Run DCE after a LoopFlatten test to reduce spurious output [nfc]

[llvm-project.git] / llvm / test / CodeGen / X86 / vector-interleaved-load-i8-stride-6.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefixes=SSE,FALLBACK0
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx  | FileCheck %s --check-prefixes=AVX,AVX1,AVX1-ONLY,FALLBACK1
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12

; These patterns are produced by LoopVectorizer for interleaved loads.

define void @load_i8_stride6_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4, ptr %out.vec5) nounwind {
; SSE-LABEL: load_i8_stride6_vf2:
; SSE:       # %bb.0:
; SSE-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT:    movdqa (%rdi), %xmm1
; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm1, %xmm3
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm3[0,3,2,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    pxor %xmm4, %xmm4
; SSE-NEXT:    movdqa %xmm1, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[1,3,2,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm5, %xmm5
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm3[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm6[1,2,2,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm4[8],xmm1[9],xmm4[9],xmm1[10],xmm4[10],xmm1[11],xmm4[11],xmm1[12],xmm4[12],xmm1[13],xmm4[13],xmm1[14],xmm4[14],xmm1[15],xmm4[15]
; SSE-NEXT:    movdqa %xmm0, %xmm4
; SSE-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[3,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,3,2,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm4, %xmm4
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[2,1,2,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm3, %xmm3
; SSE-NEXT:    psrlq $48, %xmm1
; SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movd %xmm2, %edi
; SSE-NEXT:    movw %di, (%rsi)
; SSE-NEXT:    movd %xmm5, %esi
; SSE-NEXT:    movw %si, (%rdx)
; SSE-NEXT:    movd %xmm6, %edx
; SSE-NEXT:    movw %dx, (%rcx)
; SSE-NEXT:    movd %xmm4, %ecx
; SSE-NEXT:    movw %cx, (%r8)
; SSE-NEXT:    movd %xmm3, %ecx
; SSE-NEXT:    movw %cx, (%r9)
; SSE-NEXT:    movd %xmm0, %ecx
; SSE-NEXT:    movw %cx, (%rax)
; SSE-NEXT:    retq
;
; AVX-LABEL: load_i8_stride6_vf2:
; AVX:       # %bb.0:
; AVX-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[0,6,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[1,7,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm3 = xmm0[2,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm4 = xmm0[3,9,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm5 = xmm0[4,10,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,11,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpextrw $0, %xmm1, (%rsi)
; AVX-NEXT:    vpextrw $0, %xmm2, (%rdx)
; AVX-NEXT:    vpextrw $0, %xmm3, (%rcx)
; AVX-NEXT:    vpextrw $0, %xmm4, (%r8)
; AVX-NEXT:    vpextrw $0, %xmm5, (%r9)
; AVX-NEXT:    vpextrw $0, %xmm0, (%rax)
; AVX-NEXT:    retq
  %wide.vec = load <12 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <12 x i8> %wide.vec, <12 x i8> poison, <2 x i32> <i32 0, i32 6>
  %strided.vec1 = shufflevector <12 x i8> %wide.vec, <12 x i8> poison, <2 x i32> <i32 1, i32 7>
  %strided.vec2 = shufflevector <12 x i8> %wide.vec, <12 x i8> poison, <2 x i32> <i32 2, i32 8>
  %strided.vec3 = shufflevector <12 x i8> %wide.vec, <12 x i8> poison, <2 x i32> <i32 3, i32 9>
  %strided.vec4 = shufflevector <12 x i8> %wide.vec, <12 x i8> poison, <2 x i32> <i32 4, i32 10>
  %strided.vec5 = shufflevector <12 x i8> %wide.vec, <12 x i8> poison, <2 x i32> <i32 5, i32 11>
  store <2 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <2 x i8> %strided.vec1, ptr %out.vec1, align 64
  store <2 x i8> %strided.vec2, ptr %out.vec2, align 64
  store <2 x i8> %strided.vec3, ptr %out.vec3, align 64
  store <2 x i8> %strided.vec4, ptr %out.vec4, align 64
  store <2 x i8> %strided.vec5, ptr %out.vec5, align 64
  ret void
}

define void @load_i8_stride6_vf4(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4, ptr %out.vec5) nounwind {
; SSE-LABEL: load_i8_stride6_vf4:
; SSE:       # %bb.0:
; SSE-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT:    movdqa (%rdi), %xmm5
; SSE-NEXT:    movdqa 16(%rdi), %xmm1
; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [65535,0,65535,65535,65535,65535,65535,65535]
; SSE-NEXT:    movdqa %xmm5, %xmm2
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    por %xmm2, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm2, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,1,4,5,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pxor %xmm3, %xmm3
; SSE-NEXT:    movdqa %xmm5, %xmm7
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,0,255,255,255,255,255,255,255,255,255,255,255,255]
; SSE-NEXT:    pandn %xmm1, %xmm4
; SSE-NEXT:    movdqa %xmm1, %xmm6
; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[1,0],xmm5[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[2,0],xmm5[2,3]
; SSE-NEXT:    movdqa {{.*#+}} xmm8 = [65535,65535,0,65535,65535,0,65535,65535]
; SSE-NEXT:    pand %xmm8, %xmm1
; SSE-NEXT:    pandn %xmm5, %xmm8
; SSE-NEXT:    punpckhbw {{.*#+}} xmm5 = xmm5[8],xmm3[8],xmm5[9],xmm3[9],xmm5[10],xmm3[10],xmm5[11],xmm3[11],xmm5[12],xmm3[12],xmm5[13],xmm3[13],xmm5[14],xmm3[14],xmm5[15],xmm3[15]
; SSE-NEXT:    movdqa %xmm5, %xmm9
; SSE-NEXT:    psrld $16, %xmm9
; SSE-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm3[0],xmm7[1],xmm3[1],xmm7[2],xmm3[2],xmm7[3],xmm3[3],xmm7[4],xmm3[4],xmm7[5],xmm3[5],xmm7[6],xmm3[6],xmm7[7],xmm3[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm7 = xmm7[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm7 = xmm7[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm7 = xmm7[2],xmm9[2],xmm7[3],xmm9[3]
; SSE-NEXT:    packuswb %xmm7, %xmm7
; SSE-NEXT:    por %xmm7, %xmm4
; SSE-NEXT:    movaps %xmm6, %xmm7
; SSE-NEXT:    andps %xmm2, %xmm7
; SSE-NEXT:    pshuflw {{.*#+}} xmm7 = xmm7[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm7 = xmm7[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm7 = xmm7[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm7 = xmm7[1,2,3,0,4,5,6,7]
; SSE-NEXT:    packuswb %xmm7, %xmm7
; SSE-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1],xmm6[2],xmm3[2],xmm6[3],xmm3[3],xmm6[4],xmm3[4],xmm6[5],xmm3[5],xmm6[6],xmm3[6],xmm6[7],xmm3[7]
; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[1,2],xmm5[0,3]
; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[2,0,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm6[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,2,2,3]
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    por %xmm1, %xmm8
; SSE-NEXT:    pand %xmm8, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm3[0],xmm8[1],xmm3[1],xmm8[2],xmm3[2],xmm8[3],xmm3[3],xmm8[4],xmm3[4],xmm8[5],xmm3[5],xmm8[6],xmm3[6],xmm8[7],xmm3[7]
; SSE-NEXT:    shufps {{.*#+}} xmm5 = xmm5[1,0],xmm8[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm5 = xmm5[2,0],xmm8[2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm5[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,0,1,2,4,5,6,7]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movd %xmm0, (%rsi)
; SSE-NEXT:    movd %xmm4, (%rdx)
; SSE-NEXT:    movd %xmm7, (%rcx)
; SSE-NEXT:    movd %xmm6, (%r8)
; SSE-NEXT:    movd %xmm1, (%r9)
; SSE-NEXT:    movd %xmm2, (%rax)
; SSE-NEXT:    retq
;
; AVX-LABEL: load_i8_stride6_vf4:
; AVX:       # %bb.0:
; AVX-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX-NEXT:    vmovdqa (%rdi), %xmm0
; AVX-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX-NEXT:    vpshufb {{.*#+}} xmm2 = zero,zero,zero,xmm1[2,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm3 = xmm0[0,6,12],zero,xmm0[u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX-NEXT:    vpshufb {{.*#+}} xmm3 = zero,zero,zero,xmm1[3,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm4 = xmm0[1,7,13],zero,xmm0[u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpor %xmm3, %xmm4, %xmm3
; AVX-NEXT:    vpshufb {{.*#+}} xmm4 = zero,zero,zero,xmm1[4,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm5 = xmm0[2,8,14],zero,xmm0[u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpor %xmm4, %xmm5, %xmm4
; AVX-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,xmm1[5,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm6 = xmm0[3,9,15],zero,xmm0[u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpor %xmm5, %xmm6, %xmm5
; AVX-NEXT:    vpshufb {{.*#+}} xmm6 = xmm1[0,6,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm7 = xmm0[4,10,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpunpcklwd {{.*#+}} xmm6 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
; AVX-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[1,7,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,11,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX-NEXT:    vmovd %xmm2, (%rsi)
; AVX-NEXT:    vmovd %xmm3, (%rdx)
; AVX-NEXT:    vmovd %xmm4, (%rcx)
; AVX-NEXT:    vmovd %xmm5, (%r8)
; AVX-NEXT:    vmovd %xmm6, (%r9)
; AVX-NEXT:    vmovd %xmm0, (%rax)
; AVX-NEXT:    retq
  %wide.vec = load <24 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <24 x i8> %wide.vec, <24 x i8> poison, <4 x i32> <i32 0, i32 6, i32 12, i32 18>
  %strided.vec1 = shufflevector <24 x i8> %wide.vec, <24 x i8> poison, <4 x i32> <i32 1, i32 7, i32 13, i32 19>
  %strided.vec2 = shufflevector <24 x i8> %wide.vec, <24 x i8> poison, <4 x i32> <i32 2, i32 8, i32 14, i32 20>
  %strided.vec3 = shufflevector <24 x i8> %wide.vec, <24 x i8> poison, <4 x i32> <i32 3, i32 9, i32 15, i32 21>
  %strided.vec4 = shufflevector <24 x i8> %wide.vec, <24 x i8> poison, <4 x i32> <i32 4, i32 10, i32 16, i32 22>
  %strided.vec5 = shufflevector <24 x i8> %wide.vec, <24 x i8> poison, <4 x i32> <i32 5, i32 11, i32 17, i32 23>
  store <4 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <4 x i8> %strided.vec1, ptr %out.vec1, align 64
  store <4 x i8> %strided.vec2, ptr %out.vec2, align 64
  store <4 x i8> %strided.vec3, ptr %out.vec3, align 64
  store <4 x i8> %strided.vec4, ptr %out.vec4, align 64
  store <4 x i8> %strided.vec5, ptr %out.vec5, align 64
  ret void
}

define void @load_i8_stride6_vf8(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4, ptr %out.vec5) nounwind {
; SSE-LABEL: load_i8_stride6_vf8:
; SSE:       # %bb.0:
; SSE-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT:    movdqa (%rdi), %xmm4
; SSE-NEXT:    movdqa 16(%rdi), %xmm3
; SSE-NEXT:    movdqa 32(%rdi), %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm8 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    movdqa %xmm4, %xmm1
; SSE-NEXT:    pand %xmm8, %xmm1
; SSE-NEXT:    pandn %xmm3, %xmm8
; SSE-NEXT:    por %xmm1, %xmm8
; SSE-NEXT:    movdqa {{.*#+}} xmm5 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm8, %xmm1
; SSE-NEXT:    pand %xmm5, %xmm1
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,3,2,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm1[0,1,2,3,4,7,6,7]
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [65535,65535,65535,0,65535,65535,65535,65535]
; SSE-NEXT:    pand %xmm1, %xmm6
; SSE-NEXT:    movdqa %xmm0, %xmm7
; SSE-NEXT:    pand %xmm5, %xmm7
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm7[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm2[0,1,2,3,4,5,6,5]
; SSE-NEXT:    packuswb %xmm9, %xmm9
; SSE-NEXT:    movdqa %xmm1, %xmm2
; SSE-NEXT:    pandn %xmm9, %xmm2
; SSE-NEXT:    por %xmm6, %xmm2
; SSE-NEXT:    pxor %xmm6, %xmm6
; SSE-NEXT:    movdqa %xmm8, %xmm9
; SSE-NEXT:    punpckhbw {{.*#+}} xmm9 = xmm9[8],xmm6[8],xmm9[9],xmm6[9],xmm9[10],xmm6[10],xmm9[11],xmm6[11],xmm9[12],xmm6[12],xmm9[13],xmm6[13],xmm9[14],xmm6[14],xmm9[15],xmm6[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm9 = xmm9[1,1,1,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm9[0,1,2,3,5,7,6,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm10 = [65535,65535,0,65535,0,0,65535,65535]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm6[0],xmm8[1],xmm6[1],xmm8[2],xmm6[2],xmm8[3],xmm6[3],xmm8[4],xmm6[4],xmm8[5],xmm6[5],xmm8[6],xmm6[6],xmm8[7],xmm6[7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm8 = xmm8[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm8[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm8 = xmm8[1,3,2,0,4,5,6,7]
; SSE-NEXT:    pand %xmm10, %xmm8
; SSE-NEXT:    pandn %xmm9, %xmm10
; SSE-NEXT:    por %xmm8, %xmm10
; SSE-NEXT:    packuswb %xmm10, %xmm10
; SSE-NEXT:    pand %xmm1, %xmm10
; SSE-NEXT:    movdqa %xmm0, %xmm8
; SSE-NEXT:    punpckhbw {{.*#+}} xmm8 = xmm8[8],xmm6[8],xmm8[9],xmm6[9],xmm8[10],xmm6[10],xmm8[11],xmm6[11],xmm8[12],xmm6[12],xmm8[13],xmm6[13],xmm8[14],xmm6[14],xmm8[15],xmm6[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1],xmm0[2],xmm6[2],xmm0[3],xmm6[3],xmm0[4],xmm6[4],xmm0[5],xmm6[5],xmm0[6],xmm6[6],xmm0[7],xmm6[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm0[2,2,3,3]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm8[0],xmm9[1],xmm8[1],xmm9[2],xmm8[2],xmm9[3],xmm8[3]
; SSE-NEXT:    packuswb %xmm9, %xmm9
; SSE-NEXT:    pandn %xmm9, %xmm1
; SSE-NEXT:    por %xmm10, %xmm1
; SSE-NEXT:    movdqa {{.*#+}} xmm11 = [65535,65535,0,65535,65535,0,65535,65535]
; SSE-NEXT:    movdqa %xmm11, %xmm9
; SSE-NEXT:    pandn %xmm3, %xmm9
; SSE-NEXT:    movdqa %xmm4, %xmm12
; SSE-NEXT:    pand %xmm11, %xmm12
; SSE-NEXT:    por %xmm9, %xmm12
; SSE-NEXT:    movdqa %xmm12, %xmm9
; SSE-NEXT:    pand %xmm5, %xmm9
; SSE-NEXT:    pshuflw {{.*#+}} xmm9 = xmm9[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm9[0,1,2,3,4,5,4,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm9 = xmm9[1,2,3,0,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm13 = xmm9[0,1,2,3,5,5,5,5]
; SSE-NEXT:    packuswb %xmm13, %xmm13
; SSE-NEXT:    movdqa {{.*#+}} xmm9 = [255,255,255,255,255,0,0,0,255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm9, %xmm13
; SSE-NEXT:    pshuflw {{.*#+}} xmm10 = xmm7[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm14 = xmm10[0,1,2,3,4,4,5,6]
; SSE-NEXT:    packuswb %xmm14, %xmm14
; SSE-NEXT:    movdqa %xmm9, %xmm10
; SSE-NEXT:    pandn %xmm14, %xmm10
; SSE-NEXT:    por %xmm13, %xmm10
; SSE-NEXT:    movdqa %xmm12, %xmm13
; SSE-NEXT:    punpcklbw {{.*#+}} xmm13 = xmm13[0],xmm6[0],xmm13[1],xmm6[1],xmm13[2],xmm6[2],xmm13[3],xmm6[3],xmm13[4],xmm6[4],xmm13[5],xmm6[5],xmm13[6],xmm6[6],xmm13[7],xmm6[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm13 = xmm13[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm13 = xmm13[3,1,2,1,4,5,6,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm14 = [0,65535,65535,0,65535,65535,65535,65535]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm12 = xmm12[8],xmm6[8],xmm12[9],xmm6[9],xmm12[10],xmm6[10],xmm12[11],xmm6[11],xmm12[12],xmm6[12],xmm12[13],xmm6[13],xmm12[14],xmm6[14],xmm12[15],xmm6[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm12 = xmm12[0,3,2,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm12 = xmm12[0,1,3,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm12 = xmm12[0,1,2,3,7,7,7,7]
; SSE-NEXT:    pand %xmm14, %xmm12
; SSE-NEXT:    pandn %xmm13, %xmm14
; SSE-NEXT:    por %xmm12, %xmm14
; SSE-NEXT:    packuswb %xmm14, %xmm14
; SSE-NEXT:    pand %xmm9, %xmm14
; SSE-NEXT:    movdqa %xmm8, %xmm12
; SSE-NEXT:    shufps {{.*#+}} xmm12 = xmm12[2,0],xmm0[3,0]
; SSE-NEXT:    movaps %xmm0, %xmm13
; SSE-NEXT:    shufps {{.*#+}} xmm13 = xmm13[0,1],xmm12[0,2]
; SSE-NEXT:    pshufhw {{.*#+}} xmm12 = xmm13[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm13 = xmm12[0,1,0,2]
; SSE-NEXT:    packuswb %xmm13, %xmm13
; SSE-NEXT:    movdqa %xmm9, %xmm12
; SSE-NEXT:    pandn %xmm13, %xmm12
; SSE-NEXT:    por %xmm14, %xmm12
; SSE-NEXT:    pand %xmm11, %xmm3
; SSE-NEXT:    pandn %xmm4, %xmm11
; SSE-NEXT:    por %xmm3, %xmm11
; SSE-NEXT:    pand %xmm11, %xmm5
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm5[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm3[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm4, %xmm4
; SSE-NEXT:    pand %xmm9, %xmm4
; SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm7[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[0,1,0,2]
; SSE-NEXT:    packuswb %xmm5, %xmm5
; SSE-NEXT:    movdqa %xmm9, %xmm3
; SSE-NEXT:    pandn %xmm5, %xmm3
; SSE-NEXT:    por %xmm4, %xmm3
; SSE-NEXT:    movdqa %xmm11, %xmm4
; SSE-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm6[8],xmm4[9],xmm6[9],xmm4[10],xmm6[10],xmm4[11],xmm6[11],xmm4[12],xmm6[12],xmm4[13],xmm6[13],xmm4[14],xmm6[14],xmm4[15],xmm6[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[1,1,2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm4[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa {{.*#+}} xmm5 = [65535,0,65535,65535,0,65535,65535,65535]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm11 = xmm11[0],xmm6[0],xmm11[1],xmm6[1],xmm11[2],xmm6[2],xmm11[3],xmm6[3],xmm11[4],xmm6[4],xmm11[5],xmm6[5],xmm11[6],xmm6[6],xmm11[7],xmm6[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm11[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm6[3,1,1,2,4,5,6,7]
; SSE-NEXT:    pand %xmm5, %xmm6
; SSE-NEXT:    pandn %xmm4, %xmm5
; SSE-NEXT:    por %xmm6, %xmm5
; SSE-NEXT:    packuswb %xmm5, %xmm5
; SSE-NEXT:    pand %xmm9, %xmm5
; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm8[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm8[2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,5,7]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pandn %xmm0, %xmm9
; SSE-NEXT:    por %xmm5, %xmm9
; SSE-NEXT:    movq %xmm2, (%rsi)
; SSE-NEXT:    movq %xmm1, (%rdx)
; SSE-NEXT:    movq %xmm10, (%rcx)
; SSE-NEXT:    movq %xmm12, (%r8)
; SSE-NEXT:    movq %xmm3, (%r9)
; SSE-NEXT:    movq %xmm9, (%rax)
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i8_stride6_vf8:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX1-ONLY-NEXT:    vmovdqa (%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovdqa 32(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = zero,zero,zero,xmm2[2,8,14,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm4 = xmm1[0,6,12],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm3, %xmm4, %xmm3
; AVX1-ONLY-NEXT:    vpxor %xmm4, %xmm4, %xmm4
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm3 = xmm3[0,1,2],xmm4[3],xmm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,zero,zero,zero,xmm0[4,10,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm5, %xmm3, %xmm3
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,xmm2[3,9,15,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm1[1,7,13],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm5, %xmm6, %xmm5
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm4 = xmm5[0,1,2],xmm4[3],xmm5[4,5,6,7]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,zero,zero,zero,xmm0[5,11,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm5, %xmm4, %xmm4
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,xmm2[4,10,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm1[2,8,14],zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm5, %xmm6, %xmm5
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm6 = [0,1,2,3,4,128,128,128,0,1,2,3,4,128,128,128]
; AVX1-ONLY-NEXT:    # xmm6 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm5, %xmm5
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = zero,zero,zero,zero,zero,xmm0[0,6,12,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm5, %xmm5
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = zero,zero,zero,xmm2[5,11,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm1[3,9,15],zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm8, %xmm7
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm7, %xmm7
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = zero,zero,zero,zero,zero,xmm0[1,7,13,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm8, %xmm7
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm1[4,10],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm9 = zero,zero,xmm2[0,6,12,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm8, %xmm9, %xmm8
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm8, %xmm8
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm9 = zero,zero,zero,zero,zero,xmm0[2,8,14,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm9, %xmm8, %xmm8
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[5,11],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = zero,zero,xmm2[1,7,13,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm1, %xmm1
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = zero,zero,zero,zero,zero,xmm0[3,9,15,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovq %xmm3, (%rsi)
; AVX1-ONLY-NEXT:    vmovq %xmm4, (%rdx)
; AVX1-ONLY-NEXT:    vmovq %xmm5, (%rcx)
; AVX1-ONLY-NEXT:    vmovq %xmm7, (%r8)
; AVX1-ONLY-NEXT:    vmovq %xmm8, (%r9)
; AVX1-ONLY-NEXT:    vmovq %xmm0, (%rax)
; AVX1-ONLY-NEXT:    retq
;
; AVX2-LABEL: load_i8_stride6_vf8:
; AVX2:       # %bb.0:
; AVX2-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
; AVX2-NEXT:    vmovdqa 32(%rdi), %ymm1
; AVX2-NEXT:    vpblendw {{.*#+}} ymm2 = ymm0[0,1],ymm1[2],ymm0[3,4],ymm1[5],ymm0[6,7,8,9],ymm1[10],ymm0[11,12],ymm1[13],ymm0[14,15]
; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm3
; AVX2-NEXT:    vpshufb {{.*#+}} xmm4 = zero,zero,zero,xmm3[2,8,14],zero,zero,xmm3[u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpshufb {{.*#+}} xmm5 = xmm2[0,6,12],zero,zero,zero,xmm2[4,10,u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpor %xmm4, %xmm5, %xmm4
; AVX2-NEXT:    vpshufb {{.*#+}} xmm3 = zero,zero,zero,xmm3[3,9,15],zero,zero,xmm3[u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[1,7,13],zero,zero,zero,xmm2[5,11,u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX2-NEXT:    vpblendw {{.*#+}} ymm3 = ymm1[0],ymm0[1,2],ymm1[3],ymm0[4,5],ymm1[6],ymm0[7],ymm1[8],ymm0[9,10],ymm1[11],ymm0[12,13],ymm1[14],ymm0[15]
; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm5
; AVX2-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,zero,xmm5[4,10],zero,zero,zero,xmm5[u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpshufb {{.*#+}} xmm7 = xmm3[2,8,14],zero,zero,xmm3[0,6,12,u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX2-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,xmm5[5,11],zero,zero,zero,xmm5[u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[3,9,15],zero,zero,xmm3[1,7,13,u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpor %xmm5, %xmm3, %xmm3
; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3],ymm1[4],ymm0[5,6],ymm1[7],ymm0[8],ymm1[9],ymm0[10,11],ymm1[12],ymm0[13,14],ymm1[15]
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,xmm1[0,6,12],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpshufb {{.*#+}} xmm7 = xmm0[4,10],zero,zero,zero,xmm0[2,8,14,u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpor %xmm5, %xmm7, %xmm5
; AVX2-NEXT:    vpshufb {{.*#+}} xmm1 = zero,zero,xmm1[1,7,13],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,11],zero,zero,zero,xmm0[3,9,15,u,u,u,u,u,u,u,u]
; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vmovq %xmm4, (%rsi)
; AVX2-NEXT:    vmovq %xmm2, (%rdx)
; AVX2-NEXT:    vmovq %xmm6, (%rcx)
; AVX2-NEXT:    vmovq %xmm3, (%r8)
; AVX2-NEXT:    vmovq %xmm5, (%r9)
; AVX2-NEXT:    vmovq %xmm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
  %wide.vec = load <48 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <48 x i8> %wide.vec, <48 x i8> poison, <8 x i32> <i32 0, i32 6, i32 12, i32 18, i32 24, i32 30, i32 36, i32 42>
