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v4i8_2:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    movd %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v4i8_2:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vmovd %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v4i8_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vmovd %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v4i8_2:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrld $16, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovdb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v4i8_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vmovd %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v4i8_2:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrld $16, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovdb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <4 x i32> <i32 2, i32 6, i32 10, i32 14>
  store <4 x i8> %strided.vec, <4 x i8>* %S
  ret void
}

define void @shuffle_v16i8_to_v4i8_3(<16 x i8>* %L, <4 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v4i8_3:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[3,1,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,3,4,5,6,7]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v4i8_3:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    movd %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v4i8_3:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vmovd %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v4i8_3:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vmovd %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v4i8_3:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrld $24, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovdb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v4i8_3:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vmovd %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v4i8_3:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrld $24, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovdb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <4 x i32> <i32 3, i32 7, i32 11, i32 15>
  store <4 x i8> %strided.vec, <4 x i8>* %S
  ret void
}

define void @shuffle_v8i16_to_v2i16_1(<8 x i16>* %L, <2 x i16>* %S) nounwind {
; SSE-LABEL: shuffle_v8i16_to_v2i16_1:
; SSE:       # BB#0:
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = mem[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,3,2,3,4,5,6,7]
; SSE-NEXT:    movd %xmm0, (%rsi)
; SSE-NEXT:    retq
;
; AVX-LABEL: shuffle_v8i16_to_v2i16_1:
; AVX:       # BB#0:
; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
; AVX-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[1,3,2,3,4,5,6,7]
; AVX-NEXT:    vmovd %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v8i16_to_v2i16_1:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[1,3,2,3,4,5,6,7]
; AVX512F-NEXT:    vmovd %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v8i16_to_v2i16_1:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrld $16, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovqw %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v8i16_to_v2i16_1:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
; AVX512BW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[1,3,2,3,4,5,6,7]
; AVX512BW-NEXT:    vmovd %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v8i16_to_v2i16_1:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrld $16, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovqw %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <8 x i16>, <8 x i16>* %L
  %strided.vec = shufflevector <8 x i16> %vec, <8 x i16> undef, <2 x i32> <i32 1, i32 5>
  store <2 x i16> %strided.vec, <2 x i16>* %S
  ret void
}

define void @shuffle_v8i16_to_v2i16_2(<8 x i16>* %L, <2 x i16>* %S) nounwind {
; SSE-LABEL: shuffle_v8i16_to_v2i16_2:
; SSE:       # BB#0:
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = mem[3,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,0,2,3,4,5,6,7]
; SSE-NEXT:    movd %xmm0, (%rsi)
; SSE-NEXT:    retq
;
; AVX-LABEL: shuffle_v8i16_to_v2i16_2:
; AVX:       # BB#0:
; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = mem[3,1,2,3]
; AVX-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[2,0,2,3,4,5,6,7]
; AVX-NEXT:    vmovd %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v8i16_to_v2i16_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = mem[3,1,2,3]
; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[2,0,2,3,4,5,6,7]
; AVX512F-NEXT:    vmovd %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v8i16_to_v2i16_2:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = mem[1,1,3,3]
; AVX512VL-NEXT:    vpmovqw %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v8i16_to_v2i16_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = mem[3,1,2,3]
; AVX512BW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[2,0,2,3,4,5,6,7]
; AVX512BW-NEXT:    vmovd %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v8i16_to_v2i16_2:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm0 = mem[1,1,3,3]
; AVX512BWVL-NEXT:    vpmovqw %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <8 x i16>, <8 x i16>* %L
  %strided.vec = shufflevector <8 x i16> %vec, <8 x i16> undef, <2 x i32> <i32 2, i32 6>
  store <2 x i16> %strided.vec, <2 x i16>* %S
  ret void
}

define void @shuffle_v8i16_to_v2i16_3(<8 x i16>* %L, <2 x i16>* %S) nounwind {
; SSE-LABEL: shuffle_v8i16_to_v2i16_3:
; SSE:       # BB#0:
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = mem[3,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; SSE-NEXT:    movd %xmm0, (%rsi)
; SSE-NEXT:    retq
;
; AVX-LABEL: shuffle_v8i16_to_v2i16_3:
; AVX:       # BB#0:
; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = mem[3,1,2,3]
; AVX-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; AVX-NEXT:    vmovd %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v8i16_to_v2i16_3:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = mem[3,1,2,3]
; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; AVX512F-NEXT:    vmovd %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v8i16_to_v2i16_3:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrlq $48, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovqw %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v8i16_to_v2i16_3:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = mem[3,1,2,3]
; AVX512BW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; AVX512BW-NEXT:    vmovd %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v8i16_to_v2i16_3:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrlq $48, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovqw %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <8 x i16>, <8 x i16>* %L
  %strided.vec = shufflevector <8 x i16> %vec, <8 x i16> undef, <2 x i32> <i32 3, i32 7>
  store <2 x i16> %strided.vec, <2 x i16>* %S
  ret void
}