  %strided.vec1 = shufflevector <48 x i8> %wide.vec, <48 x i8> poison, <8 x i32> <i32 1, i32 7, i32 13, i32 19, i32 25, i32 31, i32 37, i32 43>
  %strided.vec2 = shufflevector <48 x i8> %wide.vec, <48 x i8> poison, <8 x i32> <i32 2, i32 8, i32 14, i32 20, i32 26, i32 32, i32 38, i32 44>
  %strided.vec3 = shufflevector <48 x i8> %wide.vec, <48 x i8> poison, <8 x i32> <i32 3, i32 9, i32 15, i32 21, i32 27, i32 33, i32 39, i32 45>
  %strided.vec4 = shufflevector <48 x i8> %wide.vec, <48 x i8> poison, <8 x i32> <i32 4, i32 10, i32 16, i32 22, i32 28, i32 34, i32 40, i32 46>
  %strided.vec5 = shufflevector <48 x i8> %wide.vec, <48 x i8> poison, <8 x i32> <i32 5, i32 11, i32 17, i32 23, i32 29, i32 35, i32 41, i32 47>
  store <8 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <8 x i8> %strided.vec1, ptr %out.vec1, align 64
  store <8 x i8> %strided.vec2, ptr %out.vec2, align 64
  store <8 x i8> %strided.vec3, ptr %out.vec3, align 64
  store <8 x i8> %strided.vec4, ptr %out.vec4, align 64
  store <8 x i8> %strided.vec5, ptr %out.vec5, align 64
  ret void
}

define void @load_i8_stride6_vf16(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4, ptr %out.vec5) nounwind {
; SSE-LABEL: load_i8_stride6_vf16:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa 64(%rdi), %xmm11
; SSE-NEXT:    movdqa %xmm11, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa (%rdi), %xmm5
; SSE-NEXT:    movdqa 16(%rdi), %xmm2
; SSE-NEXT:    movdqa 32(%rdi), %xmm8
; SSE-NEXT:    movdqa 48(%rdi), %xmm6
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [65535,65535,0,65535,65535,0,65535,65535]
; SSE-NEXT:    movdqa %xmm4, %xmm7
; SSE-NEXT:    pandn %xmm8, %xmm7
; SSE-NEXT:    movdqa {{.*#+}} xmm10 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    pandn %xmm6, %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm4, %xmm0
; SSE-NEXT:    pandn %xmm6, %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm4, %xmm6
; SSE-NEXT:    por %xmm7, %xmm6
; SSE-NEXT:    movdqa {{.*#+}} xmm7 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm6, %xmm0
; SSE-NEXT:    pand %xmm7, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,5]
; SSE-NEXT:    packuswb %xmm3, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm9 = [65535,65535,65535,0,0,0,65535,65535]
; SSE-NEXT:    pandn %xmm0, %xmm9
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    pandn %xmm2, %xmm10
; SSE-NEXT:    pand %xmm4, %xmm11
; SSE-NEXT:    movdqa %xmm4, %xmm3
; SSE-NEXT:    pandn %xmm2, %xmm3
; SSE-NEXT:    movdqa %xmm5, %xmm14
; SSE-NEXT:    pand %xmm4, %xmm14
; SSE-NEXT:    movdqa 80(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, %xmm13
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm4, %xmm13
; SSE-NEXT:    movdqa %xmm8, %xmm15
; SSE-NEXT:    pand %xmm4, %xmm8
; SSE-NEXT:    pand %xmm4, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm4, %xmm12
; SSE-NEXT:    movdqa %xmm4, %xmm2
; SSE-NEXT:    pandn %xmm5, %xmm4
; SSE-NEXT:    pand %xmm0, %xmm5
; SSE-NEXT:    por %xmm10, %xmm5
; SSE-NEXT:    movdqa %xmm5, %xmm10
; SSE-NEXT:    pand %xmm7, %xmm10
; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm10 = xmm10[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm10 = xmm10[0,3,2,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm10[0,1,2,3,4,7,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE-NEXT:    por %xmm9, %xmm0
; SSE-NEXT:    pandn %xmm1, %xmm12
; SSE-NEXT:    por %xmm12, %xmm11
; SSE-NEXT:    movdqa %xmm11, %xmm9
; SSE-NEXT:    pand %xmm7, %xmm9
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[3,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm9 = xmm9[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[0,3,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm9[0,1,2,3,4,7,6,5]
; SSE-NEXT:    packuswb %xmm9, %xmm9
; SSE-NEXT:    movdqa {{.*#+}} xmm10 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; SSE-NEXT:    movdqa %xmm10, %xmm1
; SSE-NEXT:    pandn %xmm9, %xmm1
; SSE-NEXT:    pand %xmm10, %xmm0
; SSE-NEXT:    por %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pxor %xmm9, %xmm9
; SSE-NEXT:    movdqa %xmm6, %xmm0
; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm9[8],xmm0[9],xmm9[9],xmm0[10],xmm9[10],xmm0[11],xmm9[11],xmm0[12],xmm9[12],xmm0[13],xmm9[13],xmm0[14],xmm9[14],xmm0[15],xmm9[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm9[0],xmm6[1],xmm9[1],xmm6[2],xmm9[2],xmm6[3],xmm9[3],xmm6[4],xmm9[4],xmm6[5],xmm9[5],xmm6[6],xmm9[6],xmm6[7],xmm9[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[2,2,3,3]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE-NEXT:    psrld $16, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm6 = xmm6[2],xmm0[2],xmm6[3],xmm0[3]
; SSE-NEXT:    packuswb %xmm6, %xmm1
; SSE-NEXT:    movdqa %xmm5, %xmm0
; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm9[8],xmm0[9],xmm9[9],xmm0[10],xmm9[10],xmm0[11],xmm9[11],xmm0[12],xmm9[12],xmm0[13],xmm9[13],xmm0[14],xmm9[14],xmm0[15],xmm9[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,1,1,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm9[0],xmm5[1],xmm9[1],xmm5[2],xmm9[2],xmm5[3],xmm9[3],xmm5[4],xmm9[4],xmm5[5],xmm9[5],xmm5[6],xmm9[6],xmm5[7],xmm9[7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[1,3,2,0,4,5,6,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm6 = [65535,65535,0,65535,0,0,65535,65535]
; SSE-NEXT:    pand %xmm6, %xmm5
; SSE-NEXT:    pandn %xmm0, %xmm6
; SSE-NEXT:    por %xmm5, %xmm6
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [65535,65535,65535,0,0,0,65535,65535]
; SSE-NEXT:    pand %xmm0, %xmm6
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    por %xmm0, %xmm6
; SSE-NEXT:    movdqa %xmm11, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1],xmm0[2],xmm9[2],xmm0[3],xmm9[3],xmm0[4],xmm9[4],xmm0[5],xmm9[5],xmm0[6],xmm9[6],xmm0[7],xmm9[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,5,5]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm11 = xmm11[8],xmm9[8],xmm11[9],xmm9[9],xmm11[10],xmm9[10],xmm11[11],xmm9[11],xmm11[12],xmm9[12],xmm11[13],xmm9[13],xmm11[14],xmm9[14],xmm11[15],xmm9[15]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm11[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,7,6,4]
; SSE-NEXT:    movdqa {{.*#+}} xmm5 = [65535,65535,65535,0,65535,65535,0,65535]
; SSE-NEXT:    pand %xmm5, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm5
; SSE-NEXT:    por %xmm1, %xmm5
; SSE-NEXT:    packuswb %xmm5, %xmm0
; SSE-NEXT:    movdqa %xmm10, %xmm12
; SSE-NEXT:    pandn %xmm0, %xmm12
; SSE-NEXT:    pand %xmm10, %xmm6
; SSE-NEXT:    por %xmm6, %xmm12
; SSE-NEXT:    pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm15
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm15, %xmm0
; SSE-NEXT:    pand %xmm7, %xmm0
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,3,4,5,6,7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,5,6]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    por %xmm3, %xmm14
; SSE-NEXT:    movdqa %xmm14, %xmm1
; SSE-NEXT:    pand %xmm7, %xmm1
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,4,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,0,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm3, %xmm5
; SSE-NEXT:    pandn %xmm1, %xmm5
; SSE-NEXT:    pand %xmm3, %xmm0
; SSE-NEXT:    por %xmm0, %xmm5
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    pandn %xmm11, %xmm2
; SSE-NEXT:    por %xmm2, %xmm13
; SSE-NEXT:    movdqa %xmm13, %xmm0
; SSE-NEXT:    pand %xmm7, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,2,2,2,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,6,7,4]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa %xmm10, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm10, %xmm5
; SSE-NEXT:    por %xmm5, %xmm1
; SSE-NEXT:    movdqa %xmm15, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1],xmm0[2],xmm9[2],xmm0[3],xmm9[3],xmm0[4],xmm9[4],xmm0[5],xmm9[5],xmm0[6],xmm9[6],xmm0[7],xmm9[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm15 = xmm15[8],xmm9[8],xmm15[9],xmm9[9],xmm15[10],xmm9[10],xmm15[11],xmm9[11],xmm15[12],xmm9[12],xmm15[13],xmm9[13],xmm15[14],xmm9[14],xmm15[15],xmm9[15]
; SSE-NEXT:    movdqa %xmm15, %xmm2
; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm0[3,0]
; SSE-NEXT:    movaps %xmm0, %xmm5
; SSE-NEXT:    shufps {{.*#+}} xmm5 = xmm5[0,1],xmm2[0,2]
; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm15[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm15[2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm5[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,0,2]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,1,3,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm14, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1],xmm0[2],xmm9[2],xmm0[3],xmm9[3],xmm0[4],xmm9[4],xmm0[5],xmm9[5],xmm0[6],xmm9[6],xmm0[7],xmm9[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,1,4,5,6,7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm14 = xmm14[8],xmm9[8],xmm14[9],xmm9[9],xmm14[10],xmm9[10],xmm14[11],xmm9[11],xmm14[12],xmm9[12],xmm14[13],xmm9[13],xmm14[14],xmm9[14],xmm14[15],xmm9[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm14[0,3,2,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[0,1,3,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,7,7,7,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm6 = [0,65535,65535,0,65535,65535,65535,65535]
; SSE-NEXT:    pand %xmm6, %xmm5
; SSE-NEXT:    pandn %xmm0, %xmm6
; SSE-NEXT:    por %xmm5, %xmm6
; SSE-NEXT:    pand %xmm3, %xmm2
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    pandn %xmm6, %xmm3
; SSE-NEXT:    por %xmm2, %xmm3
; SSE-NEXT:    movdqa %xmm13, %xmm0
; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm9[8],xmm0[9],xmm9[9],xmm0[10],xmm9[10],xmm0[11],xmm9[11],xmm0[12],xmm9[12],xmm0[13],xmm9[13],xmm0[14],xmm9[14],xmm0[15],xmm9[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,5]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm13 = xmm13[0],xmm9[0],xmm13[1],xmm9[1],xmm13[2],xmm9[2],xmm13[3],xmm9[3],xmm13[4],xmm9[4],xmm13[5],xmm9[5],xmm13[6],xmm9[6],xmm13[7],xmm9[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm13[0,2,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,7,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm5 = [65535,65535,65535,65535,0,65535,65535,0]
; SSE-NEXT:    pand %xmm5, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm5
; SSE-NEXT:    por %xmm2, %xmm5
; SSE-NEXT:    pand %xmm10, %xmm3
; SSE-NEXT:    packuswb %xmm5, %xmm0
; SSE-NEXT:    pandn %xmm0, %xmm10
; SSE-NEXT:    por %xmm3, %xmm10
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm8, %xmm0
; SSE-NEXT:    pand %xmm7, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
; SSE-NEXT:    packuswb %xmm2, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,0,0,0,0,0,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm3, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm4, %xmm0
; SSE-NEXT:    pand %xmm7, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm0[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm5, %xmm5
; SSE-NEXT:    pand %xmm3, %xmm5
; SSE-NEXT:    por %xmm2, %xmm5
; SSE-NEXT:    movdqa {{.*#+}} xmm13 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    pand %xmm13, %xmm11
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Folded Reload
; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [65535,65535,65535,65535,65535,0,0,0]
; SSE-NEXT:    pand %xmm0, %xmm5
; SSE-NEXT:    por %xmm11, %xmm13
; SSE-NEXT:    pand %xmm13, %xmm7
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm7[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,2,1,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,6,5,4,7]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm0, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    por %xmm5, %xmm6
; SSE-NEXT:    movdqa %xmm8, %xmm2
; SSE-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm9[8],xmm2[9],xmm9[9],xmm2[10],xmm9[10],xmm2[11],xmm9[11],xmm2[12],xmm9[12],xmm2[13],xmm9[13],xmm2[14],xmm9[14],xmm2[15],xmm9[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm9[0],xmm8[1],xmm9[1],xmm8[2],xmm9[2],xmm8[3],xmm9[3],xmm8[4],xmm9[4],xmm8[5],xmm9[5],xmm8[6],xmm9[6],xmm8[7],xmm9[7]
; SSE-NEXT:    movdqa %xmm8, %xmm5
; SSE-NEXT:    shufps {{.*#+}} xmm8 = xmm8[1,0],xmm2[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm8 = xmm8[2,0],xmm2[2,3]
; SSE-NEXT:    psrlq $48, %xmm2
; SSE-NEXT:    psrldq {{.*#+}} xmm5 = xmm5[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm2[0],xmm5[1],xmm2[1],xmm5[2],xmm2[2],xmm5[3],xmm2[3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm8[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm7 = xmm2[0,1,2,3,4,4,5,7]
; SSE-NEXT:    packuswb %xmm5, %xmm7
; SSE-NEXT:    movdqa %xmm4, %xmm2
; SSE-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm9[8],xmm2[9],xmm9[9],xmm2[10],xmm9[10],xmm2[11],xmm9[11],xmm2[12],xmm9[12],xmm2[13],xmm9[13],xmm2[14],xmm9[14],xmm2[15],xmm9[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm2[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [65535,0,65535,65535,0,65535,65535,65535]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm9[0],xmm4[1],xmm9[1],xmm4[2],xmm9[2],xmm4[3],xmm9[3],xmm4[4],xmm9[4],xmm4[5],xmm9[5],xmm4[6],xmm9[6],xmm4[7],xmm9[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm4[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[3,1,1,2,4,5,6,7]
; SSE-NEXT:    pand %xmm2, %xmm4
; SSE-NEXT:    pandn %xmm5, %xmm2
; SSE-NEXT:    por %xmm4, %xmm2
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    pand %xmm3, %xmm2
; SSE-NEXT:    pandn %xmm7, %xmm3
; SSE-NEXT:    por %xmm3, %xmm2
; SSE-NEXT:    movdqa %xmm13, %xmm3
; SSE-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]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,3,1,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,1,3,4,5,6,7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm13 = xmm13[8],xmm9[8],xmm13[9],xmm9[9],xmm13[10],xmm9[10],xmm13[11],xmm9[11],xmm13[12],xmm9[12],xmm13[13],xmm9[13],xmm13[14],xmm9[14],xmm13[15],xmm9[15]
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535,0,65535,0,0]
; SSE-NEXT:    pand %xmm4, %xmm3
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm13[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,5,5,7,4]
; SSE-NEXT:    pandn %xmm5, %xmm4
; SSE-NEXT:    por %xmm3, %xmm4
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    packuswb %xmm4, %xmm3
; SSE-NEXT:    pandn %xmm3, %xmm0
; SSE-NEXT:    por %xmm2, %xmm0
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, (%rsi)
; SSE-NEXT:    movdqa %xmm12, (%rdx)
; SSE-NEXT:    movdqa %xmm1, (%rcx)
; SSE-NEXT:    movdqa %xmm10, (%r8)
; SSE-NEXT:    movdqa %xmm6, (%r9)
; SSE-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT:    movdqa %xmm0, (%rax)
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i8_stride6_vf16:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX1-ONLY-NEXT:    vmovdqa (%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovdqa 32(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovdqa 48(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm4 = xmm0[u,u,4,10,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = xmm3[u,u,u,u,0,6,12,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm4 = xmm5[0],xmm4[0],xmm5[1],xmm4[1]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,xmm2[2,8,14,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm1[0,6,12],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm5, %xmm6, %xmm5
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm6 = xmm5[0,1,2],xmm4[3,4,5],xmm5[6,7]
; AVX1-ONLY-NEXT:    vmovdqa 80(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = xmm4[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm4[4,10]
; AVX1-ONLY-NEXT:    vmovdqa 64(%rdi), %xmm5
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm5[u,u,u,u,u,u,u,u,u,u,u,2,8,14],zero,zero
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm8, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm9 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; AVX1-ONLY-NEXT:    vpblendvb %xmm9, %xmm6, %xmm7, %xmm6
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = xmm0[u,u,5,11,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm3[u,u,u,u,1,7,13,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm7 = xmm8[0],xmm7[0],xmm8[1],xmm7[1]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = zero,zero,zero,xmm2[3,9,15,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm1[1,7,13],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm8, %xmm10, %xmm8
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm7 = xmm8[0,1,2],xmm7[3,4,5],xmm8[6,7]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm4[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm4[5,11]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm5[u,u,u,u,u,u,u,u,u,u,u,3,9,15],zero,zero
; AVX1-ONLY-NEXT:    vpor %xmm8, %xmm10, %xmm8
; AVX1-ONLY-NEXT:    vpblendvb %xmm9, %xmm7, %xmm8, %xmm7
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm3[2,8,14,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm0[u,u,u,u,u,0,6,12,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm8 = xmm10[0],xmm8[0]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = zero,zero,zero,xmm2[4,10,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = xmm1[2,8,14],zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm10, %xmm11, %xmm10
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm11 = <0,0,0,0,0,255,255,255,255,255,255,u,u,u,u,u>
; AVX1-ONLY-NEXT:    vpblendvb %xmm11, %xmm8, %xmm10, %xmm8
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm5[u,u,u,u,u,u,u,u,u,u,u,4,10],zero,zero,zero
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = xmm4[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm4[0,6,12]
; AVX1-ONLY-NEXT:    vpor %xmm10, %xmm12, %xmm10
; AVX1-ONLY-NEXT:    vpblendvb %xmm9, %xmm8, %xmm10, %xmm8
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm3[3,9,15,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = xmm0[u,u,u,u,u,1,7,13,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm10 = xmm12[0],xmm10[0]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = zero,zero,zero,xmm2[5,11,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm1[3,9,15],zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm12, %xmm13, %xmm12
; AVX1-ONLY-NEXT:    vpblendvb %xmm11, %xmm10, %xmm12, %xmm10
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = xmm5[u,u,u,u,u,u,u,u,u,u,u,5,11],zero,zero,zero
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = xmm4[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm4[1,7,13]
; AVX1-ONLY-NEXT:    vpor %xmm11, %xmm12, %xmm11
; AVX1-ONLY-NEXT:    vpblendvb %xmm9, %xmm10, %xmm11, %xmm9
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm1[4,10],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = zero,zero,xmm2[0,6,12,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm10, %xmm11, %xmm10
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = xmm3[u,u,u,u,u,u,u,u,4,10,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = xmm0[u,u,u,u,u,u,u,u,u,u,u,u,u,2,8,14]
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm11 = xmm12[1],xmm11[1]
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm12 = <255,255,255,255,255,0,0,0,0,0,u,u,u,u,u,u>
; AVX1-ONLY-NEXT:    vpblendvb %xmm12, %xmm10, %xmm11, %xmm10
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = xmm4[u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm4[2,8,14]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm5[u,u,u,u,u,u,u,u,u,u,0,6,12],zero,zero,zero
; AVX1-ONLY-NEXT:    vpor %xmm11, %xmm13, %xmm11
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm10 = xmm10[0,1,2,3,4],xmm11[5,6,7]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[5,11],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = zero,zero,xmm2[1,7,13,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm3[u,u,u,u,u,u,u,u,5,11,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,u,u,u,u,u,u,u,u,u,3,9,15]
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm2[1]
; AVX1-ONLY-NEXT:    vpblendvb %xmm12, %xmm1, %xmm0, %xmm0
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm4[u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm4[3,9,15]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm5[u,u,u,u,u,u,u,u,u,u,1,7,13],zero,zero,zero
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5,6,7]
; AVX1-ONLY-NEXT:    vmovdqa %xmm6, (%rsi)
; AVX1-ONLY-NEXT:    vmovdqa %xmm7, (%rdx)
; AVX1-ONLY-NEXT:    vmovdqa %xmm8, (%rcx)
; AVX1-ONLY-NEXT:    vmovdqa %xmm9, (%r8)
; AVX1-ONLY-NEXT:    vmovdqa %xmm10, (%r9)
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, (%rax)
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i8_stride6_vf16:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm0 = <255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255>
; AVX2-ONLY-NEXT:    vpblendvb %ymm0, %ymm3, %ymm4, %ymm5
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm5[0,6,12],zero,zero,zero,xmm5[4,10],zero,zero,zero,xmm5[u,u,u,u,u]
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm5, %xmm6
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = zero,zero,zero,xmm6[2,8,14],zero,zero,xmm6[0,6,12,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm2
; AVX2-ONLY-NEXT:    vmovdqa 80(%rdi), %xmm0
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = xmm0[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm0[4,10]
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %xmm1
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm1[u,u,u,u,u,u,u,u,u,u,u,2,8,14],zero,zero
; AVX2-ONLY-NEXT:    vpor %xmm7, %xmm8, %xmm7
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm8 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; AVX2-ONLY-NEXT:    vpblendvb %xmm8, %xmm2, %xmm7, %xmm2
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = xmm5[1,7,13],zero,zero,zero,xmm5[5,11],zero,zero,zero,xmm5[u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,zero,xmm6[3,9,15],zero,zero,xmm6[1,7,13,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm5, %xmm6, %xmm5
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm0[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm0[5,11]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = xmm1[u,u,u,u,u,u,u,u,u,u,u,3,9,15],zero,zero
; AVX2-ONLY-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX2-ONLY-NEXT:    vpblendvb %xmm8, %xmm5, %xmm6, %xmm5
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm6 = <255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255>
; AVX2-ONLY-NEXT:    vpblendvb %ymm6, %ymm4, %ymm3, %ymm6
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm6, %xmm7
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm9 = zero,zero,zero,xmm7[4,10],zero,zero,zero,xmm7[2,8,14,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm6[2,8,14],zero,zero,xmm6[0,6,12],zero,zero,zero,xmm6[u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm9, %xmm10, %xmm9
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm1[u,u,u,u,u,u,u,u,u,u,u,4,10],zero,zero,zero
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = xmm0[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm0[0,6,12]
; AVX2-ONLY-NEXT:    vpor %xmm10, %xmm11, %xmm10
; AVX2-ONLY-NEXT:    vpblendvb %xmm8, %xmm9, %xmm10, %xmm9
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = zero,zero,zero,xmm7[5,11],zero,zero,zero,xmm7[3,9,15,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm6[3,9,15],zero,zero,xmm6[1,7,13],zero,zero,zero,xmm6[u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm7, %xmm6, %xmm6
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = xmm1[u,u,u,u,u,u,u,u,u,u,u,5,11],zero,zero,zero
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm0[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm0[1,7,13]
; AVX2-ONLY-NEXT:    vpor %xmm7, %xmm10, %xmm7
; AVX2-ONLY-NEXT:    vpblendvb %xmm8, %xmm6, %xmm7, %xmm6
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm7 = <u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u>
; AVX2-ONLY-NEXT:    vpblendvb %ymm7, %ymm4, %ymm3, %ymm3
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm3, %xmm4
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = zero,zero,xmm4[0,6,12],zero,zero,zero,xmm4[4,10,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm3[4,10],zero,zero,zero,xmm3[2,8,14],zero,zero,xmm3[u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm7, %xmm8, %xmm7
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm0[u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm0[2,8,14]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm1[u,u,u,u,u,u,u,u,u,u,0,6,12],zero,zero,zero
; AVX2-ONLY-NEXT:    vpor %xmm8, %xmm10, %xmm8
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} xmm7 = xmm7[0,1,2,3,4],xmm8[5,6,7]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm4 = zero,zero,xmm4[1,7,13],zero,zero,zero,xmm4[5,11,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[5,11],zero,zero,zero,xmm3[3,9,15],zero,zero,xmm3[u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm4, %xmm3, %xmm3
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm0[3,9,15]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[u,u,u,u,u,u,u,u,u,u,1,7,13],zero,zero,zero
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} xmm0 = xmm3[0,1,2,3,4],xmm0[5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa %xmm2, (%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %xmm5, (%rdx)
; AVX2-ONLY-NEXT:    vmovdqa %xmm9, (%rcx)
; AVX2-ONLY-NEXT:    vmovdqa %xmm6, (%r8)
; AVX2-ONLY-NEXT:    vmovdqa %xmm7, (%r9)
; AVX2-ONLY-NEXT:    vmovdqa %xmm0, (%rax)
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-LABEL: load_i8_stride6_vf16:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm0 = [65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535]
; AVX512F-NEXT:    vmovdqa (%rdi), %ymm3
; AVX512F-NEXT:    vmovdqa 32(%rdi), %ymm4
; AVX512F-NEXT:    vmovdqa %ymm0, %ymm5
; AVX512F-NEXT:    vpternlogq $202, %ymm4, %ymm3, %ymm5
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm1 = xmm5[0,6,12],zero,zero,zero,xmm5[4,10],zero,zero,zero,xmm5[u,u,u,u,u]
; AVX512F-NEXT:    vextracti128 $1, %ymm5, %xmm6
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm2 = zero,zero,zero,xmm6[2,8,14],zero,zero,xmm6[0,6,12,u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm1, %xmm2, %xmm7
; AVX512F-NEXT:    vmovdqa 80(%rdi), %xmm2
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm8 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm2[4,10]
; AVX512F-NEXT:    vmovdqa 64(%rdi), %xmm1
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm9 = xmm1[u,u,u,u,u,u,u,u,u,u,u,2,8,14],zero,zero
; AVX512F-NEXT:    vpor %xmm8, %xmm9, %xmm8
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm9 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; AVX512F-NEXT:    vpternlogq $184, %xmm7, %xmm9, %xmm8
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm5 = xmm5[1,7,13],zero,zero,zero,xmm5[5,11],zero,zero,zero,xmm5[u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,zero,xmm6[3,9,15],zero,zero,xmm6[1,7,13,u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm5, %xmm6, %xmm5
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm6 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm2[5,11]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm7 = xmm1[u,u,u,u,u,u,u,u,u,u,u,3,9,15],zero,zero
; AVX512F-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX512F-NEXT:    vpternlogq $184, %xmm5, %xmm9, %xmm6
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm5 = xmm1[u,u,u,u,u,u,u,u,u,u,u,4,10],zero,zero,zero
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm7 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm2[0,6,12]
; AVX512F-NEXT:    vpor %xmm5, %xmm7, %xmm5
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm7 = [65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535]
; AVX512F-NEXT:    vpternlogq $202, %ymm3, %ymm4, %ymm7
; AVX512F-NEXT:    vextracti128 $1, %ymm7, %xmm10
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm11 = zero,zero,zero,xmm10[4,10],zero,zero,zero,xmm10[2,8,14,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm12 = xmm7[2,8,14],zero,zero,xmm7[0,6,12],zero,zero,zero,xmm7[u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm11, %xmm12, %xmm11
; AVX512F-NEXT:    vpternlogq $226, %xmm5, %xmm9, %xmm11
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm5 = xmm1[u,u,u,u,u,u,u,u,u,u,u,5,11],zero,zero,zero
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm12 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm2[1,7,13]
; AVX512F-NEXT:    vpor %xmm5, %xmm12, %xmm5
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm10 = zero,zero,zero,xmm10[5,11],zero,zero,zero,xmm10[3,9,15,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm7 = xmm7[3,9,15],zero,zero,xmm7[1,7,13],zero,zero,zero,xmm7[u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm7, %xmm10, %xmm7
; AVX512F-NEXT:    vpternlogq $226, %xmm5, %xmm9, %xmm7
; AVX512F-NEXT:    vpternlogq $202, %ymm3, %ymm4, %ymm0
; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm4 = zero,zero,xmm3[0,6,12],zero,zero,zero,xmm3[4,10,u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm5 = xmm2[u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm2[2,8,14]
; AVX512F-NEXT:    vpblendw {{.*#+}} xmm4 = xmm4[0,1,2,3,4],xmm5[5,6,7]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm5 = xmm0[4,10],zero,zero,zero,xmm0[2,8,14],zero,zero,xmm0[u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm9 = xmm1[u,u,u,u,u,u,u,u,u,u,0,6,12],zero,zero,zero
; AVX512F-NEXT:    vpblendw {{.*#+}} xmm5 = xmm5[0,1,2,3,4],xmm9[5,6,7]
; AVX512F-NEXT:    vpor %xmm4, %xmm5, %xmm4
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm3 = zero,zero,xmm3[1,7,13],zero,zero,zero,xmm3[5,11,u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm2[3,9,15]
; AVX512F-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3,4],xmm2[5,6,7]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,11],zero,zero,zero,xmm0[3,9,15],zero,zero,xmm0[u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[u,u,u,u,u,u,u,u,u,u,1,7,13],zero,zero,zero
; AVX512F-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5,6,7]
; AVX512F-NEXT:    vpor %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqa %xmm8, (%rsi)
; AVX512F-NEXT:    vmovdqa %xmm6, (%rdx)
; AVX512F-NEXT:    vmovdqa %xmm11, (%rcx)
; AVX512F-NEXT:    vmovdqa %xmm7, (%r8)
; AVX512F-NEXT:    vmovdqa %xmm4, (%r9)
; AVX512F-NEXT:    vmovdqa %xmm0, (%rax)
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride6_vf16:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX512BW-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512BW-NEXT:    vmovdqa 32(%rdi), %ymm1
; AVX512BW-NEXT:    movw $18724, %r10w # imm = 0x4924
; AVX512BW-NEXT:    kmovd %r10d, %k1
; AVX512BW-NEXT:    vpblendmw %ymm1, %ymm0, %ymm2 {%k1}
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm3 = xmm2[0,6,12],zero,zero,zero,xmm2[4,10],zero,zero,zero,xmm2[u,u,u,u,u]
; AVX512BW-NEXT:    vextracti128 $1, %ymm2, %xmm4
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,xmm4[2,8,14],zero,zero,xmm4[0,6,12,u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm3, %xmm5, %xmm3
; AVX512BW-NEXT:    vmovdqa 80(%rdi), %xmm5
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm6 = xmm5[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm5[4,10]
; AVX512BW-NEXT:    vmovdqa 64(%rdi), %xmm7
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm8 = xmm7[u,u,u,u,u,u,u,u,u,u,u,2,8,14],zero,zero
; AVX512BW-NEXT:    vpor %xmm6, %xmm8, %xmm6
; AVX512BW-NEXT:    movw $-2048, %di # imm = 0xF800
; AVX512BW-NEXT:    kmovd %edi, %k2
; AVX512BW-NEXT:    vmovdqu8 %xmm6, %xmm3 {%k2}
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[1,7,13],zero,zero,zero,xmm2[5,11],zero,zero,zero,xmm2[u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm4 = zero,zero,zero,xmm4[3,9,15],zero,zero,xmm4[1,7,13,u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm2, %xmm4, %xmm2
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm4 = xmm5[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm5[5,11]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm6 = xmm7[u,u,u,u,u,u,u,u,u,u,u,3,9,15],zero,zero
; AVX512BW-NEXT:    vpor %xmm4, %xmm6, %xmm4
; AVX512BW-NEXT:    vmovdqu8 %xmm4, %xmm2 {%k2}
; AVX512BW-NEXT:    movw $9362, %di # imm = 0x2492
; AVX512BW-NEXT:    kmovd %edi, %k3
; AVX512BW-NEXT:    vpblendmw %ymm0, %ymm1, %ymm4 {%k3}
; AVX512BW-NEXT:    vextracti128 $1, %ymm4, %xmm6
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm8 = zero,zero,zero,xmm6[4,10],zero,zero,zero,xmm6[2,8,14,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm9 = xmm4[2,8,14],zero,zero,xmm4[0,6,12],zero,zero,zero,xmm4[u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm8, %xmm9, %xmm8
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm9 = xmm7[u,u,u,u,u,u,u,u,u,u,u,4,10],zero,zero,zero
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm10 = xmm5[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm5[0,6,12]
; AVX512BW-NEXT:    vpor %xmm9, %xmm10, %xmm9
; AVX512BW-NEXT:    vmovdqu8 %xmm9, %xmm8 {%k2}
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,zero,xmm6[5,11],zero,zero,zero,xmm6[3,9,15,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm4 = xmm4[3,9,15],zero,zero,xmm4[1,7,13],zero,zero,zero,xmm4[u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm6, %xmm4, %xmm4
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm6 = xmm7[u,u,u,u,u,u,u,u,u,u,u,5,11],zero,zero,zero
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm9 = xmm5[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm5[1,7,13]
; AVX512BW-NEXT:    vpor %xmm6, %xmm9, %xmm6
; AVX512BW-NEXT:    vmovdqu8 %xmm6, %xmm4 {%k2}
; AVX512BW-NEXT:    vmovdqu16 %ymm0, %ymm1 {%k1}
; AVX512BW-NEXT:    vextracti128 $1, %ymm1, %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,xmm0[0,6,12],zero,zero,zero,xmm0[4,10,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm9 = xmm5[u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm5[2,8,14]
; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm6 = xmm6[0,1,2,3,4],xmm9[5,6,7]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm9 = xmm1[4,10],zero,zero,zero,xmm1[2,8,14],zero,zero,xmm1[u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm10 = xmm7[u,u,u,u,u,u,u,u,u,u,0,6,12],zero,zero,zero
; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm9 = xmm9[0,1,2,3,4],xmm10[5,6,7]
; AVX512BW-NEXT:    vpor %xmm6, %xmm9, %xmm6
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = zero,zero,xmm0[1,7,13],zero,zero,zero,xmm0[5,11,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm5 = xmm5[u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm5[3,9,15]
; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm5[5,6,7]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[5,11],zero,zero,zero,xmm1[3,9,15],zero,zero,xmm1[u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm5 = xmm7[u,u,u,u,u,u,u,u,u,u,1,7,13],zero,zero,zero
; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3,4],xmm5[5,6,7]
; AVX512BW-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX512BW-NEXT:    vmovdqa %xmm3, (%rsi)
; AVX512BW-NEXT:    vmovdqa %xmm2, (%rdx)
; AVX512BW-NEXT:    vmovdqa %xmm8, (%rcx)
; AVX512BW-NEXT:    vmovdqa %xmm4, (%r8)
; AVX512BW-NEXT:    vmovdqa %xmm6, (%r9)
; AVX512BW-NEXT:    vmovdqa %xmm0, (%rax)
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <96 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <96 x i8> %wide.vec, <96 x i8> poison, <16 x i32> <i32 0, i32 6, i32 12, i32 18, i32 24, i32 30, i32 36, i32 42, i32 48, i32 54, i32 60, i32 66, i32 72, i32 78, i32 84, i32 90>
  %strided.vec1 = shufflevector <96 x i8> %wide.vec, <96 x i8> poison, <16 x i32> <i32 1, i32 7, i32 13, i32 19, i32 25, i32 31, i32 37, i32 43, i32 49, i32 55, i32 61, i32 67, i32 73, i32 79, i32 85, i32 91>
  %strided.vec2 = shufflevector <96 x i8> %wide.vec, <96 x i8> poison, <16 x i32> <i32 2, i32 8, i32 14, i32 20, i32 26, i32 32, i32 38, i32 44, i32 50, i32 56, i32 62, i32 68, i32 74, i32 80, i32 86, i32 92>
  %strided.vec3 = shufflevector <96 x i8> %wide.vec, <96 x i8> poison, <16 x i32> <i32 3, i32 9, i32 15, i32 21, i32 27, i32 33, i32 39, i32 45, i32 51, i32 57, i32 63, i32 69, i32 75, i32 81, i32 87, i32 93>
  %strided.vec4 = shufflevector <96 x i8> %wide.vec, <96 x i8> poison, <16 x i32> <i32 4, i32 10, i32 16, i32 22, i32 28, i32 34, i32 40, i32 46, i32 52, i32 58, i32 64, i32 70, i32 76, i32 82, i32 88, i32 94>
  %strided.vec5 = shufflevector <96 x i8> %wide.vec, <96 x i8> poison, <16 x i32> <i32 5, i32 11, i32 17, i32 23, i32 29, i32 35, i32 41, i32 47, i32 53, i32 59, i32 65, i32 71, i32 77, i32 83, i32 89, i32 95>
  store <16 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <16 x i8> %strided.vec1, ptr %out.vec1, align 64
  store <16 x i8> %strided.vec2, ptr %out.vec2, align 64
  store <16 x i8> %strided.vec3, ptr %out.vec3, align 64
  store <16 x i8> %strided.vec4, ptr %out.vec4, align 64
  store <16 x i8> %strided.vec5, ptr %out.vec5, align 64
  ret void
}

define void @load_i8_stride6_vf32(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4, ptr %out.vec5) nounwind {
; SSE-LABEL: load_i8_stride6_vf32:
; SSE:       # %bb.0:
; SSE-NEXT:    subq $280, %rsp # imm = 0x118
; SSE-NEXT:    movdqa 64(%rdi), %xmm10
; SSE-NEXT:    movdqa %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 80(%rdi), %xmm9
; SSE-NEXT:    movdqa (%rdi), %xmm14
; SSE-NEXT:    movdqa 16(%rdi), %xmm2
; SSE-NEXT:    movdqa 32(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 48(%rdi), %xmm5
; SSE-NEXT:    movdqa {{.*#+}} xmm7 = [65535,65535,0,65535,65535,0,65535,65535]
; SSE-NEXT:    movdqa %xmm7, %xmm0
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm12 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    movdqa %xmm12, %xmm1
; SSE-NEXT:    pandn %xmm5, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    pandn %xmm5, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm5, %xmm15
; SSE-NEXT:    pand %xmm7, %xmm15
; SSE-NEXT:    por %xmm0, %xmm15
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm15, %xmm0
; SSE-NEXT:    pand %xmm1, %xmm0
; SSE-NEXT:    movdqa %xmm1, %xmm3
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,5]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [65535,65535,65535,0,0,0,65535,65535]
; SSE-NEXT:    movdqa %xmm12, %xmm1
; SSE-NEXT:    pandn %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm14, %xmm11
; SSE-NEXT:    pand %xmm12, %xmm11
; SSE-NEXT:    por %xmm1, %xmm11
; SSE-NEXT:    movdqa %xmm11, %xmm1
; SSE-NEXT:    pand %xmm3, %xmm1
; SSE-NEXT:    movdqa %xmm3, %xmm6
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,3,2,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,7,6,7]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    pand %xmm4, %xmm1
; SSE-NEXT:    movdqa %xmm4, %xmm3
; SSE-NEXT:    pandn %xmm0, %xmm3
; SSE-NEXT:    por %xmm3, %xmm1
; SSE-NEXT:    movdqa %xmm7, %xmm0
; SSE-NEXT:    pandn %xmm9, %xmm0
; SSE-NEXT:    pand %xmm7, %xmm10
; SSE-NEXT:    por %xmm0, %xmm10
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    pand %xmm6, %xmm0
; SSE-NEXT:    movdqa %xmm6, %xmm8
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,3,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,5]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm5 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; SSE-NEXT:    movdqa %xmm5, %xmm3
; SSE-NEXT:    pandn %xmm0, %xmm3
; SSE-NEXT:    pand %xmm5, %xmm1
; SSE-NEXT:    por %xmm1, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 128(%rdi), %xmm13
; SSE-NEXT:    movdqa %xmm7, %xmm0
; SSE-NEXT:    pandn %xmm13, %xmm0
; SSE-NEXT:    movdqa 144(%rdi), %xmm6
; SSE-NEXT:    movdqa %xmm12, %xmm3
; SSE-NEXT:    pandn %xmm6, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm3
; SSE-NEXT:    pandn %xmm6, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm7, %xmm6
; SSE-NEXT:    por %xmm0, %xmm6
; SSE-NEXT:    movdqa %xmm6, %xmm0
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,5]
; SSE-NEXT:    packuswb %xmm3, %xmm0
; SSE-NEXT:    movdqa %xmm4, %xmm8
; SSE-NEXT:    pandn %xmm0, %xmm8
; SSE-NEXT:    movdqa %xmm7, %xmm0
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    pandn %xmm14, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 112(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm12, %xmm4
; SSE-NEXT:    pandn %xmm1, %xmm4
; SSE-NEXT:    movdqa 160(%rdi), %xmm7
; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm7
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    pandn %xmm2, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm14
; SSE-NEXT:    pandn %xmm9, %xmm12
; SSE-NEXT:    movdqa %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm9
; SSE-NEXT:    movdqa %xmm9, (%rsp) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    pandn %xmm1, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 96(%rdi), %xmm3
; SSE-NEXT:    movdqa %xmm3, %xmm5
; SSE-NEXT:    pand %xmm0, %xmm5
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 176(%rdi), %xmm5
; SSE-NEXT:    movdqa %xmm5, %xmm12
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm12
; SSE-NEXT:    movdqa %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm9, %xmm12
; SSE-NEXT:    pand %xmm0, %xmm9
; SSE-NEXT:    movdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm13, %xmm2
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, %xmm2
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pandn %xmm3, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm3, %xmm2
; SSE-NEXT:    pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE-NEXT:    por %xmm4, %xmm2
; SSE-NEXT:    movdqa %xmm2, %xmm4
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm1, %xmm4
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm4[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,3,2,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm4[0,1,2,3,4,7,6,7]
; SSE-NEXT:    packuswb %xmm4, %xmm4
; SSE-NEXT:    pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm4
; SSE-NEXT:    por %xmm8, %xmm4
; SSE-NEXT:    pandn %xmm5, %xmm0
; SSE-NEXT:    por %xmm0, %xmm7
; SSE-NEXT:    movdqa %xmm7, %xmm0
; SSE-NEXT:    pand %xmm1, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,3,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,5]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; SSE-NEXT:    movdqa %xmm3, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm3, %xmm4
; SSE-NEXT:    por %xmm4, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pxor %xmm9, %xmm9
; SSE-NEXT:    movdqa %xmm15, %xmm0
; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm9[8],xmm0[9],xmm9[9],xmm0[10],xmm9[10],xmm0[11],xmm9[11],xmm0[12],xmm9[12],xmm0[13],xmm9[13],xmm0[14],xmm9[14],xmm0[15],xmm9[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm9[0],xmm15[1],xmm9[1],xmm15[2],xmm9[2],xmm15[3],xmm9[3],xmm15[4],xmm9[4],xmm15[5],xmm9[5],xmm15[6],xmm9[6],xmm15[7],xmm9[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm15[2,2,3,3]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1],xmm5[2],xmm0[2],xmm5[3],xmm0[3]
; SSE-NEXT:    psrld $16, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm15 = xmm15[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm15 = xmm15[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm15 = xmm15[2],xmm0[2],xmm15[3],xmm0[3]
; SSE-NEXT:    packuswb %xmm15, %xmm5
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [65535,65535,65535,0,0,0,65535,65535]
; SSE-NEXT:    movdqa %xmm4, %xmm1
; SSE-NEXT:    pandn %xmm5, %xmm1
; SSE-NEXT:    movdqa %xmm11, %xmm5
; SSE-NEXT:    punpckhbw {{.*#+}} xmm5 = xmm5[8],xmm9[8],xmm5[9],xmm9[9],xmm5[10],xmm9[10],xmm5[11],xmm9[11],xmm5[12],xmm9[12],xmm5[13],xmm9[13],xmm5[14],xmm9[14],xmm5[15],xmm9[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[1,1,1,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,5,7,6,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm15 = [65535,65535,0,65535,0,0,65535,65535]
; SSE-NEXT:    movdqa %xmm15, %xmm0
; SSE-NEXT:    pandn %xmm5, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm11 = xmm11[0],xmm9[0],xmm11[1],xmm9[1],xmm11[2],xmm9[2],xmm11[3],xmm9[3],xmm11[4],xmm9[4],xmm11[5],xmm9[5],xmm11[6],xmm9[6],xmm11[7],xmm9[7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm11[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[1,3,2,0,4,5,6,7]
; SSE-NEXT:    pand %xmm15, %xmm5
; SSE-NEXT:    por %xmm0, %xmm5
; SSE-NEXT:    packuswb %xmm5, %xmm5
; SSE-NEXT:    pand %xmm4, %xmm5
; SSE-NEXT:    por %xmm1, %xmm5
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1],xmm0[2],xmm9[2],xmm0[3],xmm9[3],xmm0[4],xmm9[4],xmm0[5],xmm9[5],xmm0[6],xmm9[6],xmm0[7],xmm9[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa {{.*#+}} xmm8 = [65535,65535,65535,0,65535,65535,0,65535]
; SSE-NEXT:    movdqa %xmm8, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    punpckhbw {{.*#+}} xmm10 = xmm10[8],xmm9[8],xmm10[9],xmm9[9],xmm10[10],xmm9[10],xmm10[11],xmm9[11],xmm10[12],xmm9[12],xmm10[13],xmm9[13],xmm10[14],xmm9[14],xmm10[15],xmm9[15]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm10[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,7,6,4]
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa %xmm3, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm3, %xmm5
; SSE-NEXT:    por %xmm5, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm6, %xmm0
; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm9[8],xmm0[9],xmm9[9],xmm0[10],xmm9[10],xmm0[11],xmm9[11],xmm0[12],xmm9[12],xmm0[13],xmm9[13],xmm0[14],xmm9[14],xmm0[15],xmm9[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm9[0],xmm6[1],xmm9[1],xmm6[2],xmm9[2],xmm6[3],xmm9[3],xmm6[4],xmm9[4],xmm6[5],xmm9[5],xmm6[6],xmm9[6],xmm6[7],xmm9[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[2,2,3,3]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE-NEXT:    psrld $16, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm6[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm5 = xmm5[2],xmm0[2],xmm5[3],xmm0[3]
; SSE-NEXT:    packuswb %xmm5, %xmm1
; SSE-NEXT:    movdqa %xmm2, %xmm0
; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm9[8],xmm0[9],xmm9[9],xmm0[10],xmm9[10],xmm0[11],xmm9[11],xmm0[12],xmm9[12],xmm0[13],xmm9[13],xmm0[14],xmm9[14],xmm0[15],xmm9[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,1,1,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm9[0],xmm2[1],xmm9[1],xmm2[2],xmm9[2],xmm2[3],xmm9[3],xmm2[4],xmm9[4],xmm2[5],xmm9[5],xmm2[6],xmm9[6],xmm2[7],xmm9[7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[1,3,2,0,4,5,6,7]
; SSE-NEXT:    pand %xmm15, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm15
; SSE-NEXT:    por %xmm2, %xmm15
; SSE-NEXT:    packuswb %xmm15, %xmm15
; SSE-NEXT:    movdqa %xmm4, %xmm0
; SSE-NEXT:    pand %xmm4, %xmm15
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    por %xmm0, %xmm15