define void @shuffle_v16i8_to_v2i8_1(<16 x i8>* %L, <2 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v2i8_1:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, %eax
; SSE2-NEXT:    movw %ax, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v2i8_1:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[1,9,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    pextrw $0, %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v2i8_1:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,9,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v2i8_1:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,9,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v2i8_1:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512VL-NEXT:    vpsrlw $8, %xmm0, %xmm0
; AVX512VL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v2i8_1:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,9,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v2i8_1:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrlw $8, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <2 x i32> <i32 1, i32 9>
  store <2 x i8> %strided.vec, <2 x i8>* %S
  ret void
}

define void @shuffle_v16i8_to_v2i8_2(<16 x i8>* %L, <2 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v2i8_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,3,2,3,4,5,6,7]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, %eax
; SSE2-NEXT:    movw %ax, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v2i8_2:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[2,10,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    pextrw $0, %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v2i8_2:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,10,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v2i8_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,10,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v2i8_2:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrld $16, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v2i8_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,10,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v2i8_2:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrld $16, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <2 x i32> <i32 2, i32 10>
  store <2 x i8> %strided.vec, <2 x i8>* %S
  ret void
}

define void @shuffle_v16i8_to_v2i8_3(<16 x i8>* %L, <2 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v2i8_3:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, %eax
; SSE2-NEXT:    movw %ax, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v2i8_3:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[3,11,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    pextrw $0, %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v2i8_3:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[3,11,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v2i8_3:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[3,11,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v2i8_3:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrld $24, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v2i8_3:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[3,11,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v2i8_3:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrld $24, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <2 x i32> <i32 3, i32 11>
  store <2 x i8> %strided.vec, <2 x i8>* %S
  ret void
}

define void @shuffle_v16i8_to_v2i8_4(<16 x i8>* %L, <2 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v2i8_4:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,0,2,3,4,5,6,7]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, %eax
; SSE2-NEXT:    movw %ax, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v2i8_4:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[4,12,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    pextrw $0, %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v2i8_4:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[4,12,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v2i8_4:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[4,12,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v2i8_4:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = mem[1,1,3,3]
; AVX512VL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v2i8_4:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[4,12,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v2i8_4:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm0 = mem[1,1,3,3]
; AVX512BWVL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <2 x i32> <i32 4, i32 12>
  store <2 x i8> %strided.vec, <2 x i8>* %S
  ret void
}

define void @shuffle_v16i8_to_v2i8_5(<16 x i8>* %L, <2 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v2i8_5:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, %eax
; SSE2-NEXT:    movw %ax, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v2i8_5:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[5,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    pextrw $0, %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v2i8_5:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v2i8_5:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v2i8_5:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrlq $40, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v2i8_5:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v2i8_5:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrlq $40, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <2 x i32> <i32 5, i32 13>
  store <2 x i8> %strided.vec, <2 x i8>* %S
  ret void
}

define void @shuffle_v16i8_to_v2i8_6(<16 x i8>* %L, <2 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v2i8_6:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, %eax
; SSE2-NEXT:    movw %ax, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v2i8_6:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[6,14,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    pextrw $0, %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v2i8_6:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6,14,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v2i8_6:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6,14,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v2i8_6:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrlq $48, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v2i8_6:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6,14,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v2i8_6:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrlq $48, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <2 x i32> <i32 6, i32 14>
  store <2 x i8> %strided.vec, <2 x i8>* %S
  ret void
}

define void @shuffle_v16i8_to_v2i8_7(<16 x i8>* %L, <2 x i8>* %S) nounwind {
; SSE2-LABEL: shuffle_v16i8_to_v2i8_7:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
; SSE2-NEXT:    packuswb %xmm0, %xmm0
; SSE2-NEXT:    movd %xmm0, %eax
; SSE2-NEXT:    movw %ax, (%rsi)
; SSE2-NEXT:    retq
;
; SSE42-LABEL: shuffle_v16i8_to_v2i8_7:
; SSE42:       # BB#0:
; SSE42-NEXT:    movdqa (%rdi), %xmm0
; SSE42-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[7,15,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; SSE42-NEXT:    pextrw $0, %xmm0, (%rsi)
; SSE42-NEXT:    retq
;
; AVX-LABEL: shuffle_v16i8_to_v2i8_7:
; AVX:       # BB#0:
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[7,15,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX-NEXT:    retq
;
; AVX512F-LABEL: shuffle_v16i8_to_v2i8_7:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[7,15,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: shuffle_v16i8_to_v2i8_7:
; AVX512VL:       # BB#0:
; AVX512VL-NEXT:    vpsrlq $56, (%rdi), %xmm0
; AVX512VL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: shuffle_v16i8_to_v2i8_7:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[7,15,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rsi)
; AVX512BW-NEXT:    retq
;
; AVX512BWVL-LABEL: shuffle_v16i8_to_v2i8_7:
; AVX512BWVL:       # BB#0:
; AVX512BWVL-NEXT:    vpsrlq $56, (%rdi), %xmm0
; AVX512BWVL-NEXT:    vpmovqb %xmm0, (%rsi)
; AVX512BWVL-NEXT:    retq
  %vec = load <16 x i8>, <16 x i8>* %L
  %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <2 x i32> <i32 7, i32 15>
  store <2 x i8> %strided.vec, <2 x i8>* %S
  ret void
}