; SSE-NEXT:    movdqa %xmm7, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1],xmm0[2],xmm9[2],xmm0[3],xmm9[3],xmm0[4],xmm9[4],xmm0[5],xmm9[5],xmm0[6],xmm9[6],xmm0[7],xmm9[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,5,5]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm7 = xmm7[8],xmm9[8],xmm7[9],xmm9[9],xmm7[10],xmm9[10],xmm7[11],xmm9[11],xmm7[12],xmm9[12],xmm7[13],xmm9[13],xmm7[14],xmm9[14],xmm7[15],xmm9[15]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm7[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,7,6,4]
; SSE-NEXT:    pand %xmm8, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm8
; SSE-NEXT:    por %xmm1, %xmm8
; SSE-NEXT:    packuswb %xmm8, %xmm0
; SSE-NEXT:    movdqa %xmm3, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm3, %xmm15
; SSE-NEXT:    movdqa %xmm3, %xmm8
; SSE-NEXT:    por %xmm15, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    pand %xmm4, %xmm12
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm12, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm11 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,3,4,5,6,7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm0[0,1,2,3,4,4,5,6]
; SSE-NEXT:    packuswb %xmm1, %xmm2
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm14, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,4,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[1,2,3,0,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm0[0,1,2,3,5,5,5,5]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    pandn %xmm1, %xmm3
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    por %xmm2, %xmm3
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    movdqa (%rsp), %xmm15 # 16-byte Reload
; SSE-NEXT:    por %xmm1, %xmm15
; SSE-NEXT:    movdqa %xmm15, %xmm1
; SSE-NEXT:    pand %xmm11, %xmm1
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,2,2,2,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,6,7,4]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa %xmm8, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    pand %xmm8, %xmm3
; SSE-NEXT:    por %xmm3, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm4, %xmm13
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm13, %xmm1
; SSE-NEXT:    pand %xmm11, %xmm1
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm1[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[1,2,3,3,4,5,6,7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,5,6]
; SSE-NEXT:    packuswb %xmm2, %xmm1
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm4, %xmm2
; SSE-NEXT:    pand %xmm11, %xmm2
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,4,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[1,2,3,0,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,5,5,5,5]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    pandn %xmm2, %xmm3
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    por %xmm1, %xmm3
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    por %xmm1, %xmm10
; SSE-NEXT:    movdqa %xmm10, %xmm1
; SSE-NEXT:    pand %xmm11, %xmm1
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,2,2,2,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,6,7,4]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa %xmm8, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    pand %xmm8, %xmm3
; SSE-NEXT:    por %xmm3, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm12, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm9[0],xmm1[1],xmm9[1],xmm1[2],xmm9[2],xmm1[3],xmm9[3],xmm1[4],xmm9[4],xmm1[5],xmm9[5],xmm1[6],xmm9[6],xmm1[7],xmm9[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm12 = xmm12[8],xmm9[8],xmm12[9],xmm9[9],xmm12[10],xmm9[10],xmm12[11],xmm9[11],xmm12[12],xmm9[12],xmm12[13],xmm9[13],xmm12[14],xmm9[14],xmm12[15],xmm9[15]
; SSE-NEXT:    movdqa %xmm12, %xmm3
; SSE-NEXT:    shufps {{.*#+}} xmm3 = xmm3[2,0],xmm1[3,0]
; SSE-NEXT:    movaps %xmm1, %xmm5
; SSE-NEXT:    shufps {{.*#+}} xmm5 = xmm5[0,1],xmm3[0,2]
; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm12[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm12[2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm5[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,1,0,2]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,1,3,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm1, %xmm3
; SSE-NEXT:    movdqa %xmm14, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm9[0],xmm1[1],xmm9[1],xmm1[2],xmm9[2],xmm1[3],xmm9[3],xmm1[4],xmm9[4],xmm1[5],xmm9[5],xmm1[6],xmm9[6],xmm1[7],xmm9[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm1[3,1,2,1,4,5,6,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [0,65535,65535,0,65535,65535,65535,65535]
; SSE-NEXT:    movdqa %xmm1, %xmm6
; SSE-NEXT:    pandn %xmm5, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm14 = xmm14[8],xmm9[8],xmm14[9],xmm9[9],xmm14[10],xmm9[10],xmm14[11],xmm9[11],xmm14[12],xmm9[12],xmm14[13],xmm9[13],xmm14[14],xmm9[14],xmm14[15],xmm9[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm14[0,3,2,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[0,1,3,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,7,7,7,7]
; SSE-NEXT:    pand %xmm1, %xmm5
; SSE-NEXT:    por %xmm6, %xmm5
; SSE-NEXT:    packuswb %xmm5, %xmm5
; SSE-NEXT:    movdqa %xmm0, %xmm6
; SSE-NEXT:    pandn %xmm5, %xmm6
; SSE-NEXT:    pand %xmm0, %xmm3
; SSE-NEXT:    por %xmm3, %xmm6
; SSE-NEXT:    movdqa %xmm15, %xmm3
; SSE-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm9[8],xmm3[9],xmm9[9],xmm3[10],xmm9[10],xmm3[11],xmm9[11],xmm3[12],xmm9[12],xmm3[13],xmm9[13],xmm3[14],xmm9[14],xmm3[15],xmm9[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm3[0,1,2,3,7,5,6,5]
; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [65535,65535,65535,65535,0,65535,65535,0]
; SSE-NEXT:    movdqa %xmm3, %xmm7
; SSE-NEXT:    pandn %xmm5, %xmm7
; SSE-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm9[0],xmm15[1],xmm9[1],xmm15[2],xmm9[2],xmm15[3],xmm9[3],xmm15[4],xmm9[4],xmm15[5],xmm9[5],xmm15[6],xmm9[6],xmm15[7],xmm9[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm15[0,2,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,4,5,7,7]
; SSE-NEXT:    pand %xmm3, %xmm5
; SSE-NEXT:    por %xmm7, %xmm5
; SSE-NEXT:    packuswb %xmm5, %xmm5
; SSE-NEXT:    movdqa %xmm8, %xmm15
; SSE-NEXT:    pandn %xmm5, %xmm15
; SSE-NEXT:    pand %xmm8, %xmm6
; SSE-NEXT:    por %xmm6, %xmm15
; SSE-NEXT:    movdqa %xmm13, %xmm5
; SSE-NEXT:    pxor %xmm2, %xmm2
; SSE-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm2[0],xmm5[1],xmm2[1],xmm5[2],xmm2[2],xmm5[3],xmm2[3],xmm5[4],xmm2[4],xmm5[5],xmm2[5],xmm5[6],xmm2[6],xmm5[7],xmm2[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm13 = xmm13[8],xmm2[8],xmm13[9],xmm2[9],xmm13[10],xmm2[10],xmm13[11],xmm2[11],xmm13[12],xmm2[12],xmm13[13],xmm2[13],xmm13[14],xmm2[14],xmm13[15],xmm2[15]
; SSE-NEXT:    movdqa %xmm13, %xmm6
; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[2,0],xmm5[3,0]
; SSE-NEXT:    movaps %xmm5, %xmm7
; SSE-NEXT:    shufps {{.*#+}} xmm7 = xmm7[0,1],xmm6[0,2]
; SSE-NEXT:    shufps {{.*#+}} xmm5 = xmm5[1,0],xmm13[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm5 = xmm5[2,0],xmm13[2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm7[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,1,0,2]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[0,1,3,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm5, %xmm6
; SSE-NEXT:    movdqa %xmm4, %xmm5
; SSE-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm2[0],xmm5[1],xmm2[1],xmm5[2],xmm2[2],xmm5[3],xmm2[3],xmm5[4],xmm2[4],xmm5[5],xmm2[5],xmm5[6],xmm2[6],xmm5[7],xmm2[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[3,1,2,1,4,5,6,7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm2[8],xmm4[9],xmm2[9],xmm4[10],xmm2[10],xmm4[11],xmm2[11],xmm4[12],xmm2[12],xmm4[13],xmm2[13],xmm4[14],xmm2[14],xmm4[15],xmm2[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm7 = xmm4[0,3,2,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm7 = xmm7[0,1,3,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm7 = xmm7[0,1,2,3,7,7,7,7]
; SSE-NEXT:    pand %xmm1, %xmm7
; SSE-NEXT:    pandn %xmm5, %xmm1
; SSE-NEXT:    por %xmm7, %xmm1
; SSE-NEXT:    pand %xmm0, %xmm6
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    por %xmm6, %xmm0
; SSE-NEXT:    movdqa %xmm10, %xmm1
; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm2[8],xmm1[9],xmm2[9],xmm1[10],xmm2[10],xmm1[11],xmm2[11],xmm1[12],xmm2[12],xmm1[13],xmm2[13],xmm1[14],xmm2[14],xmm1[15],xmm2[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,7,5,6,5]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm2[0],xmm10[1],xmm2[1],xmm10[2],xmm2[2],xmm10[3],xmm2[3],xmm10[4],xmm2[4],xmm10[5],xmm2[5],xmm10[6],xmm2[6],xmm10[7],xmm2[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm10[0,2,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,4,5,7,7]
; SSE-NEXT:    pand %xmm3, %xmm5
; SSE-NEXT:    pandn %xmm1, %xmm3
; SSE-NEXT:    por %xmm5, %xmm3
; SSE-NEXT:    movdqa %xmm8, %xmm4
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    packuswb %xmm3, %xmm1
; SSE-NEXT:    pandn %xmm1, %xmm4
; SSE-NEXT:    por %xmm0, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm9, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm5 = [255,255,255,255,255,0,0,0,0,0,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm5, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm0[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm3, %xmm3
; SSE-NEXT:    pand %xmm5, %xmm3
; SSE-NEXT:    por %xmm1, %xmm3
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; SSE-NEXT:    movdqa {{.*#+}} xmm13 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    pand %xmm13, %xmm12
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm12, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,1,2,1,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,1,0,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,4,7]
; SSE-NEXT:    packuswb %xmm0, %xmm6
; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [65535,65535,65535,65535,65535,0,0,0]
; SSE-NEXT:    movdqa %xmm0, %xmm8
; SSE-NEXT:    pandn %xmm6, %xmm8
; SSE-NEXT:    pand %xmm0, %xmm3
; SSE-NEXT:    por %xmm3, %xmm8
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm14, %xmm3
; SSE-NEXT:    pand %xmm11, %xmm3
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm3[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm6[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,1,0,2]
; SSE-NEXT:    packuswb %xmm6, %xmm3
; SSE-NEXT:    movdqa %xmm5, %xmm6
; SSE-NEXT:    pandn %xmm3, %xmm6
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm1, %xmm3
; SSE-NEXT:    pand %xmm11, %xmm3
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm7 = xmm3[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm7, %xmm7
; SSE-NEXT:    pand %xmm5, %xmm7
; SSE-NEXT:    por %xmm6, %xmm7
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    pand %xmm13, %xmm3
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Folded Reload
; SSE-NEXT:    por %xmm3, %xmm13
; SSE-NEXT:    movdqa %xmm11, %xmm3
; SSE-NEXT:    pand %xmm13, %xmm3
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,2,1,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,0,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,6,5,4,7]
; SSE-NEXT:    packuswb %xmm3, %xmm6
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    pandn %xmm6, %xmm3
; SSE-NEXT:    pand %xmm0, %xmm7
; SSE-NEXT:    por %xmm7, %xmm3
; SSE-NEXT:    movdqa %xmm9, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm2[8],xmm6[9],xmm2[9],xmm6[10],xmm2[10],xmm6[11],xmm2[11],xmm6[12],xmm2[12],xmm6[13],xmm2[13],xmm6[14],xmm2[14],xmm6[15],xmm2[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm9 = xmm9[0],xmm2[0],xmm9[1],xmm2[1],xmm9[2],xmm2[2],xmm9[3],xmm2[3],xmm9[4],xmm2[4],xmm9[5],xmm2[5],xmm9[6],xmm2[6],xmm9[7],xmm2[7]
; SSE-NEXT:    movdqa %xmm9, %xmm7
; SSE-NEXT:    shufps {{.*#+}} xmm9 = xmm9[1,0],xmm6[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm9 = xmm9[2,0],xmm6[2,3]
; SSE-NEXT:    psrlq $48, %xmm6
; SSE-NEXT:    psrldq {{.*#+}} xmm7 = xmm7[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm9[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,4,4,5,7]
; SSE-NEXT:    packuswb %xmm7, %xmm6
; SSE-NEXT:    movdqa %xmm5, %xmm7
; SSE-NEXT:    pandn %xmm6, %xmm7
; SSE-NEXT:    movdqa %xmm10, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm2[8],xmm6[9],xmm2[9],xmm6[10],xmm2[10],xmm6[11],xmm2[11],xmm6[12],xmm2[12],xmm6[13],xmm2[13],xmm6[14],xmm2[14],xmm6[15],xmm2[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm6[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa {{.*#+}} xmm6 = [65535,0,65535,65535,0,65535,65535,65535]
; SSE-NEXT:    movdqa %xmm6, %xmm11
; SSE-NEXT:    pandn %xmm9, %xmm11
; SSE-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm2[0],xmm10[1],xmm2[1],xmm10[2],xmm2[2],xmm10[3],xmm2[3],xmm10[4],xmm2[4],xmm10[5],xmm2[5],xmm10[6],xmm2[6],xmm10[7],xmm2[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm10[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm10 = xmm9[3,1,1,2,4,5,6,7]
; SSE-NEXT:    pand %xmm6, %xmm10
; SSE-NEXT:    por %xmm11, %xmm10
; SSE-NEXT:    packuswb %xmm10, %xmm10
; SSE-NEXT:    pand %xmm5, %xmm10
; SSE-NEXT:    por %xmm7, %xmm10
; SSE-NEXT:    movdqa %xmm12, %xmm7
; SSE-NEXT:    punpckhbw {{.*#+}} xmm12 = xmm12[8],xmm2[8],xmm12[9],xmm2[9],xmm12[10],xmm2[10],xmm12[11],xmm2[11],xmm12[12],xmm2[12],xmm12[13],xmm2[13],xmm12[14],xmm2[14],xmm12[15],xmm2[15]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm12[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[0,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm11 = xmm9[0,1,2,3,5,5,7,4]
; SSE-NEXT:    movdqa {{.*#+}} xmm9 = [65535,65535,65535,65535,0,65535,0,0]
; SSE-NEXT:    movdqa %xmm9, %xmm12
; SSE-NEXT:    pandn %xmm11, %xmm12
; SSE-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm2[0],xmm7[1],xmm2[1],xmm7[2],xmm2[2],xmm7[3],xmm2[3],xmm7[4],xmm2[4],xmm7[5],xmm2[5],xmm7[6],xmm2[6],xmm7[7],xmm2[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm7 = xmm7[0,3,1,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm7 = xmm7[0,1,1,3,4,5,6,7]
; SSE-NEXT:    pand %xmm9, %xmm7
; SSE-NEXT:    por %xmm7, %xmm12
; SSE-NEXT:    packuswb %xmm12, %xmm11
; SSE-NEXT:    movdqa %xmm0, %xmm7
; SSE-NEXT:    pandn %xmm11, %xmm7
; SSE-NEXT:    pand %xmm0, %xmm10
; SSE-NEXT:    por %xmm10, %xmm7
; SSE-NEXT:    movdqa %xmm14, %xmm10
; SSE-NEXT:    punpckhbw {{.*#+}} xmm10 = xmm10[8],xmm2[8],xmm10[9],xmm2[9],xmm10[10],xmm2[10],xmm10[11],xmm2[11],xmm10[12],xmm2[12],xmm10[13],xmm2[13],xmm10[14],xmm2[14],xmm10[15],xmm2[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm14 = xmm14[0],xmm2[0],xmm14[1],xmm2[1],xmm14[2],xmm2[2],xmm14[3],xmm2[3],xmm14[4],xmm2[4],xmm14[5],xmm2[5],xmm14[6],xmm2[6],xmm14[7],xmm2[7]
; SSE-NEXT:    movdqa %xmm14, %xmm11
; SSE-NEXT:    shufps {{.*#+}} xmm14 = xmm14[1,0],xmm10[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm14 = xmm14[2,0],xmm10[2,3]
; SSE-NEXT:    psrlq $48, %xmm10
; SSE-NEXT:    psrldq {{.*#+}} xmm11 = xmm11[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE-NEXT:    punpcklwd {{.*#+}} xmm11 = xmm11[0],xmm10[0],xmm11[1],xmm10[1],xmm11[2],xmm10[2],xmm11[3],xmm10[3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm10 = xmm14[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm10 = xmm10[0,1,2,3,4,4,5,7]
; SSE-NEXT:    packuswb %xmm11, %xmm10
; SSE-NEXT:    movdqa %xmm1, %xmm11
; SSE-NEXT:    punpckhbw {{.*#+}} xmm11 = xmm11[8],xmm2[8],xmm11[9],xmm2[9],xmm11[10],xmm2[10],xmm11[11],xmm2[11],xmm11[12],xmm2[12],xmm11[13],xmm2[13],xmm11[14],xmm2[14],xmm11[15],xmm2[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm11 = xmm11[1,1,2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm11 = xmm11[0,1,2,3,5,5,5,5]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm12 = xmm1[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm12 = xmm12[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm12 = xmm12[3,1,1,2,4,5,6,7]
; SSE-NEXT:    pand %xmm6, %xmm12
; SSE-NEXT:    pandn %xmm11, %xmm6
; SSE-NEXT:    por %xmm12, %xmm6
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    pand %xmm5, %xmm6
; SSE-NEXT:    pandn %xmm10, %xmm5
; SSE-NEXT:    por %xmm5, %xmm6
; SSE-NEXT:    movdqa %xmm13, %xmm5
; SSE-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm2[0],xmm5[1],xmm2[1],xmm5[2],xmm2[2],xmm5[3],xmm2[3],xmm5[4],xmm2[4],xmm5[5],xmm2[5],xmm5[6],xmm2[6],xmm5[7],xmm2[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm13 = xmm13[8],xmm2[8],xmm13[9],xmm2[9],xmm13[10],xmm2[10],xmm13[11],xmm2[11],xmm13[12],xmm2[12],xmm13[13],xmm2[13],xmm13[14],xmm2[14],xmm13[15],xmm2[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm5[0,3,1,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,1,1,3,4,5,6,7]
; SSE-NEXT:    pand %xmm9, %xmm4
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm13[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,5,5,7,4]
; SSE-NEXT:    pandn %xmm5, %xmm9
; SSE-NEXT:    por %xmm4, %xmm9
; SSE-NEXT:    pand %xmm0, %xmm6
; SSE-NEXT:    packuswb %xmm9, %xmm4
; SSE-NEXT:    pandn %xmm4, %xmm0
; SSE-NEXT:    por %xmm6, %xmm0
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    movaps %xmm4, 16(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    movaps %xmm4, (%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 16(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, (%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 16(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, (%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 16(%r8)
; SSE-NEXT:    movdqa %xmm15, (%r8)
; SSE-NEXT:    movdqa %xmm3, 16(%r9)
; SSE-NEXT:    movdqa %xmm8, (%r9)
; SSE-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT:    movdqa %xmm0, 16(%rax)
; SSE-NEXT:    movdqa %xmm7, (%rax)
; SSE-NEXT:    addq $280, %rsp # imm = 0x118
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i8_stride6_vf32:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    subq $168, %rsp
; AVX1-ONLY-NEXT:    vmovdqa 128(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm1[u,u,4,10,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, %xmm4
; AVX1-ONLY-NEXT:    vmovdqa 144(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm2[u,u,u,u,0,6,12,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm5
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; AVX1-ONLY-NEXT:    vmovdqa 176(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm2[4,10]
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm6
; AVX1-ONLY-NEXT:    vmovdqa 160(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm3[u,u,u,u,u,u,u,u,u,u,u,2,8,14],zero,zero
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm7
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm2, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,u,u,255,255,255,255,255,0,0,0,0,0>
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, (%rsp) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm4[u,u,5,11,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm5[u,u,u,u,1,7,13,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm6[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm6[5,11]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[u,u,u,u,u,u,u,u,u,u,u,3,9,15],zero,zero
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 32(%rdi), %xmm14
; AVX1-ONLY-NEXT:    vmovdqa 48(%rdi), %xmm15
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm0 = [2,8,14,0,2,8,14,0,2,8,14,0,2,8,14,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm15, %xmm2
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm1 = [0,0,6,12,0,0,6,12,0,0,6,12,0,0,6,12]
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm14, %xmm3
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0]
; AVX1-ONLY-NEXT:    vmovdqa (%rdi), %xmm9
; AVX1-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm8
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = zero,zero,zero,xmm8[4,10,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm9[2,8,14],zero,zero,xmm9[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm10, %xmm10
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,0,0,0,0,255,255,255,255,255,255,u,u,u,u,u>
; AVX1-ONLY-NEXT:    vpblendvb %xmm2, %xmm3, %xmm10, %xmm3
; AVX1-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm3 = [3,9,15,0,3,9,15,0,3,9,15,0,3,9,15,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm15, %xmm10
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm11 = [0,1,7,13,0,1,7,13,0,1,7,13,0,1,7,13]
; AVX1-ONLY-NEXT:    vpshufb %xmm11, %xmm14, %xmm12
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm10 = xmm12[0],xmm10[0]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = zero,zero,zero,xmm8[5,11,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm9[3,9,15],zero,zero,xmm9[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm12, %xmm13, %xmm12
; AVX1-ONLY-NEXT:    vpblendvb %xmm2, %xmm10, %xmm12, %xmm2
; AVX1-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm5, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm4, %xmm1
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX1-ONLY-NEXT:    vmovdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm7[u,u,u,u,u,u,u,u,u,u,u,4,10],zero,zero,zero
; AVX1-ONLY-NEXT:    vmovdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm6[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm6[0,6,12]
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm2 = <u,u,u,u,u,255,255,255,255,255,255,0,0,0,0,0>
; AVX1-ONLY-NEXT:    vpblendvb %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm5, %xmm0
; AVX1-ONLY-NEXT:    vpshufb %xmm11, %xmm4, %xmm1
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm7[u,u,u,u,u,u,u,u,u,u,u,5,11],zero,zero,zero
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = xmm6[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm6[1,7,13]
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm3, %xmm1
; AVX1-ONLY-NEXT:    vpblendvb %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm9[4,10],zero,zero,zero,xmm9[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = zero,zero,xmm8[0,6,12,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm1
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm15[u,u,u,u,u,u,u,u,4,10,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm14[u,u,u,u,u,u,u,u,u,u,u,u,u,2,8,14]
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm2 = xmm2[1],xmm0[1]
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <255,255,255,255,255,0,0,0,0,0,u,u,u,u,u,u>
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm9[5,11],zero,zero,zero,xmm9[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = zero,zero,xmm8[1,7,13,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm15[u,u,u,u,u,u,u,u,5,11,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = xmm14[u,u,u,u,u,u,u,u,u,u,u,u,u,3,9,15]
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm2 = xmm3[1],xmm2[1]
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm0
; AVX1-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm10 = [128,128,128,2,8,14,0,0,128,128,128,2,8,14,0,0]
; AVX1-ONLY-NEXT:    # xmm10 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa 112(%rdi), %xmm5
; AVX1-ONLY-NEXT:    vpshufb %xmm10, %xmm5, %xmm1
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm6 = [0,6,12,128,128,128,0,0,0,6,12,128,128,128,0,0]
; AVX1-ONLY-NEXT:    # xmm6 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa 96(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa 80(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm2[4,10]
; AVX1-ONLY-NEXT:    vmovdqa 64(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm4[u,u,u,u,u,u,u,u,u,u,u,2,8,14],zero,zero
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm13, %xmm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm14[u,u,4,10,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm15[u,u,u,u,0,6,12,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm13[0],xmm1[0],xmm13[1],xmm1[1]
; AVX1-ONLY-NEXT:    vpshufb %xmm10, %xmm8, %xmm10
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm9, %xmm6
; AVX1-ONLY-NEXT:    vpor %xmm6, %xmm10, %xmm6
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm1 = xmm6[0,1,2],xmm1[3,4,5],xmm6[6,7]
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm6 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255]
; AVX1-ONLY-NEXT:    vandnps %ymm0, %ymm6, %ymm0
; AVX1-ONLY-NEXT:    vandps %ymm6, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm0, %ymm1, %ymm0
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm1 = [65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,0,0,0,0,0]
; AVX1-ONLY-NEXT:    vandps %ymm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, (%rsp), %ymm0, %ymm10 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm10, %ymm1, %ymm10
; AVX1-ONLY-NEXT:    vorps %ymm0, %ymm10, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, (%rsp) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm7 = [128,128,128,3,9,15,0,0,128,128,128,3,9,15,0,0]
; AVX1-ONLY-NEXT:    # xmm7 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm5, %xmm13
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm0 = [1,7,13,128,128,128,0,0,1,7,13,128,128,128,0,0]
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm3, %xmm12
; AVX1-ONLY-NEXT:    vpor %xmm13, %xmm12, %xmm12
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,zero,xmm2[5,11]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = xmm4[u,u,u,u,u,u,u,u,u,u,u,3,9,15],zero,zero
; AVX1-ONLY-NEXT:    vpor %xmm13, %xmm11, %xmm11
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm12, %ymm11, %ymm11
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = xmm14[u,u,5,11,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm15[u,u,u,u,1,7,13,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm12 = xmm13[0],xmm12[0],xmm13[1],xmm12[1]
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm8, %xmm7
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm9, %xmm0
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm0, %xmm0
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm12[3,4,5],xmm0[6,7]
; AVX1-ONLY-NEXT:    vandnps %ymm11, %ymm6, %ymm7
; AVX1-ONLY-NEXT:    vandps %ymm6, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vorps %ymm7, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vandps %ymm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm6 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm6, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = zero,zero,zero,xmm5[4,10,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm3[2,8,14],zero,zero,xmm3[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm6, %xmm1
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm4[u,u,u,u,u,u,u,u,u,u,u,4,10],zero,zero,zero
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm2[0,6,12]
; AVX1-ONLY-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm6, %ymm1
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm6 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255]
; AVX1-ONLY-NEXT:    vandnps %ymm1, %ymm6, %ymm1
; AVX1-ONLY-NEXT:    vandps {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm7, %ymm7
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm10 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX1-ONLY-NEXT:    vandps %ymm7, %ymm10, %ymm7
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm8 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm8, %ymm10, %ymm8
; AVX1-ONLY-NEXT:    vorps %ymm7, %ymm8, %ymm13
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = zero,zero,zero,xmm5[5,11,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm3[3,9,15],zero,zero,xmm3[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm8, %xmm7
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm4[u,u,u,u,u,u,u,u,u,u,u,5,11],zero,zero,zero
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm9 = xmm2[u,u,u,u,u,u,u,u,u,u,u],zero,zero,xmm2[1,7,13]
; AVX1-ONLY-NEXT:    vpor %xmm8, %xmm9, %xmm8
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm7, %ymm8, %ymm7
; AVX1-ONLY-NEXT:    vandnps %ymm7, %ymm6, %ymm7
; AVX1-ONLY-NEXT:    vandps {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm6 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vorps %ymm7, %ymm6, %ymm6
; AVX1-ONLY-NEXT:    vandps %ymm6, %ymm10, %ymm6
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm7 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm7, %ymm10, %ymm7
; AVX1-ONLY-NEXT:    vorps %ymm7, %ymm6, %ymm14
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm3[4,10],zero,zero,zero,xmm3[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = zero,zero,xmm5[0,6,12,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm7 = [0,0,128,128,128,2,8,14,0,0,128,128,128,2,8,14]
; AVX1-ONLY-NEXT:    # xmm7 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm2, %xmm8
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm9 = [0,0,0,6,12,128,128,128,0,0,0,6,12,128,128,128]
; AVX1-ONLY-NEXT:    # xmm9 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm9, %xmm4, %xmm11
; AVX1-ONLY-NEXT:    vpor %xmm8, %xmm11, %xmm8
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm6, %ymm8, %ymm6
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm8 = [0,0,0,0,0,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535]
; AVX1-ONLY-NEXT:    vandnps {{[-0-9]+}}(%r{{[sb]}}p), %ymm8, %ymm11 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vandps %ymm6, %ymm8, %ymm6
; AVX1-ONLY-NEXT:    vorps %ymm6, %ymm11, %ymm6
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm1, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm9, %xmm0, %xmm9
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm9, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm9 = xmm15[u,u,u,u,u,u,u,u,4,10,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = xmm12[u,u,u,u,u,u,u,u,u,u,u,u,u,2,8,14]
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm9 = xmm11[1],xmm9[1]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm7 = xmm9[0,1,2,3,4],xmm7[5,6,7]
; AVX1-ONLY-NEXT:    vandps %ymm6, %ymm10, %ymm6
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm7, %ymm0, %ymm7
; AVX1-ONLY-NEXT:    vandnps %ymm7, %ymm10, %ymm7
; AVX1-ONLY-NEXT:    vorps %ymm7, %ymm6, %ymm6
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[5,11],zero,zero,zero,xmm3[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,xmm5[1,7,13,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm3, %xmm5, %xmm3
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm5 = [0,0,128,128,128,3,9,15,0,0,128,128,128,3,9,15]
; AVX1-ONLY-NEXT:    # xmm5 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm7 = [0,0,1,7,13,128,128,128,0,0,1,7,13,128,128,128]
; AVX1-ONLY-NEXT:    # xmm7 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm4, %xmm4
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm4, %xmm2
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm2
; AVX1-ONLY-NEXT:    vandnps {{[-0-9]+}}(%r{{[sb]}}p), %ymm8, %ymm3 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vandps %ymm2, %ymm8, %ymm2
; AVX1-ONLY-NEXT:    vorps %ymm3, %ymm2, %ymm2
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm1, %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm0, %xmm4
; AVX1-ONLY-NEXT:    vpor %xmm3, %xmm4, %xmm3
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm4 = xmm15[u,u,u,u,u,u,u,u,5,11,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = xmm12[u,u,u,u,u,u,u,u,u,u,u,u,u,3,9,15]
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm4 = xmm5[1],xmm4[1]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm3 = xmm4[0,1,2,3,4],xmm3[5,6,7]
; AVX1-ONLY-NEXT:    vandps %ymm2, %ymm10, %ymm2
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX1-ONLY-NEXT:    vandnps %ymm3, %ymm10, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm2, %ymm1
; AVX1-ONLY-NEXT:    vmovups (%rsp), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%rdx)
; AVX1-ONLY-NEXT:    vmovaps %ymm13, (%rcx)
; AVX1-ONLY-NEXT:    vmovaps %ymm14, (%r8)
; AVX1-ONLY-NEXT:    vmovaps %ymm6, (%r9)
; AVX1-ONLY-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX1-ONLY-NEXT:    vmovaps %ymm1, (%rax)
; AVX1-ONLY-NEXT:    addq $168, %rsp
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i8_stride6_vf32:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    vmovdqa 160(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 96(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm8 = <255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255>
; AVX2-ONLY-NEXT:    vpblendvb %ymm8, %ymm2, %ymm3, %ymm9
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = xmm9[0,6,12],zero,zero,zero,xmm9[4,10],zero,zero,zero,xmm9[u,u,u,u,u]
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm9, %xmm10
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,zero,xmm10[2,8,14],zero,zero,xmm10[0,6,12,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm5, %xmm6, %xmm11
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm7 = <u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0>
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm5 = ymm0[0,1],ymm1[0,1]
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm6 = ymm0[2,3],ymm1[2,3]
; AVX2-ONLY-NEXT:    vpblendvb %ymm7, %ymm5, %ymm6, %ymm1
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm0 = ymm1[u,u,u,u,u,u,u,u,u,u,u,2,8,14,4,10,16,22,28,18,24,30,u,u,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm7 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; AVX2-ONLY-NEXT:    vpblendvb %ymm7, %ymm11, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm9 = xmm9[1,7,13],zero,zero,zero,xmm9[5,11],zero,zero,zero,xmm9[u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = zero,zero,zero,xmm10[3,9,15],zero,zero,xmm10[1,7,13,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm9, %xmm10, %xmm9
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,u,u,u,3,9,15,5,11,17,23,29,19,25,31,u,u,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpblendvb %ymm7, %ymm9, %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm11 = <255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255>
; AVX2-ONLY-NEXT:    vpblendvb %ymm11, %ymm3, %ymm2, %ymm9
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm9, %xmm10
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = zero,zero,zero,xmm10[4,10],zero,zero,zero,xmm10[2,8,14,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm9[2,8,14],zero,zero,xmm9[0,6,12],zero,zero,zero,xmm9[u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm12, %xmm13, %xmm12
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm13 = <255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0>
; AVX2-ONLY-NEXT:    vpblendvb %ymm13, %ymm6, %ymm5, %ymm13
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm14 = ymm13[u,u,u,u,u,u,u,u,u,u,u,4,10,0,6,12,18,24,30,20,26,u,u,u,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpblendvb %ymm7, %ymm12, %ymm14, %ymm12
; AVX2-ONLY-NEXT:    vmovdqa 128(%rdi), %ymm14
; AVX2-ONLY-NEXT:    vpblendvb %ymm8, %ymm14, %ymm4, %ymm8
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = zero,zero,zero,xmm10[5,11],zero,zero,zero,xmm10[3,9,15,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm9 = xmm9[3,9,15],zero,zero,xmm9[1,7,13],zero,zero,zero,xmm9[u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm10, %xmm9, %xmm9
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm8[u,u,u,u,u,0,6,12],zero,zero,zero,xmm8[4,10],zero,zero,zero
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm13 = ymm13[u,u,u,u,u,u,u,u,u,u,u,5,11,1,7,13,19,25,31,21,27,u,u,u,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpblendvb %ymm7, %ymm9, %ymm13, %ymm13
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm8, %xmm9
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm7 = xmm9[u,u,u,u,u],zero,zero,zero,xmm9[2,8,14],zero,zero,xmm9[0,6,12]
; AVX2-ONLY-NEXT:    vpor %xmm7, %xmm10, %xmm7
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm7, %ymm0, %ymm7
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm10 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX2-ONLY-NEXT:    vpblendvb %ymm10, %ymm12, %ymm7, %ymm7
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm12 = <u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u>
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm8 = xmm8[u,u,u,u,u,1,7,13],zero,zero,zero,xmm8[5,11],zero,zero,zero
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm9 = xmm9[u,u,u,u,u],zero,zero,zero,xmm9[3,9,15],zero,zero,xmm9[1,7,13]
; AVX2-ONLY-NEXT:    vpor %xmm8, %xmm9, %xmm8
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm4, %ymm14, %ymm9
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm8, %ymm0, %ymm8
; AVX2-ONLY-NEXT:    vpblendvb %ymm10, %ymm13, %ymm8, %ymm8
; AVX2-ONLY-NEXT:    vpblendvb %ymm11, %ymm4, %ymm14, %ymm4
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm4, %xmm11
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm13 = xmm11[u,u,u,u,u],zero,zero,zero,xmm11[4,10],zero,zero,zero,xmm11[2,8,14]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm14 = xmm4[u,u,u,u,u,2,8,14],zero,zero,xmm4[0,6,12],zero,zero,zero
; AVX2-ONLY-NEXT:    vpor %xmm13, %xmm14, %xmm13
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm3, %ymm2, %ymm2
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm2, %xmm3
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm12 = <0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u>
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm6, %ymm5, %ymm5
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,xmm3[0,6,12],zero,zero,zero,xmm3[4,10,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm12 = xmm2[4,10],zero,zero,zero,xmm2[2,8,14],zero,zero,xmm2[u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm6, %xmm12, %xmm6
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm12 = ymm5[u,u,u,u,u,u,u,u,u,u,0,6,12,2,8,14,20,26,16,22,28,u,u,u,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} xmm6 = xmm6[0,1,2,3,4],xmm12[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm6 = ymm6[0,1,2,3],ymm12[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm13, %ymm0, %ymm12
; AVX2-ONLY-NEXT:    vpblendvb %ymm10, %ymm6, %ymm12, %ymm6
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm11 = xmm11[u,u,u,u,u],zero,zero,zero,xmm11[5,11],zero,zero,zero,xmm11[3,9,15]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm4 = xmm4[u,u,u,u,u,3,9,15],zero,zero,xmm4[1,7,13],zero,zero,zero
; AVX2-ONLY-NEXT:    vpor %xmm4, %xmm11, %xmm4
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = zero,zero,xmm3[1,7,13],zero,zero,zero,xmm3[5,11,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[5,11],zero,zero,zero,xmm2[3,9,15],zero,zero,xmm2[u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm3 = ymm5[u,u,u,u,u,u,u,u,u,u,1,7,13,3,9,15,21,27,17,23,29,u,u,u,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0,1,2,3,4],xmm3[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7]
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm9, %xmm3
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm4, %ymm0, %ymm4
; AVX2-ONLY-NEXT:    vpblendvb %ymm10, %ymm2, %ymm4, %ymm2
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm4 = xmm3[u,u,u,u,u,u],zero,zero,xmm3[0,6,12],zero,zero,zero,xmm3[4,10]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm5 = xmm9[u,u,u,u,u,u,4,10],zero,zero,zero,xmm9[2,8,14],zero,zero
; AVX2-ONLY-NEXT:    vpor %xmm4, %xmm5, %xmm4
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm4, %ymm0, %ymm4
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} ymm4 = ymm0[0,1,2],ymm4[3,4,5,6,7],ymm0[8,9,10],ymm4[11,12,13,14,15]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm4[4,5,6,7]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[u,u,u,u,u,u],zero,zero,xmm3[1,7,13],zero,zero,zero,xmm3[5,11]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} xmm4 = xmm9[u,u,u,u,u,u,5,11],zero,zero,zero,xmm9[3,9,15],zero,zero
; AVX2-ONLY-NEXT:    vpor %xmm3, %xmm4, %xmm3
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} ymm3 = ymm1[0,1,2],ymm3[3,4,5,6,7],ymm1[8,9,10],ymm3[11,12,13,14,15]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, (%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %ymm1, (%rdx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm7, (%rcx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm8, (%r8)
; AVX2-ONLY-NEXT:    vmovdqa %ymm6, (%r9)
; AVX2-ONLY-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, (%rax)
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-LABEL: load_i8_stride6_vf32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm0 = [65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535]
; AVX512F-NEXT:    vmovdqa64 (%rdi), %ymm17
; AVX512F-NEXT:    vmovdqa 32(%rdi), %ymm3
; AVX512F-NEXT:    vmovdqa 64(%rdi), %ymm1
; AVX512F-NEXT:    vmovdqa 128(%rdi), %ymm6
; AVX512F-NEXT:    vmovdqa %ymm0, %ymm7
; AVX512F-NEXT:    vpternlogq $202, %ymm3, %ymm17, %ymm7
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm4 = xmm7[0,6,12],zero,zero,zero,xmm7[4,10],zero,zero,zero,xmm7[u,u,u,u,u]
; AVX512F-NEXT:    vextracti128 $1, %ymm7, %xmm8
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,xmm8[2,8,14],zero,zero,xmm8[0,6,12,u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm4, %xmm5, %xmm4
; AVX512F-NEXT:    vperm2i128 {{.*#+}} ymm5 = ymm1[2,3],mem[2,3]
; AVX512F-NEXT:    vinserti128 $1, 96(%rdi), %ymm1, %ymm1
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm9 = [65535,65535,0,65535,65535,0,65535,65535,65535,0,65535,65535,0,65535,65535,0]
; AVX512F-NEXT:    vmovdqa %ymm9, %ymm10
; AVX512F-NEXT:    vpternlogq $202, %ymm5, %ymm1, %ymm10
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm11 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm10[2,8,14,4,10,16,22,28,18,24,30],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm16 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255]
; AVX512F-NEXT:    vpternlogq $248, %ymm16, %ymm4, %ymm11
; AVX512F-NEXT:    vmovdqa 160(%rdi), %ymm13
; AVX512F-NEXT:    vmovdqa %ymm0, %ymm14
; AVX512F-NEXT:    vpternlogq $202, %ymm6, %ymm13, %ymm14
; AVX512F-NEXT:    vextracti128 $1, %ymm14, %xmm15
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm4 = xmm15[u,u,u,u,u,u],zero,zero,xmm15[0,6,12],zero,zero,zero,xmm15[4,10]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm12 = xmm14[u,u,u,u,u,u,4,10],zero,zero,zero,xmm14[2,8,14],zero,zero
; AVX512F-NEXT:    vpor %xmm4, %xmm12, %xmm4
; AVX512F-NEXT:    vinserti128 $1, %xmm4, %ymm0, %ymm4
; AVX512F-NEXT:    vpblendw {{.*#+}} ymm4 = ymm11[0,1,2],ymm4[3,4,5,6,7],ymm11[8,9,10],ymm4[11,12,13,14,15]
; AVX512F-NEXT:    vpblendd {{.*#+}} ymm2 = ymm11[0,1,2,3],ymm4[4,5,6,7]
; AVX512F-NEXT:    vmovdqa64 %ymm2, %ymm18
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm7 = xmm7[1,7,13],zero,zero,zero,xmm7[5,11],zero,zero,zero,xmm7[u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm8 = zero,zero,zero,xmm8[3,9,15],zero,zero,xmm8[1,7,13,u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm7, %xmm8, %xmm7
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm8 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm10[3,9,15,5,11,17,23,29,19,25,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; AVX512F-NEXT:    vpternlogq $248, %ymm16, %ymm7, %ymm8
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm7 = xmm15[u,u,u,u,u,u],zero,zero,xmm15[1,7,13],zero,zero,zero,xmm15[5,11]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm10 = xmm14[u,u,u,u,u,u,5,11],zero,zero,zero,xmm14[3,9,15],zero,zero
; AVX512F-NEXT:    vpor %xmm7, %xmm10, %xmm7
; AVX512F-NEXT:    vinserti128 $1, %xmm7, %ymm0, %ymm7
; AVX512F-NEXT:    vpblendw {{.*#+}} ymm7 = ymm8[0,1,2],ymm7[3,4,5,6,7],ymm8[8,9,10],ymm7[11,12,13,14,15]
; AVX512F-NEXT:    vpblendd {{.*#+}} ymm7 = ymm8[0,1,2,3],ymm7[4,5,6,7]
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm8 = [65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535]
; AVX512F-NEXT:    vmovdqa %ymm8, %ymm10
; AVX512F-NEXT:    vpternlogq $202, %ymm17, %ymm3, %ymm10
; AVX512F-NEXT:    vextracti128 $1, %ymm10, %xmm11
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm12 = zero,zero,zero,xmm11[4,10],zero,zero,zero,xmm11[2,8,14,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm14 = xmm10[2,8,14],zero,zero,xmm10[0,6,12],zero,zero,zero,xmm10[u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm12, %xmm14, %xmm12
; AVX512F-NEXT:    vpternlogq $202, %ymm1, %ymm5, %ymm9
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm14 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm9[4,10,0,6,12,18,24,30,20,26,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm16 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255]
; AVX512F-NEXT:    vpternlogq $248, %ymm16, %ymm12, %ymm14
; AVX512F-NEXT:    vmovdqa %ymm0, %ymm12
; AVX512F-NEXT:    vpternlogq $202, %ymm13, %ymm6, %ymm12
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm15 = xmm12[u,u,u,u,u,0,6,12],zero,zero,zero,xmm12[4,10],zero,zero,zero
; AVX512F-NEXT:    vextracti128 $1, %ymm12, %xmm2
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm4 = xmm2[u,u,u,u,u],zero,zero,zero,xmm2[2,8,14],zero,zero,xmm2[0,6,12]
; AVX512F-NEXT:    vpor %xmm4, %xmm15, %xmm4
; AVX512F-NEXT:    vinserti128 $1, %xmm4, %ymm0, %ymm4
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm15 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX512F-NEXT:    vpternlogq $184, %ymm14, %ymm15, %ymm4
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm11 = zero,zero,zero,xmm11[5,11],zero,zero,zero,xmm11[3,9,15,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm10 = xmm10[3,9,15],zero,zero,xmm10[1,7,13],zero,zero,zero,xmm10[u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm11, %xmm10, %xmm10
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm9 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm9[5,11,1,7,13,19,25,31,21,27,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpternlogq $248, %ymm16, %ymm10, %ymm9
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm10 = xmm12[u,u,u,u,u,1,7,13],zero,zero,zero,xmm12[5,11],zero,zero,zero
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[u,u,u,u,u],zero,zero,zero,xmm2[3,9,15],zero,zero,xmm2[1,7,13]
; AVX512F-NEXT:    vpor %xmm2, %xmm10, %xmm2
; AVX512F-NEXT:    vinserti128 $1, %xmm2, %ymm0, %ymm2
; AVX512F-NEXT:    vpternlogq $184, %ymm9, %ymm15, %ymm2
; AVX512F-NEXT:    vpternlogq $202, %ymm6, %ymm13, %ymm8
; AVX512F-NEXT:    vextracti128 $1, %ymm8, %xmm6
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm9 = xmm6[u,u,u,u,u],zero,zero,zero,xmm6[4,10],zero,zero,zero,xmm6[2,8,14]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm10 = xmm8[u,u,u,u,u,2,8,14],zero,zero,xmm8[0,6,12],zero,zero,zero
; AVX512F-NEXT:    vpor %xmm9, %xmm10, %xmm9
; AVX512F-NEXT:    vinserti128 $1, %xmm9, %ymm0, %ymm9
; AVX512F-NEXT:    vpternlogq $202, %ymm17, %ymm3, %ymm0
; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm10 = zero,zero,xmm3[0,6,12],zero,zero,zero,xmm3[4,10,u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm11 = xmm0[4,10],zero,zero,zero,xmm0[2,8,14],zero,zero,xmm0[u,u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm10, %xmm11, %xmm10
; AVX512F-NEXT:    vpternlogq $216, {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm5, %ymm1
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm5 = ymm1[u,u,u,u,u,u,u,u,u,u,0,6,12,2,8,14,20,26,16,22,28,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpblendw {{.*#+}} xmm10 = xmm10[0,1,2,3,4],xmm5[5,6,7]
; AVX512F-NEXT:    vpblendd {{.*#+}} ymm5 = ymm10[0,1,2,3],ymm5[4,5,6,7]
; AVX512F-NEXT:    vpternlogq $226, %ymm9, %ymm15, %ymm5
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm6 = xmm6[u,u,u,u,u],zero,zero,zero,xmm6[5,11],zero,zero,zero,xmm6[3,9,15]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm8 = xmm8[u,u,u,u,u,3,9,15],zero,zero,xmm8[1,7,13],zero,zero,zero
; AVX512F-NEXT:    vpor %xmm6, %xmm8, %xmm6
; AVX512F-NEXT:    vinserti128 $1, %xmm6, %ymm0, %ymm6
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm3 = zero,zero,xmm3[1,7,13],zero,zero,zero,xmm3[5,11,u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,11],zero,zero,zero,xmm0[3,9,15],zero,zero,xmm0[u,u,u,u,u,u]
; AVX512F-NEXT:    vpor %xmm3, %xmm0, %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,u,u,1,7,13,3,9,15,21,27,17,23,29,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5,6,7]
; AVX512F-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX512F-NEXT:    vpternlogq $226, %ymm6, %ymm15, %ymm0
; AVX512F-NEXT:    vmovdqa64 %ymm18, (%rsi)
; AVX512F-NEXT:    vmovdqa %ymm7, (%rdx)
; AVX512F-NEXT:    vmovdqa %ymm4, (%rcx)
; AVX512F-NEXT:    vmovdqa %ymm2, (%r8)
; AVX512F-NEXT:    vmovdqa %ymm5, (%r9)
; AVX512F-NEXT:    vmovdqa %ymm0, (%rax)
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride6_vf32:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX512BW-NEXT:    vmovdqa (%rdi), %ymm4
; AVX512BW-NEXT:    vmovdqa 32(%rdi), %ymm0
; AVX512BW-NEXT:    vmovdqa 64(%rdi), %ymm3
; AVX512BW-NEXT:    vmovdqa 128(%rdi), %ymm2
; AVX512BW-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm3[2,3],mem[2,3]
; AVX512BW-NEXT:    vinserti128 $1, 96(%rdi), %ymm3, %ymm8
; AVX512BW-NEXT:    movw $-28124, %r10w # imm = 0x9224
; AVX512BW-NEXT:    kmovd %r10d, %k2
; AVX512BW-NEXT:    vpblendmw %ymm1, %ymm8, %ymm6 {%k2}
; AVX512BW-NEXT:    movw $18724, %r10w # imm = 0x4924
; AVX512BW-NEXT:    kmovd %r10d, %k1
; AVX512BW-NEXT:    vpblendmw %ymm0, %ymm4, %ymm7 {%k1}
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm3 = xmm7[0,6,12],zero,zero,zero,xmm7[4,10],zero,zero,zero,xmm7[u,u,u,u,u]
; AVX512BW-NEXT:    vextracti128 $1, %ymm7, %xmm9
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm5 = zero,zero,zero,xmm9[2,8,14],zero,zero,xmm9[0,6,12,u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm3, %xmm5, %xmm5
; AVX512BW-NEXT:    movl $4192256, %r10d # imm = 0x3FF800
; AVX512BW-NEXT:    kmovd %r10d, %k3
; AVX512BW-NEXT:    vpshufb {{.*#+}} ymm5 {%k3} = ymm6[u,u,u,u,u,u,u,u,u,u,u,2,8,14,4,10,16,22,28,18,24,30,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vmovdqa 160(%rdi), %ymm3
; AVX512BW-NEXT:    vpblendmw %ymm2, %ymm3, %ymm10 {%k1}
; AVX512BW-NEXT:    vextracti128 $1, %ymm10, %xmm11
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm12 = xmm11[u,u,u,u,u,u],zero,zero,xmm11[0,6,12],zero,zero,zero,xmm11[4,10]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm13 = xmm10[u,u,u,u,u,u,4,10],zero,zero,zero,xmm10[2,8,14],zero,zero
; AVX512BW-NEXT:    vpor %xmm12, %xmm13, %xmm12
; AVX512BW-NEXT:    vinserti128 $1, %xmm12, %ymm0, %ymm12
; AVX512BW-NEXT:    vpblendw {{.*#+}} ymm12 = ymm5[0,1,2],ymm12[3,4,5,6,7],ymm5[8,9,10],ymm12[11,12,13,14,15]
; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm5 = ymm5[0,1,2,3],ymm12[4,5,6,7]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm7 = xmm7[1,7,13],zero,zero,zero,xmm7[5,11],zero,zero,zero,xmm7[u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm9 = zero,zero,zero,xmm9[3,9,15],zero,zero,xmm9[1,7,13,u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm7, %xmm9, %xmm7
; AVX512BW-NEXT:    vpshufb {{.*#+}} ymm7 {%k3} = ymm6[u,u,u,u,u,u,u,u,u,u,u,3,9,15,5,11,17,23,29,19,25,31,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm6 = xmm11[u,u,u,u,u,u],zero,zero,xmm11[1,7,13],zero,zero,zero,xmm11[5,11]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm9 = xmm10[u,u,u,u,u,u,5,11],zero,zero,zero,xmm10[3,9,15],zero,zero
; AVX512BW-NEXT:    vpor %xmm6, %xmm9, %xmm6
; AVX512BW-NEXT:    vinserti128 $1, %xmm6, %ymm0, %ymm6
; AVX512BW-NEXT:    vpblendw {{.*#+}} ymm6 = ymm7[0,1,2],ymm6[3,4,5,6,7],ymm7[8,9,10],ymm6[11,12,13,14,15]
; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm6 = ymm7[0,1,2,3],ymm6[4,5,6,7]
; AVX512BW-NEXT:    vpblendmw %ymm8, %ymm1, %ymm9 {%k2}
; AVX512BW-NEXT:    movw $9362, %di # imm = 0x2492
; AVX512BW-NEXT:    kmovd %edi, %k3
; AVX512BW-NEXT:    vpblendmw %ymm4, %ymm0, %ymm10 {%k3}
; AVX512BW-NEXT:    vextracti128 $1, %ymm10, %xmm11
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm7 = zero,zero,zero,xmm11[4,10],zero,zero,zero,xmm11[2,8,14,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm12 = xmm10[2,8,14],zero,zero,xmm10[0,6,12],zero,zero,zero,xmm10[u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm7, %xmm12, %xmm7
; AVX512BW-NEXT:    movl $2095104, %edi # imm = 0x1FF800
; AVX512BW-NEXT:    kmovd %edi, %k4
; AVX512BW-NEXT:    vpshufb {{.*#+}} ymm7 {%k4} = ymm9[u,u,u,u,u,u,u,u,u,u,u,4,10,0,6,12,18,24,30,20,26,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpblendmw %ymm3, %ymm2, %ymm12 {%k1}
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm13 = xmm12[u,u,u,u,u,0,6,12],zero,zero,zero,xmm12[4,10],zero,zero,zero
; AVX512BW-NEXT:    vextracti128 $1, %ymm12, %xmm14
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm15 = xmm14[u,u,u,u,u],zero,zero,zero,xmm14[2,8,14],zero,zero,xmm14[0,6,12]
; AVX512BW-NEXT:    vpor %xmm13, %xmm15, %xmm13
; AVX512BW-NEXT:    vinserti128 $1, %xmm13, %ymm0, %ymm13
; AVX512BW-NEXT:    movl $-2097152, %edi # imm = 0xFFE00000
; AVX512BW-NEXT:    kmovd %edi, %k2
; AVX512BW-NEXT:    vmovdqu8 %ymm13, %ymm7 {%k2}
; AVX512BW-NEXT:    movw $9289, %di # imm = 0x2449
; AVX512BW-NEXT:    kmovd %edi, %k5
; AVX512BW-NEXT:    vmovdqu16 %ymm8, %ymm1 {%k5}
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm8 = zero,zero,zero,xmm11[5,11],zero,zero,zero,xmm11[3,9,15,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm10 = xmm10[3,9,15],zero,zero,xmm10[1,7,13],zero,zero,zero,xmm10[u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm8, %xmm10, %xmm8
; AVX512BW-NEXT:    vpshufb {{.*#+}} ymm8 {%k4} = ymm9[u,u,u,u,u,u,u,u,u,u,u,5,11,1,7,13,19,25,31,21,27,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm9 = xmm12[u,u,u,u,u,1,7,13],zero,zero,zero,xmm12[5,11],zero,zero,zero
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm10 = xmm14[u,u,u,u,u],zero,zero,zero,xmm14[3,9,15],zero,zero,xmm14[1,7,13]
; AVX512BW-NEXT:    vpor %xmm9, %xmm10, %xmm9
; AVX512BW-NEXT:    vinserti128 $1, %xmm9, %ymm0, %ymm9
; AVX512BW-NEXT:    vmovdqu8 %ymm9, %ymm8 {%k2}
; AVX512BW-NEXT:    vpshufb {{.*#+}} ymm9 = ymm1[u,u,u,u,u,u,u,u,u,u,0,6,12,2,8,14,20,26,16,22,28,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vmovdqu16 %ymm4, %ymm0 {%k1}
; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm4
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm10 = zero,zero,xmm4[0,6,12],zero,zero,zero,xmm4[4,10,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm11 = xmm0[4,10],zero,zero,zero,xmm0[2,8,14],zero,zero,xmm0[u,u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm10, %xmm11, %xmm10
; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm10 = xmm10[0,1,2,3,4],xmm9[5,6,7]
; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm9 = ymm10[0,1,2,3],ymm9[4,5,6,7]
; AVX512BW-NEXT:    vmovdqu16 %ymm2, %ymm3 {%k3}
; AVX512BW-NEXT:    vextracti128 $1, %ymm3, %xmm2
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm10 = xmm2[u,u,u,u,u],zero,zero,zero,xmm2[4,10],zero,zero,zero,xmm2[2,8,14]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm11 = xmm3[u,u,u,u,u,2,8,14],zero,zero,xmm3[0,6,12],zero,zero,zero
; AVX512BW-NEXT:    vpor %xmm10, %xmm11, %xmm10
; AVX512BW-NEXT:    vinserti128 $1, %xmm10, %ymm0, %ymm10
; AVX512BW-NEXT:    vmovdqu8 %ymm10, %ymm9 {%k2}
; AVX512BW-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,u,u,1,7,13,3,9,15,21,27,17,23,29,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm4 = zero,zero,xmm4[1,7,13],zero,zero,zero,xmm4[5,11,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[5,11],zero,zero,zero,xmm0[3,9,15],zero,zero,xmm0[u,u,u,u,u,u]
; AVX512BW-NEXT:    vpor %xmm4, %xmm0, %xmm0
; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5,6,7]
; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm1 = xmm2[u,u,u,u,u],zero,zero,zero,xmm2[5,11],zero,zero,zero,xmm2[3,9,15]
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm2 = xmm3[u,u,u,u,u,3,9,15],zero,zero,xmm3[1,7,13],zero,zero,zero
; AVX512BW-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX512BW-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX512BW-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k2}
; AVX512BW-NEXT:    vmovdqa %ymm5, (%rsi)
; AVX512BW-NEXT:    vmovdqa %ymm6, (%rdx)
; AVX512BW-NEXT:    vmovdqa %ymm7, (%rcx)
; AVX512BW-NEXT:    vmovdqa %ymm8, (%r8)
; AVX512BW-NEXT:    vmovdqa %ymm9, (%r9)
; AVX512BW-NEXT:    vmovdqa %ymm0, (%rax)
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <192 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <192 x i8> %wide.vec, <192 x i8> poison, <32 x i32> <i32 0, i32 6, i32 12, i32 18, i32 24, i32 30, i32 36, i32 42, i32 48, i32 54, i32 60, i32 66, i32 72, i32 78, i32 84, i32 90, i32 96, i32 102, i32 108, i32 114, i32 120, i32 126, i32 132, i32 138, i32 144, i32 150, i32 156, i32 162, i32 168, i32 174, i32 180, i32 186>
  %strided.vec1 = shufflevector <192 x i8> %wide.vec, <192 x i8> poison, <32 x i32> <i32 1, i32 7, i32 13, i32 19, i32 25, i32 31, i32 37, i32 43, i32 49, i32 55, i32 61, i32 67, i32 73, i32 79, i32 85, i32 91, i32 97, i32 103, i32 109, i32 115, i32 121, i32 127, i32 133, i32 139, i32 145, i32 151, i32 157, i32 163, i32 169, i32 175, i32 181, i32 187>
  %strided.vec2 = shufflevector <192 x i8> %wide.vec, <192 x i8> poison, <32 x i32> <i32 2, i32 8, i32 14, i32 20, i32 26, i32 32, i32 38, i32 44, i32 50, i32 56, i32 62, i32 68, i32 74, i32 80, i32 86, i32 92, i32 98, i32 104, i32 110, i32 116, i32 122, i32 128, i32 134, i32 140, i32 146, i32 152, i32 158, i32 164, i32 170, i32 176, i32 182, i32 188>
  %strided.vec3 = shufflevector <192 x i8> %wide.vec, <192 x i8> poison, <32 x i32> <i32 3, i32 9, i32 15, i32 21, i32 27, i32 33, i32 39, i32 45, i32 51, i32 57, i32 63, i32 69, i32 75, i32 81, i32 87, i32 93, i32 99, i32 105, i32 111, i32 117, i32 123, i32 129, i32 135, i32 141, i32 147, i32 153, i32 159, i32 165, i32 171, i32 177, i32 183, i32 189>
  %strided.vec4 = shufflevector <192 x i8> %wide.vec, <192 x i8> poison, <32 x i32> <i32 4, i32 10, i32 16, i32 22, i32 28, i32 34, i32 40, i32 46, i32 52, i32 58, i32 64, i32 70, i32 76, i32 82, i32 88, i32 94, i32 100, i32 106, i32 112, i32 118, i32 124, i32 130, i32 136, i32 142, i32 148, i32 154, i32 160, i32 166, i32 172, i32 178, i32 184, i32 190>
  %strided.vec5 = shufflevector <192 x i8> %wide.vec, <192 x i8> poison, <32 x i32> <i32 5, i32 11, i32 17, i32 23, i32 29, i32 35, i32 41, i32 47, i32 53, i32 59, i32 65, i32 71, i32 77, i32 83, i32 89, i32 95, i32 101, i32 107, i32 113, i32 119, i32 125, i32 131, i32 137, i32 143, i32 149, i32 155, i32 161, i32 167, i32 173, i32 179, i32 185, i32 191>
  store <32 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <32 x i8> %strided.vec1, ptr %out.vec1, align 64
  store <32 x i8> %strided.vec2, ptr %out.vec2, align 64
  store <32 x i8> %strided.vec3, ptr %out.vec3, align 64
  store <32 x i8> %strided.vec4, ptr %out.vec4, align 64
  store <32 x i8> %strided.vec5, ptr %out.vec5, align 64
  ret void
}

define void @load_i8_stride6_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4, ptr %out.vec5) nounwind {
; SSE-LABEL: load_i8_stride6_vf64:
; SSE:       # %bb.0:
; SSE-NEXT:    subq $808, %rsp # imm = 0x328
; SSE-NEXT:    movdqa 64(%rdi), %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 80(%rdi), %xmm5
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa (%rdi), %xmm7
; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 16(%rdi), %xmm6
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 32(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 48(%rdi), %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm9 = [65535,65535,0,65535,65535,0,65535,65535]
; SSE-NEXT:    movdqa %xmm9, %xmm1
; SSE-NEXT:    pandn %xmm2, %xmm1
; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    movdqa %xmm3, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm9, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm9, %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm1, %xmm0
; SSE-NEXT:    movdqa %xmm1, %xmm11
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,5]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm8 = [65535,65535,65535,0,0,0,65535,65535]
; SSE-NEXT:    movdqa %xmm3, %xmm1
; SSE-NEXT:    pandn %xmm6, %xmm1
; SSE-NEXT:    movdqa %xmm7, %xmm2
; SSE-NEXT:    pand %xmm3, %xmm2
; SSE-NEXT:    por %xmm1, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm2, %xmm1
; SSE-NEXT:    pand %xmm11, %xmm1
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,3,2,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,7,6,7]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    pand %xmm8, %xmm1
; SSE-NEXT:    movdqa %xmm8, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    por %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm9, %xmm6
; SSE-NEXT:    movdqa %xmm9, %xmm0
; SSE-NEXT:    pandn %xmm5, %xmm0
; SSE-NEXT:    pand %xmm9, %xmm4
; SSE-NEXT:    por %xmm0, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm4, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,3,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,5]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; SSE-NEXT:    movdqa %xmm4, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    pand %xmm4, %xmm1
; SSE-NEXT:    por %xmm1, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 320(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm9, %xmm0
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    movdqa 336(%rdi), %xmm13
; SSE-NEXT:    movdqa %xmm3, %xmm1
; SSE-NEXT:    pandn %xmm13, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm9, %xmm1
; SSE-NEXT:    pandn %xmm13, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm9, %xmm13
; SSE-NEXT:    por %xmm0, %xmm13
; SSE-NEXT:    movdqa %xmm13, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,5]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa %xmm8, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    movdqa 304(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm3, %xmm0
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    movdqa 288(%rdi), %xmm7
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    pand %xmm3, %xmm1
; SSE-NEXT:    por %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm1, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    por %xmm2, %xmm0
; SSE-NEXT:    movdqa 368(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm9, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    movdqa 352(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm9, %xmm1
; SSE-NEXT:    por %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm1, %xmm2
; SSE-NEXT:    pand %xmm11, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[3,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,7,6,5]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm4, %xmm1
; SSE-NEXT:    pandn %xmm2, %xmm1
; SSE-NEXT:    pand %xmm4, %xmm0
; SSE-NEXT:    movdqa %xmm4, %xmm9
; SSE-NEXT:    por %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 224(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm6, %xmm0
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    movdqa 240(%rdi), %xmm14
; SSE-NEXT:    movdqa %xmm3, %xmm1
; SSE-NEXT:    pandn %xmm14, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm6, %xmm1
; SSE-NEXT:    pandn %xmm14, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm6, %xmm14
; SSE-NEXT:    por %xmm0, %xmm14
; SSE-NEXT:    movdqa %xmm14, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,5]
; SSE-NEXT:    packuswb %xmm2, %xmm0
; SSE-NEXT:    movdqa %xmm8, %xmm5
; SSE-NEXT:    movdqa %xmm8, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    movdqa 208(%rdi), %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm3, %xmm1
; SSE-NEXT:    movdqa %xmm3, %xmm0
; SSE-NEXT:    pandn %xmm4, %xmm0
; SSE-NEXT:    movdqa 192(%rdi), %xmm3
; SSE-NEXT:    movdqa %xmm3, %xmm4
; SSE-NEXT:    pand %xmm1, %xmm4
; SSE-NEXT:    movdqa %xmm1, %xmm8
; SSE-NEXT:    por %xmm0, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm4, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pand %xmm5, %xmm0
; SSE-NEXT:    movdqa %xmm5, %xmm1
; SSE-NEXT:    por %xmm2, %xmm0
; SSE-NEXT:    movdqa 272(%rdi), %xmm15
; SSE-NEXT:    movdqa %xmm6, %xmm2
; SSE-NEXT:    pandn %xmm15, %xmm2
; SSE-NEXT:    movdqa 256(%rdi), %xmm12
; SSE-NEXT:    movdqa %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm6, %xmm12
; SSE-NEXT:    por %xmm2, %xmm12
; SSE-NEXT:    movdqa %xmm12, %xmm2
; SSE-NEXT:    pand %xmm11, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[3,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,7,6,5]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm9, %xmm4
; SSE-NEXT:    pandn %xmm2, %xmm4
; SSE-NEXT:    pand %xmm9, %xmm0
; SSE-NEXT:    por %xmm0, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 128(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, (%rsp) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm6, %xmm0
; SSE-NEXT:    pandn %xmm2, %xmm0
; SSE-NEXT:    movdqa 144(%rdi), %xmm10
; SSE-NEXT:    movdqa %xmm8, %xmm4
; SSE-NEXT:    pandn %xmm10, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm6, %xmm4
; SSE-NEXT:    pandn %xmm10, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm6, %xmm10
; SSE-NEXT:    por %xmm0, %xmm10
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,5]
; SSE-NEXT:    packuswb %xmm5, %xmm0
; SSE-NEXT:    movdqa %xmm1, %xmm11
; SSE-NEXT:    pandn %xmm0, %xmm11
; SSE-NEXT:    movdqa %xmm6, %xmm0
; SSE-NEXT:    movdqa %xmm6, %xmm4
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    pandn %xmm1, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm6, %xmm2
; SSE-NEXT:    pandn %xmm7, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm4
; SSE-NEXT:    movdqa %xmm6, %xmm2
; SSE-NEXT:    pandn %xmm3, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm3, %xmm5
; SSE-NEXT:    movdqa 112(%rdi), %xmm6
; SSE-NEXT:    movdqa %xmm8, %xmm7
; SSE-NEXT:    movdqa %xmm8, %xmm9
; SSE-NEXT:    pandn %xmm6, %xmm9
; SSE-NEXT:    movdqa 160(%rdi), %xmm8
; SSE-NEXT:    movdqa %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm8
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    pandn %xmm2, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    pandn %xmm3, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    pandn %xmm1, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm3
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    pandn %xmm4, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm4, %xmm3
; SSE-NEXT:    pand %xmm0, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    pandn %xmm4, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm5
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm3
; SSE-NEXT:    pandn %xmm15, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm15
; SSE-NEXT:    movdqa %xmm15, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    pandn %xmm6, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm6, %xmm15
; SSE-NEXT:    movdqa 96(%rdi), %xmm5
; SSE-NEXT:    movdqa %xmm5, %xmm3
; SSE-NEXT:    pand %xmm0, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 176(%rdi), %xmm6
; SSE-NEXT:    movdqa %xmm6, %xmm3
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa (%rsp), %xmm1 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, (%rsp) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, %xmm1
; SSE-NEXT:    pand %xmm0, %xmm15
; SSE-NEXT:    movdqa %xmm15, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pandn %xmm5, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm5, %xmm3
; SSE-NEXT:    pand %xmm7, %xmm3
; SSE-NEXT:    por %xmm9, %xmm3
; SSE-NEXT:    movdqa %xmm3, %xmm5
; SSE-NEXT:    movdqa {{.*#+}} xmm7 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm7, %xmm5
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,2,1,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[0,3,2,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,4,7,6,7]
; SSE-NEXT:    packuswb %xmm5, %xmm5
; SSE-NEXT:    movdqa {{.*#+}} xmm9 = [65535,65535,65535,0,0,0,65535,65535]
; SSE-NEXT:    pand %xmm9, %xmm5
; SSE-NEXT:    por %xmm11, %xmm5
; SSE-NEXT:    pandn %xmm6, %xmm0
; SSE-NEXT:    por %xmm0, %xmm8
; SSE-NEXT:    movdqa %xmm8, %xmm0
; SSE-NEXT:    pand %xmm7, %xmm0
; SSE-NEXT:    movdqa %xmm7, %xmm11
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,3,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,5]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm7 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm7, %xmm5
; SSE-NEXT:    por %xmm5, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pxor %xmm5, %xmm5
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm1, %xmm0
; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm5[8],xmm0[9],xmm5[9],xmm0[10],xmm5[10],xmm0[11],xmm5[11],xmm0[12],xmm5[12],xmm0[13],xmm5[13],xmm0[14],xmm5[14],xmm0[15],xmm5[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3],xmm1[4],xmm5[4],xmm1[5],xmm5[5],xmm1[6],xmm5[6],xmm1[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[2,2,3,3]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3]
; SSE-NEXT:    psrld $16, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm15 = xmm1[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm15 = xmm15[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm15 = xmm15[2],xmm0[2],xmm15[3],xmm0[3]
; SSE-NEXT:    packuswb %xmm15, %xmm4
; SSE-NEXT:    movdqa %xmm9, %xmm1
; SSE-NEXT:    pandn %xmm4, %xmm1
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm2, %xmm4
; SSE-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm5[8],xmm4[9],xmm5[9],xmm4[10],xmm5[10],xmm4[11],xmm5[11],xmm4[12],xmm5[12],xmm4[13],xmm5[13],xmm4[14],xmm5[14],xmm4[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[1,1,1,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm4[0,1,2,3,5,7,6,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [65535,65535,0,65535,0,0,65535,65535]
; SSE-NEXT:    movdqa %xmm0, %xmm15
; SSE-NEXT:    pandn %xmm4, %xmm15
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm2[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[1,3,2,0,4,5,6,7]
; SSE-NEXT:    pand %xmm0, %xmm4
; SSE-NEXT:    por %xmm15, %xmm4
; SSE-NEXT:    packuswb %xmm4, %xmm4
; SSE-NEXT:    pand %xmm9, %xmm4
; SSE-NEXT:    por %xmm1, %xmm4
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm6, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3],xmm1[4],xmm5[4],xmm1[5],xmm5[5],xmm1[6],xmm5[6],xmm1[7],xmm5[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm15 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [65535,65535,65535,0,65535,65535,0,65535]
; SSE-NEXT:    movdqa %xmm2, %xmm1
; SSE-NEXT:    pandn %xmm15, %xmm1
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm5[8],xmm6[9],xmm5[9],xmm6[10],xmm5[10],xmm6[11],xmm5[11],xmm6[12],xmm5[12],xmm6[13],xmm5[13],xmm6[14],xmm5[14],xmm6[15],xmm5[15]
; SSE-NEXT:    pshuflw {{.*#+}} xmm15 = xmm6[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm15 = xmm15[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm15 = xmm15[0,1,2,3,5,7,6,4]
; SSE-NEXT:    pand %xmm2, %xmm15
; SSE-NEXT:    por %xmm1, %xmm15
; SSE-NEXT:    packuswb %xmm15, %xmm1
; SSE-NEXT:    movdqa %xmm7, %xmm15
; SSE-NEXT:    pandn %xmm1, %xmm15
; SSE-NEXT:    pand %xmm7, %xmm4
; SSE-NEXT:    por %xmm4, %xmm15
; SSE-NEXT:    movdqa %xmm15, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm13, %xmm1
; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm5[8],xmm1[9],xmm5[9],xmm1[10],xmm5[10],xmm1[11],xmm5[11],xmm1[12],xmm5[12],xmm1[13],xmm5[13],xmm1[14],xmm5[14],xmm1[15],xmm5[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm13 = xmm13[0],xmm5[0],xmm13[1],xmm5[1],xmm13[2],xmm5[2],xmm13[3],xmm5[3],xmm13[4],xmm5[4],xmm13[5],xmm5[5],xmm13[6],xmm5[6],xmm13[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm13[2,2,3,3]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3]
; SSE-NEXT:    psrld $16, %xmm1
; SSE-NEXT:    pshufd {{.*#+}} xmm13 = xmm13[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm13 = xmm13[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm13 = xmm13[2],xmm1[2],xmm13[3],xmm1[3]
; SSE-NEXT:    packuswb %xmm13, %xmm4
; SSE-NEXT:    movdqa %xmm9, %xmm15
; SSE-NEXT:    movdqa %xmm9, %xmm1
; SSE-NEXT:    pandn %xmm4, %xmm1
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm6, %xmm4
; SSE-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm5[8],xmm4[9],xmm5[9],xmm4[10],xmm5[10],xmm4[11],xmm5[11],xmm4[12],xmm5[12],xmm4[13],xmm5[13],xmm4[14],xmm5[14],xmm4[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[1,1,1,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm4[0,1,2,3,5,7,6,7]
; SSE-NEXT:    movdqa %xmm0, %xmm13
; SSE-NEXT:    pandn %xmm4, %xmm13
; SSE-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm5[0],xmm6[1],xmm5[1],xmm6[2],xmm5[2],xmm6[3],xmm5[3],xmm6[4],xmm5[4],xmm6[5],xmm5[5],xmm6[6],xmm5[6],xmm6[7],xmm5[7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm6[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[1,3,2,0,4,5,6,7]
; SSE-NEXT:    pand %xmm0, %xmm4
; SSE-NEXT:    por %xmm13, %xmm4
; SSE-NEXT:    packuswb %xmm4, %xmm4
; SSE-NEXT:    pand %xmm9, %xmm4
; SSE-NEXT:    por %xmm1, %xmm4
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm6, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3],xmm1[4],xmm5[4],xmm1[5],xmm5[5],xmm1[6],xmm5[6],xmm1[7],xmm5[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa %xmm2, %xmm13
; SSE-NEXT:    pandn %xmm1, %xmm13
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm5[8],xmm6[9],xmm5[9],xmm6[10],xmm5[10],xmm6[11],xmm5[11],xmm6[12],xmm5[12],xmm6[13],xmm5[13],xmm6[14],xmm5[14],xmm6[15],xmm5[15]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm6[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,7,6,4]
; SSE-NEXT:    pand %xmm2, %xmm1
; SSE-NEXT:    por %xmm13, %xmm1
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa %xmm7, %xmm13
; SSE-NEXT:    pandn %xmm1, %xmm13
; SSE-NEXT:    pand %xmm7, %xmm4
; SSE-NEXT:    por %xmm4, %xmm13
; SSE-NEXT:    movdqa %xmm13, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm14, %xmm1
; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm5[8],xmm1[9],xmm5[9],xmm1[10],xmm5[10],xmm1[11],xmm5[11],xmm1[12],xmm5[12],xmm1[13],xmm5[13],xmm1[14],xmm5[14],xmm1[15],xmm5[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm14 = xmm14[0],xmm5[0],xmm14[1],xmm5[1],xmm14[2],xmm5[2],xmm14[3],xmm5[3],xmm14[4],xmm5[4],xmm14[5],xmm5[5],xmm14[6],xmm5[6],xmm14[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm14[2,2,3,3]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3]
; SSE-NEXT:    psrld $16, %xmm1
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm14[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm9[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm9 = xmm9[2],xmm1[2],xmm9[3],xmm1[3]
; SSE-NEXT:    packuswb %xmm9, %xmm4
; SSE-NEXT:    movdqa %xmm15, %xmm1
; SSE-NEXT:    pandn %xmm4, %xmm1
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm6, %xmm4
; SSE-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm5[8],xmm4[9],xmm5[9],xmm4[10],xmm5[10],xmm4[11],xmm5[11],xmm4[12],xmm5[12],xmm4[13],xmm5[13],xmm4[14],xmm5[14],xmm4[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[1,1,1,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm4[0,1,2,3,5,7,6,7]
; SSE-NEXT:    movdqa %xmm0, %xmm9
; SSE-NEXT:    pandn %xmm4, %xmm9
; SSE-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm5[0],xmm6[1],xmm5[1],xmm6[2],xmm5[2],xmm6[3],xmm5[3],xmm6[4],xmm5[4],xmm6[5],xmm5[5],xmm6[6],xmm5[6],xmm6[7],xmm5[7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm6[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[1,3,2,0,4,5,6,7]
; SSE-NEXT:    pand %xmm0, %xmm4
; SSE-NEXT:    por %xmm9, %xmm4
; SSE-NEXT:    packuswb %xmm4, %xmm4
; SSE-NEXT:    pand %xmm15, %xmm4
; SSE-NEXT:    por %xmm1, %xmm4
; SSE-NEXT:    movdqa %xmm12, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3],xmm1[4],xmm5[4],xmm1[5],xmm5[5],xmm1[6],xmm5[6],xmm1[7],xmm5[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa %xmm2, %xmm9
; SSE-NEXT:    pandn %xmm1, %xmm9
; SSE-NEXT:    punpckhbw {{.*#+}} xmm12 = xmm12[8],xmm5[8],xmm12[9],xmm5[9],xmm12[10],xmm5[10],xmm12[11],xmm5[11],xmm12[12],xmm5[12],xmm12[13],xmm5[13],xmm12[14],xmm5[14],xmm12[15],xmm5[15]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm12[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,7,6,4]
; SSE-NEXT:    pand %xmm2, %xmm1
; SSE-NEXT:    por %xmm9, %xmm1
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa %xmm7, %xmm9
; SSE-NEXT:    pandn %xmm1, %xmm9
; SSE-NEXT:    pand %xmm7, %xmm4
; SSE-NEXT:    por %xmm4, %xmm9
; SSE-NEXT:    movdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm10, %xmm1
; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm5[8],xmm1[9],xmm5[9],xmm1[10],xmm5[10],xmm1[11],xmm5[11],xmm1[12],xmm5[12],xmm1[13],xmm5[13],xmm1[14],xmm5[14],xmm1[15],xmm5[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm5[0],xmm10[1],xmm5[1],xmm10[2],xmm5[2],xmm10[3],xmm5[3],xmm10[4],xmm5[4],xmm10[5],xmm5[5],xmm10[6],xmm5[6],xmm10[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm10[2,2,3,3]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3]
; SSE-NEXT:    psrld $16, %xmm1
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm10[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm6 = xmm6[2],xmm1[2],xmm6[3],xmm1[3]
; SSE-NEXT:    packuswb %xmm6, %xmm4
; SSE-NEXT:    movdqa %xmm3, %xmm1
; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm5[8],xmm1[9],xmm5[9],xmm1[10],xmm5[10],xmm1[11],xmm5[11],xmm1[12],xmm5[12],xmm1[13],xmm5[13],xmm1[14],xmm5[14],xmm1[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,1,1,1,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,7,6,7]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1],xmm3[2],xmm5[2],xmm3[3],xmm5[3],xmm3[4],xmm5[4],xmm3[5],xmm5[5],xmm3[6],xmm5[6],xmm3[7],xmm5[7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[1,3,2,0,4,5,6,7]
; SSE-NEXT:    pand %xmm0, %xmm3
; SSE-NEXT:    pandn %xmm1, %xmm0
; SSE-NEXT:    por %xmm3, %xmm0
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa %xmm15, %xmm1
; SSE-NEXT:    pand %xmm15, %xmm0
; SSE-NEXT:    pandn %xmm4, %xmm1
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    movdqa %xmm8, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3],xmm1[4],xmm5[4],xmm1[5],xmm5[5],xmm1[6],xmm5[6],xmm1[7],xmm5[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm8 = xmm8[8],xmm5[8],xmm8[9],xmm5[9],xmm8[10],xmm5[10],xmm8[11],xmm5[11],xmm8[12],xmm5[12],xmm8[13],xmm5[13],xmm8[14],xmm5[14],xmm8[15],xmm5[15]
; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm8[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,5,7,6,4]
; SSE-NEXT:    pand %xmm2, %xmm3
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    por %xmm3, %xmm2
; SSE-NEXT:    packuswb %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm7, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    pand %xmm7, %xmm0
; SSE-NEXT:    por %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{.*#+}} xmm12 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pand %xmm12, %xmm0
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,3,4,5,6,7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,5,6]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm6, %xmm1
; SSE-NEXT:    pand %xmm11, %xmm1
; SSE-NEXT:    movdqa %xmm11, %xmm4
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,4,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,0,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm3, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    pand %xmm3, %xmm0
; SSE-NEXT:    por %xmm0, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; SSE-NEXT:    por %xmm0, %xmm14
; SSE-NEXT:    movdqa %xmm14, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,2,2,2,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,6,7,4]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm7, %xmm2
; SSE-NEXT:    por %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    pand %xmm12, %xmm1
; SSE-NEXT:    movdqa %xmm12, %xmm10
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm1, %xmm0
; SSE-NEXT:    movdqa %xmm1, %xmm12
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,3,4,5,6,7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,5,6]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm15, %xmm1
; SSE-NEXT:    pand %xmm11, %xmm1
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,4,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,0,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa %xmm3, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    pand %xmm3, %xmm0
; SSE-NEXT:    por %xmm0, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    por %xmm0, %xmm9
; SSE-NEXT:    movdqa %xmm9, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,2,2,2,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,6,7,4]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm7, %xmm2
; SSE-NEXT:    por %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; SSE-NEXT:    pand %xmm10, %xmm8
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm8, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,3,4,5,6,7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,5,6]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm13, %xmm1
; SSE-NEXT:    pand %xmm11, %xmm1
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,4,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,0,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa %xmm3, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    pand %xmm3, %xmm0
; SSE-NEXT:    por %xmm0, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    por %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm1, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,2,2,2,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,6,7,4]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm7, %xmm2
; SSE-NEXT:    por %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    pand %xmm10, %xmm7
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm7, %xmm0
; SSE-NEXT:    pand %xmm11, %xmm0
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,3,4,5,6,7]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,5,6]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm11, %xmm1
; SSE-NEXT:    pand %xmm4, %xmm1
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,4,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[1,2,3,0,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,5,5]
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    movdqa %xmm3, %xmm2
; SSE-NEXT:    pandn %xmm1, %xmm2
; SSE-NEXT:    pand %xmm3, %xmm0
; SSE-NEXT:    por %xmm0, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    por %xmm0, %xmm10
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    pand %xmm4, %xmm0
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,0]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,2,2,2,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,6,7,4]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; SSE-NEXT:    movdqa %xmm4, %xmm1
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    pand %xmm4, %xmm2
; SSE-NEXT:    por %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm4, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1],xmm0[2],xmm5[2],xmm0[3],xmm5[3],xmm0[4],xmm5[4],xmm0[5],xmm5[5],xmm0[6],xmm5[6],xmm0[7],xmm5[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm5[8],xmm4[9],xmm5[9],xmm4[10],xmm5[10],xmm4[11],xmm5[11],xmm4[12],xmm5[12],xmm4[13],xmm5[13],xmm4[14],xmm5[14],xmm4[15],xmm5[15]
; SSE-NEXT:    movdqa %xmm4, %xmm1
; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
; SSE-NEXT:    movaps %xmm0, %xmm2
; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,1],xmm1[0,2]
; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm4[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm4[2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm2[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,2]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,1,3,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm6, %xmm0
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1],xmm0[2],xmm5[2],xmm0[3],xmm5[3],xmm0[4],xmm5[4],xmm0[5],xmm5[5],xmm0[6],xmm5[6],xmm0[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm0[3,1,2,1,4,5,6,7]
; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [0,65535,65535,0,65535,65535,65535,65535]
; SSE-NEXT:    movdqa %xmm0, %xmm4
; SSE-NEXT:    pandn %xmm2, %xmm4
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm5[8],xmm6[9],xmm5[9],xmm6[10],xmm5[10],xmm6[11],xmm5[11],xmm6[12],xmm5[12],xmm6[13],xmm5[13],xmm6[14],xmm5[14],xmm6[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm6[0,3,2,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,3,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,7,7,7]
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    por %xmm4, %xmm2
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm3, %xmm4
; SSE-NEXT:    pandn %xmm2, %xmm4
; SSE-NEXT:    pand %xmm3, %xmm1
; SSE-NEXT:    por %xmm1, %xmm4
; SSE-NEXT:    movdqa %xmm14, %xmm1
; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm5[8],xmm1[9],xmm5[9],xmm1[10],xmm5[10],xmm1[11],xmm5[11],xmm1[12],xmm5[12],xmm1[13],xmm5[13],xmm1[14],xmm5[14],xmm1[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm1[0,1,2,3,7,5,6,5]
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [65535,65535,65535,65535,0,65535,65535,0]
; SSE-NEXT:    movdqa %xmm1, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    punpcklbw {{.*#+}} xmm14 = xmm14[0],xmm5[0],xmm14[1],xmm5[1],xmm14[2],xmm5[2],xmm14[3],xmm5[3],xmm14[4],xmm5[4],xmm14[5],xmm5[5],xmm14[6],xmm5[6],xmm14[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm14[0,2,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,7,7]
; SSE-NEXT:    pand %xmm1, %xmm2
; SSE-NEXT:    por %xmm6, %xmm2
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa {{.*#+}} xmm14 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; SSE-NEXT:    movdqa %xmm14, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    pand %xmm14, %xmm4
; SSE-NEXT:    por %xmm4, %xmm6
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm12, %xmm2
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm12 = xmm12[8],xmm5[8],xmm12[9],xmm5[9],xmm12[10],xmm5[10],xmm12[11],xmm5[11],xmm12[12],xmm5[12],xmm12[13],xmm5[13],xmm12[14],xmm5[14],xmm12[15],xmm5[15]
; SSE-NEXT:    movdqa %xmm12, %xmm4
; SSE-NEXT:    shufps {{.*#+}} xmm4 = xmm4[2,0],xmm2[3,0]
; SSE-NEXT:    movaps %xmm2, %xmm6
; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[0,1],xmm4[0,2]
; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm12[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm12[2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm6[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,1,0,2]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,3,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm2, %xmm4
; SSE-NEXT:    movdqa %xmm15, %xmm2
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,1,4,5,6,7]
; SSE-NEXT:    movdqa %xmm0, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm15 = xmm15[8],xmm5[8],xmm15[9],xmm5[9],xmm15[10],xmm5[10],xmm15[11],xmm5[11],xmm15[12],xmm5[12],xmm15[13],xmm5[13],xmm15[14],xmm5[14],xmm15[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm15[0,3,2,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,3,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,7,7,7]
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    por %xmm6, %xmm2
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm3, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    pand %xmm3, %xmm4
; SSE-NEXT:    por %xmm4, %xmm6
; SSE-NEXT:    movdqa %xmm9, %xmm2
; SSE-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm5[8],xmm2[9],xmm5[9],xmm2[10],xmm5[10],xmm2[11],xmm5[11],xmm2[12],xmm5[12],xmm2[13],xmm5[13],xmm2[14],xmm5[14],xmm2[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,5,6,5]
; SSE-NEXT:    movdqa %xmm1, %xmm4
; SSE-NEXT:    pandn %xmm2, %xmm4
; SSE-NEXT:    punpcklbw {{.*#+}} xmm9 = xmm9[0],xmm5[0],xmm9[1],xmm5[1],xmm9[2],xmm5[2],xmm9[3],xmm5[3],xmm9[4],xmm5[4],xmm9[5],xmm5[5],xmm9[6],xmm5[6],xmm9[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm9[0,2,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,7,7]
; SSE-NEXT:    pand %xmm1, %xmm2
; SSE-NEXT:    por %xmm4, %xmm2
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm14, %xmm4
; SSE-NEXT:    pandn %xmm2, %xmm4
; SSE-NEXT:    pand %xmm14, %xmm6
; SSE-NEXT:    por %xmm6, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm8, %xmm2
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm8 = xmm8[8],xmm5[8],xmm8[9],xmm5[9],xmm8[10],xmm5[10],xmm8[11],xmm5[11],xmm8[12],xmm5[12],xmm8[13],xmm5[13],xmm8[14],xmm5[14],xmm8[15],xmm5[15]
; SSE-NEXT:    movdqa %xmm8, %xmm4
; SSE-NEXT:    shufps {{.*#+}} xmm4 = xmm4[2,0],xmm2[3,0]
; SSE-NEXT:    movaps %xmm2, %xmm6
; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[0,1],xmm4[0,2]
; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm8[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm8[2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm6[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,1,0,2]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,3,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm2, %xmm4
; SSE-NEXT:    movdqa %xmm13, %xmm2
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,1,4,5,6,7]
; SSE-NEXT:    movdqa %xmm0, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm13 = xmm13[8],xmm5[8],xmm13[9],xmm5[9],xmm13[10],xmm5[10],xmm13[11],xmm5[11],xmm13[12],xmm5[12],xmm13[13],xmm5[13],xmm13[14],xmm5[14],xmm13[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm13[0,3,2,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,3,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,7,7,7]
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    por %xmm6, %xmm2
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm3, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    pand %xmm3, %xmm4
; SSE-NEXT:    por %xmm4, %xmm6
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm8, %xmm2
; SSE-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm5[8],xmm2[9],xmm5[9],xmm2[10],xmm5[10],xmm2[11],xmm5[11],xmm2[12],xmm5[12],xmm2[13],xmm5[13],xmm2[14],xmm5[14],xmm2[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,5,6,5]
; SSE-NEXT:    movdqa %xmm1, %xmm4
; SSE-NEXT:    pandn %xmm2, %xmm4
; SSE-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm5[0],xmm8[1],xmm5[1],xmm8[2],xmm5[2],xmm8[3],xmm5[3],xmm8[4],xmm5[4],xmm8[5],xmm5[5],xmm8[6],xmm5[6],xmm8[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm8[0,2,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,7,7]
; SSE-NEXT:    pand %xmm1, %xmm2
; SSE-NEXT:    por %xmm4, %xmm2
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm14, %xmm4
; SSE-NEXT:    pandn %xmm2, %xmm4
; SSE-NEXT:    pand %xmm14, %xmm6
; SSE-NEXT:    por %xmm6, %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm2
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm7 = xmm7[8],xmm5[8],xmm7[9],xmm5[9],xmm7[10],xmm5[10],xmm7[11],xmm5[11],xmm7[12],xmm5[12],xmm7[13],xmm5[13],xmm7[14],xmm5[14],xmm7[15],xmm5[15]
; SSE-NEXT:    movdqa %xmm7, %xmm4
; SSE-NEXT:    shufps {{.*#+}} xmm4 = xmm4[2,0],xmm2[3,0]
; SSE-NEXT:    movaps %xmm2, %xmm6
; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[0,1],xmm4[0,2]
; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm7[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm7[2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm4 = xmm6[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,1,0,2]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,3,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm2, %xmm4
; SSE-NEXT:    movdqa %xmm11, %xmm2
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[3,1,2,1,4,5,6,7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm11 = xmm11[8],xmm5[8],xmm11[9],xmm5[9],xmm11[10],xmm5[10],xmm11[11],xmm5[11],xmm11[12],xmm5[12],xmm11[13],xmm5[13],xmm11[14],xmm5[14],xmm11[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm11[0,3,2,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm6[0,1,3,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,7,7,7,7]
; SSE-NEXT:    pand %xmm0, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm0
; SSE-NEXT:    por %xmm6, %xmm0
; SSE-NEXT:    pand %xmm3, %xmm4
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pandn %xmm0, %xmm3
; SSE-NEXT:    por %xmm4, %xmm3
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm5[8],xmm0[9],xmm5[9],xmm0[10],xmm5[10],xmm0[11],xmm5[11],xmm0[12],xmm5[12],xmm0[13],xmm5[13],xmm0[14],xmm5[14],xmm0[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,1]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,5,6,5]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm5[0],xmm10[1],xmm5[1],xmm10[2],xmm5[2],xmm10[3],xmm5[3],xmm10[4],xmm5[4],xmm10[5],xmm5[5],xmm10[6],xmm5[6],xmm10[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm10[0,2,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,7,7]
; SSE-NEXT:    pand %xmm1, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm1
; SSE-NEXT:    por %xmm2, %xmm1
; SSE-NEXT:    movdqa %xmm14, %xmm2
; SSE-NEXT:    pand %xmm14, %xmm3
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    por %xmm3, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm8 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm13 = [255,255,255,255,255,0,0,0,0,0,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm13, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm11, %xmm0
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pand %xmm13, %xmm0
; SSE-NEXT:    por %xmm2, %xmm0
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    movdqa {{.*#+}} xmm6 = [65535,0,65535,65535,0,65535,65535,0]
; SSE-NEXT:    pand %xmm6, %xmm9
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm9, %xmm2
; SSE-NEXT:    pand %xmm8, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,2,1,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,6,5,4,7]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535,65535,0,0,0]
; SSE-NEXT:    movdqa %xmm4, %xmm1
; SSE-NEXT:    pandn %xmm2, %xmm1
; SSE-NEXT:    pand %xmm4, %xmm0
; SSE-NEXT:    por %xmm0, %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm12, %xmm0
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
; SSE-NEXT:    packuswb %xmm2, %xmm0
; SSE-NEXT:    movdqa %xmm13, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm3, %xmm0
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pand %xmm13, %xmm0
; SSE-NEXT:    por %xmm2, %xmm0
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    pand %xmm6, %xmm1
; SSE-NEXT:    movdqa %xmm6, %xmm7
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm1, %xmm2
; SSE-NEXT:    pand %xmm8, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,2,1,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,6,5,4,7]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm4, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    pand %xmm4, %xmm0
; SSE-NEXT:    por %xmm0, %xmm6
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
; SSE-NEXT:    packuswb %xmm2, %xmm0
; SSE-NEXT:    movdqa %xmm13, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm14, %xmm0
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pand %xmm13, %xmm0
; SSE-NEXT:    por %xmm2, %xmm0
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    pand %xmm7, %xmm15
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm15, %xmm2
; SSE-NEXT:    pand %xmm8, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,2,1,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,6,5,4,7]
; SSE-NEXT:    packuswb %xmm2, %xmm2
; SSE-NEXT:    movdqa %xmm4, %xmm6
; SSE-NEXT:    pandn %xmm2, %xmm6
; SSE-NEXT:    pand %xmm4, %xmm0
; SSE-NEXT:    por %xmm0, %xmm6
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,1,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[2,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
; SSE-NEXT:    packuswb %xmm2, %xmm0
; SSE-NEXT:    movdqa %xmm13, %xmm2
; SSE-NEXT:    pandn %xmm0, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    por {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pand %xmm8, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm0[2,1,0,3,4,5,6,7]
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    pand %xmm13, %xmm6
; SSE-NEXT:    por %xmm2, %xmm6
; SSE-NEXT:    movdqa %xmm7, %xmm2
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pand %xmm7, %xmm0
; SSE-NEXT:    pandn {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    por %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm8, %xmm0
; SSE-NEXT:    pand %xmm2, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,1,2,1,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,1,0,3,4,5,6,7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,4,7]
; SSE-NEXT:    packuswb %xmm0, %xmm2
; SSE-NEXT:    movdqa %xmm4, %xmm7
; SSE-NEXT:    pandn %xmm2, %xmm7
; SSE-NEXT:    pand %xmm4, %xmm6
; SSE-NEXT:    por %xmm6, %xmm7
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    movdqa %xmm10, %xmm2
; SSE-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm5[8],xmm2[9],xmm5[9],xmm2[10],xmm5[10],xmm2[11],xmm5[11],xmm2[12],xmm5[12],xmm2[13],xmm5[13],xmm2[14],xmm5[14],xmm2[15],xmm5[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1],xmm0[2],xmm5[2],xmm0[3],xmm5[3],xmm0[4],xmm5[4],xmm0[5],xmm5[5],xmm0[6],xmm5[6],xmm0[7],xmm5[7]
; SSE-NEXT:    movdqa %xmm0, %xmm6
; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm2[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm2[2,3]
; SSE-NEXT:    psrlq $48, %xmm2
; SSE-NEXT:    psrldq {{.*#+}} xmm6 = xmm6[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1],xmm6[2],xmm2[2],xmm6[3],xmm2[3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm0[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,4,5,7]
; SSE-NEXT:    packuswb %xmm6, %xmm2
; SSE-NEXT:    movdqa %xmm13, %xmm8
; SSE-NEXT:    pandn %xmm2, %xmm8
; SSE-NEXT:    movdqa %xmm11, %xmm2
; SSE-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm5[8],xmm2[9],xmm5[9],xmm2[10],xmm5[10],xmm2[11],xmm5[11],xmm2[12],xmm5[12],xmm2[13],xmm5[13],xmm2[14],xmm5[14],xmm2[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa {{.*#+}} xmm0 = [65535,0,65535,65535,0,65535,65535,65535]
; SSE-NEXT:    movdqa %xmm0, %xmm10
; SSE-NEXT:    pandn %xmm2, %xmm10
; SSE-NEXT:    punpcklbw {{.*#+}} xmm11 = xmm11[0],xmm5[0],xmm11[1],xmm5[1],xmm11[2],xmm5[2],xmm11[3],xmm5[3],xmm11[4],xmm5[4],xmm11[5],xmm5[5],xmm11[6],xmm5[6],xmm11[7],xmm5[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm2 = xmm11[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm2[3,1,1,2,4,5,6,7]
; SSE-NEXT:    pand %xmm0, %xmm6
; SSE-NEXT:    por %xmm10, %xmm6
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    pand %xmm13, %xmm6
; SSE-NEXT:    por %xmm8, %xmm6
; SSE-NEXT:    movdqa %xmm9, %xmm2
; SSE-NEXT:    punpckhbw {{.*#+}} xmm9 = xmm9[8],xmm5[8],xmm9[9],xmm5[9],xmm9[10],xmm5[10],xmm9[11],xmm5[11],xmm9[12],xmm5[12],xmm9[13],xmm5[13],xmm9[14],xmm5[14],xmm9[15],xmm5[15]
; SSE-NEXT:    pshufhw {{.*#+}} xmm8 = xmm9[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm8[0,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm8 = xmm8[0,1,2,3,5,5,7,4]
; SSE-NEXT:    movdqa {{.*#+}} xmm11 = [65535,65535,65535,65535,0,65535,0,0]
; SSE-NEXT:    movdqa %xmm11, %xmm10
; SSE-NEXT:    pandn %xmm8, %xmm10
; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,3,1,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,1,3,4,5,6,7]
; SSE-NEXT:    pand %xmm11, %xmm2
; SSE-NEXT:    por %xmm2, %xmm10
; SSE-NEXT:    packuswb %xmm10, %xmm8
; SSE-NEXT:    movdqa %xmm4, %xmm2
; SSE-NEXT:    pandn %xmm8, %xmm2
; SSE-NEXT:    pand %xmm4, %xmm6
; SSE-NEXT:    por %xmm6, %xmm2
; SSE-NEXT:    movdqa %xmm12, %xmm9
; SSE-NEXT:    movdqa %xmm12, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm5[8],xmm6[9],xmm5[9],xmm6[10],xmm5[10],xmm6[11],xmm5[11],xmm6[12],xmm5[12],xmm6[13],xmm5[13],xmm6[14],xmm5[14],xmm6[15],xmm5[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm9 = xmm9[0],xmm5[0],xmm9[1],xmm5[1],xmm9[2],xmm5[2],xmm9[3],xmm5[3],xmm9[4],xmm5[4],xmm9[5],xmm5[5],xmm9[6],xmm5[6],xmm9[7],xmm5[7]
; SSE-NEXT:    movdqa %xmm9, %xmm8
; SSE-NEXT:    shufps {{.*#+}} xmm9 = xmm9[1,0],xmm6[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm9 = xmm9[2,0],xmm6[2,3]
; SSE-NEXT:    psrlq $48, %xmm6
; SSE-NEXT:    psrldq {{.*#+}} xmm8 = xmm8[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm6[0],xmm8[1],xmm6[1],xmm8[2],xmm6[2],xmm8[3],xmm6[3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm9[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,4,4,5,7]
; SSE-NEXT:    packuswb %xmm8, %xmm6
; SSE-NEXT:    movdqa %xmm13, %xmm8
; SSE-NEXT:    pandn %xmm6, %xmm8
; SSE-NEXT:    movdqa %xmm3, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm5[8],xmm6[9],xmm5[9],xmm6[10],xmm5[10],xmm6[11],xmm5[11],xmm6[12],xmm5[12],xmm6[13],xmm5[13],xmm6[14],xmm5[14],xmm6[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa %xmm0, %xmm10
; SSE-NEXT:    pandn %xmm6, %xmm10
; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1],xmm3[2],xmm5[2],xmm3[3],xmm5[3],xmm3[4],xmm5[4],xmm3[5],xmm5[5],xmm3[6],xmm5[6],xmm3[7],xmm5[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm3[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm6[3,1,1,2,4,5,6,7]
; SSE-NEXT:    pand %xmm0, %xmm6
; SSE-NEXT:    por %xmm10, %xmm6
; SSE-NEXT:    packuswb %xmm6, %xmm6
; SSE-NEXT:    pand %xmm13, %xmm6
; SSE-NEXT:    por %xmm8, %xmm6
; SSE-NEXT:    movdqa %xmm1, %xmm8
; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm5[8],xmm1[9],xmm5[9],xmm1[10],xmm5[10],xmm1[11],xmm5[11],xmm1[12],xmm5[12],xmm1[13],xmm5[13],xmm1[14],xmm5[14],xmm1[15],xmm5[15]
; SSE-NEXT:    pshufhw {{.*#+}} xmm10 = xmm1[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[0,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm10 = xmm10[0,1,2,3,5,5,7,4]
; SSE-NEXT:    movdqa %xmm11, %xmm12
; SSE-NEXT:    pandn %xmm10, %xmm12
; SSE-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm5[0],xmm8[1],xmm5[1],xmm8[2],xmm5[2],xmm8[3],xmm5[3],xmm8[4],xmm5[4],xmm8[5],xmm5[5],xmm8[6],xmm5[6],xmm8[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm8[0,3,1,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm8 = xmm8[0,1,1,3,4,5,6,7]
; SSE-NEXT:    pand %xmm11, %xmm8
; SSE-NEXT:    por %xmm8, %xmm12
; SSE-NEXT:    packuswb %xmm12, %xmm8
; SSE-NEXT:    movdqa %xmm4, %xmm12
; SSE-NEXT:    pandn %xmm8, %xmm12
; SSE-NEXT:    pand %xmm4, %xmm6
; SSE-NEXT:    por %xmm6, %xmm12
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm9, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm5[8],xmm6[9],xmm5[9],xmm6[10],xmm5[10],xmm6[11],xmm5[11],xmm6[12],xmm5[12],xmm6[13],xmm5[13],xmm6[14],xmm5[14],xmm6[15],xmm5[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm9 = xmm9[0],xmm5[0],xmm9[1],xmm5[1],xmm9[2],xmm5[2],xmm9[3],xmm5[3],xmm9[4],xmm5[4],xmm9[5],xmm5[5],xmm9[6],xmm5[6],xmm9[7],xmm5[7]
; SSE-NEXT:    movdqa %xmm9, %xmm8
; SSE-NEXT:    shufps {{.*#+}} xmm9 = xmm9[1,0],xmm6[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm9 = xmm9[2,0],xmm6[2,3]
; SSE-NEXT:    psrlq $48, %xmm6
; SSE-NEXT:    psrldq {{.*#+}} xmm8 = xmm8[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm6[0],xmm8[1],xmm6[1],xmm8[2],xmm6[2],xmm8[3],xmm6[3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm9[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,4,4,5,7]
; SSE-NEXT:    packuswb %xmm8, %xmm6
; SSE-NEXT:    movdqa %xmm13, %xmm8
; SSE-NEXT:    pandn %xmm6, %xmm8
; SSE-NEXT:    movdqa %xmm14, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm5[8],xmm6[9],xmm5[9],xmm6[10],xmm5[10],xmm6[11],xmm5[11],xmm6[12],xmm5[12],xmm6[13],xmm5[13],xmm6[14],xmm5[14],xmm6[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm6[0,1,2,3,5,5,5,5]
; SSE-NEXT:    movdqa %xmm0, %xmm9
; SSE-NEXT:    pandn %xmm6, %xmm9
; SSE-NEXT:    punpcklbw {{.*#+}} xmm14 = xmm14[0],xmm5[0],xmm14[1],xmm5[1],xmm14[2],xmm5[2],xmm14[3],xmm5[3],xmm14[4],xmm5[4],xmm14[5],xmm5[5],xmm14[6],xmm5[6],xmm14[7],xmm5[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm6 = xmm14[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm10 = xmm6[3,1,1,2,4,5,6,7]
; SSE-NEXT:    pand %xmm0, %xmm10
; SSE-NEXT:    por %xmm9, %xmm10
; SSE-NEXT:    packuswb %xmm10, %xmm10
; SSE-NEXT:    pand %xmm13, %xmm10
; SSE-NEXT:    por %xmm8, %xmm10
; SSE-NEXT:    movdqa %xmm15, %xmm6
; SSE-NEXT:    punpckhbw {{.*#+}} xmm15 = xmm15[8],xmm5[8],xmm15[9],xmm5[9],xmm15[10],xmm5[10],xmm15[11],xmm5[11],xmm15[12],xmm5[12],xmm15[13],xmm5[13],xmm15[14],xmm5[14],xmm15[15],xmm5[15]
; SSE-NEXT:    pshufhw {{.*#+}} xmm8 = xmm15[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm8[0,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm8 = xmm8[0,1,2,3,5,5,7,4]
; SSE-NEXT:    movdqa %xmm11, %xmm9
; SSE-NEXT:    pandn %xmm8, %xmm9
; SSE-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm5[0],xmm6[1],xmm5[1],xmm6[2],xmm5[2],xmm6[3],xmm5[3],xmm6[4],xmm5[4],xmm6[5],xmm5[5],xmm6[6],xmm5[6],xmm6[7],xmm5[7]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[0,3,1,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm6 = xmm6[0,1,1,3,4,5,6,7]
; SSE-NEXT:    pand %xmm11, %xmm6
; SSE-NEXT:    por %xmm6, %xmm9
; SSE-NEXT:    packuswb %xmm9, %xmm8
; SSE-NEXT:    movdqa %xmm4, %xmm6
; SSE-NEXT:    pandn %xmm8, %xmm6
; SSE-NEXT:    pand %xmm4, %xmm10
; SSE-NEXT:    por %xmm10, %xmm6
; SSE-NEXT:    movdqa (%rsp), %xmm1 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm1, %xmm8
; SSE-NEXT:    punpckhbw {{.*#+}} xmm8 = xmm8[8],xmm5[8],xmm8[9],xmm5[9],xmm8[10],xmm5[10],xmm8[11],xmm5[11],xmm8[12],xmm5[12],xmm8[13],xmm5[13],xmm8[14],xmm5[14],xmm8[15],xmm5[15]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3],xmm1[4],xmm5[4],xmm1[5],xmm5[5],xmm1[6],xmm5[6],xmm1[7],xmm5[7]
; SSE-NEXT:    movdqa %xmm1, %xmm9
; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm8[0,0]
; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm8[2,3]
; SSE-NEXT:    psrlq $48, %xmm8
; SSE-NEXT:    psrldq {{.*#+}} xmm9 = xmm9[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm8[0],xmm9[1],xmm8[1],xmm9[2],xmm8[2],xmm9[3],xmm8[3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm8 = xmm1[3,1,2,3,4,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm8[0,1,0,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm8 = xmm8[0,1,2,3,4,4,5,7]
; SSE-NEXT:    packuswb %xmm9, %xmm8
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm14, %xmm9
; SSE-NEXT:    punpckhbw {{.*#+}} xmm9 = xmm9[8],xmm5[8],xmm9[9],xmm5[9],xmm9[10],xmm5[10],xmm9[11],xmm5[11],xmm9[12],xmm5[12],xmm9[13],xmm5[13],xmm9[14],xmm5[14],xmm9[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[1,1,2,3]
; SSE-NEXT:    pshufhw {{.*#+}} xmm9 = xmm9[0,1,2,3,5,5,5,5]
; SSE-NEXT:    punpcklbw {{.*#+}} xmm14 = xmm14[0],xmm5[0],xmm14[1],xmm5[1],xmm14[2],xmm5[2],xmm14[3],xmm5[3],xmm14[4],xmm5[4],xmm14[5],xmm5[5],xmm14[6],xmm5[6],xmm14[7],xmm5[7]
; SSE-NEXT:    pshufhw {{.*#+}} xmm10 = xmm14[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[0,2,2,3]
; SSE-NEXT:    pshuflw {{.*#+}} xmm10 = xmm10[3,1,1,2,4,5,6,7]
; SSE-NEXT:    pand %xmm0, %xmm10
; SSE-NEXT:    pandn %xmm9, %xmm0
; SSE-NEXT:    por %xmm10, %xmm0
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    pand %xmm13, %xmm0
; SSE-NEXT:    pandn %xmm8, %xmm13
; SSE-NEXT:    por %xmm13, %xmm0
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm14, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3],xmm1[4],xmm5[4],xmm1[5],xmm5[5],xmm1[6],xmm5[6],xmm1[7],xmm5[7]
; SSE-NEXT:    punpckhbw {{.*#+}} xmm14 = xmm14[8],xmm5[8],xmm14[9],xmm5[9],xmm14[10],xmm5[10],xmm14[11],xmm5[11],xmm14[12],xmm5[12],xmm14[13],xmm5[13],xmm14[14],xmm5[14],xmm14[15],xmm5[15]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,3,1,1]
; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,1,1,3,4,5,6,7]
; SSE-NEXT:    pand %xmm11, %xmm1
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm14[0,1,2,3,7,5,6,7]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,1,2,0]
; SSE-NEXT:    pshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,5,5,7,4]
; SSE-NEXT:    pandn %xmm5, %xmm11
; SSE-NEXT:    por %xmm1, %xmm11
; SSE-NEXT:    pand %xmm4, %xmm0
; SSE-NEXT:    packuswb %xmm11, %xmm1
; SSE-NEXT:    pandn %xmm1, %xmm4
; SSE-NEXT:    por %xmm0, %xmm4
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 16(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 32(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 48(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, (%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 16(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 32(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 48(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, (%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 16(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 32(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 48(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, (%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 16(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 32(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 48(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, (%r8)
; SSE-NEXT:    movdqa %xmm7, 16(%r9)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 32(%r9)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 48(%r9)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, (%r9)
; SSE-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT:    movdqa %xmm4, 16(%rax)
; SSE-NEXT:    movdqa %xmm6, 32(%rax)
; SSE-NEXT:    movdqa %xmm12, 48(%rax)
; SSE-NEXT:    movdqa %xmm2, (%rax)
; SSE-NEXT:    addq $808, %rsp # imm = 0x328
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i8_stride6_vf64:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    subq $808, %rsp # imm = 0x328
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm6 = [0,0,0,128,128,128,4,10,0,0,0,128,128,128,4,10]
; AVX1-ONLY-NEXT:    # xmm6 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm7 = [0,0,0,2,8,14,128,128,0,0,0,2,8,14,128,128]
; AVX1-ONLY-NEXT:    # xmm7 = mem[0,0]
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm2 = [0,0,4,10,0,0,4,10,0,0,4,10,0,0,4,10]
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm3 = [0,6,12,0,0,6,12,0,0,6,12,0,0,6,12,0]
; AVX1-ONLY-NEXT:    vmovdqa 128(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm4
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, %xmm8
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 144(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm10
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; AVX1-ONLY-NEXT:    vmovdqa 176(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm2, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm9
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 160(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm11
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm2, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,u,u,255,255,255,255,255,0,0,0,0,0>
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 320(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm12
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 336(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm13
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; AVX1-ONLY-NEXT:    vmovdqa 368(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm14
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 352(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm4, %xmm3
; AVX1-ONLY-NEXT:    vmovdqa %xmm4, %xmm15
; AVX1-ONLY-NEXT:    vmovdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm2 = [0,0,5,11,0,0,5,11,0,0,5,11,0,0,5,11]
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm3 = [1,7,13,0,1,7,13,0,1,7,13,0,1,7,13,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm8, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm4
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm10, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm5
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm3 = [0,0,0,128,128,128,5,11,0,0,0,128,128,128,5,11]
; AVX1-ONLY-NEXT:    # xmm3 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm6 = [0,0,0,3,9,15,128,128,0,0,0,3,9,15,128,128]
; AVX1-ONLY-NEXT:    # xmm6 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm9, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm7
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm11, %xmm3
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm12, %xmm1
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm13, %xmm2
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm14, %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm15, %xmm3
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 32(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vmovdqa 48(%rdi), %xmm11
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm0 = [2,8,14,0,2,8,14,0,2,8,14,0,2,8,14,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm11, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, %xmm8
; AVX1-ONLY-NEXT:    vmovdqa %xmm11, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm15 = [0,0,6,12,0,0,6,12,0,0,6,12,0,0,6,12]
; AVX1-ONLY-NEXT:    vpshufb %xmm15, %xmm4, %xmm3
; AVX1-ONLY-NEXT:    vmovdqa %xmm4, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0]
; AVX1-ONLY-NEXT:    vmovdqa (%rdi), %xmm10
; AVX1-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm14
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm4 = [128,128,128,4,10,0,0,0,128,128,128,4,10,0,0,0]
; AVX1-ONLY-NEXT:    # xmm4 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm5 = [2,8,14,128,128,0,0,0,2,8,14,128,128,0,0,0]
; AVX1-ONLY-NEXT:    # xmm5 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm14, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm14, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa %xmm4, %xmm12
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm10, %xmm4
; AVX1-ONLY-NEXT:    vmovdqa %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa %xmm5, %xmm6
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm4, %xmm4
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm13 = <0,0,0,0,0,255,255,255,255,255,255,u,u,u,u,u>
; AVX1-ONLY-NEXT:    vpblendvb %xmm13, %xmm3, %xmm4, %xmm3
; AVX1-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 224(%rdi), %xmm5
; AVX1-ONLY-NEXT:    vmovdqa 240(%rdi), %xmm9
; AVX1-ONLY-NEXT:    vpshufb %xmm8, %xmm9, %xmm3
; AVX1-ONLY-NEXT:    vmovdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm15, %xmm5, %xmm4
; AVX1-ONLY-NEXT:    vmovdqa %xmm5, %xmm8
; AVX1-ONLY-NEXT:    vmovdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm4[0],xmm3[0]
; AVX1-ONLY-NEXT:    vmovdqa 208(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovdqa 192(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm1, %xmm4
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm2, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpor %xmm4, %xmm5, %xmm4
; AVX1-ONLY-NEXT:    vpblendvb %xmm13, %xmm3, %xmm4, %xmm3
; AVX1-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm3 = [3,9,15,0,3,9,15,0,3,9,15,0,3,9,15,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm11, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm11
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm12 = [0,1,7,13,0,1,7,13,0,1,7,13,0,1,7,13]
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm0, %xmm6
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm5 = xmm6[0],xmm5[0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm4 = [128,128,128,5,11,0,0,0,128,128,128,5,11,0,0,0]
; AVX1-ONLY-NEXT:    # xmm4 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm3 = [3,9,15,128,128,0,0,0,3,9,15,128,128,0,0,0]
; AVX1-ONLY-NEXT:    # xmm3 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm14, %xmm6
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm10, %xmm7
; AVX1-ONLY-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX1-ONLY-NEXT:    vpblendvb %xmm13, %xmm5, %xmm6, %xmm0
; AVX1-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm11, %xmm9, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa %xmm11, %xmm14
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm8, %xmm6
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm5 = xmm6[0],xmm5[0]
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm1, %xmm6
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm2, %xmm7
; AVX1-ONLY-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX1-ONLY-NEXT:    vpblendvb %xmm13, %xmm5, %xmm6, %xmm0
; AVX1-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm1 = [2,8,14,0,2,8,14,0,2,8,14,0,2,8,14,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm9, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm15, %xmm8, %xmm5
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm5[0],xmm2[0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm7 = [0,0,0,4,10,128,128,128,0,0,0,4,10,128,128,128]
; AVX1-ONLY-NEXT:    # xmm7 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm3 = [0,0,0,128,128,0,6,12,0,0,0,128,128,0,6,12]
; AVX1-ONLY-NEXT:    # xmm3 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm11, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm10, %xmm6
; AVX1-ONLY-NEXT:    vpor %xmm5, %xmm6, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm6 = <u,u,u,u,u,255,255,255,255,255,255,0,0,0,0,0>
; AVX1-ONLY-NEXT:    vpblendvb %xmm6, %xmm2, %xmm5, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm13, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm15, %xmm5, %xmm1
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm7, %xmm15, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm7, %xmm2
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vpblendvb %xmm6, %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm14, %xmm9, %xmm0
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm8, %xmm1
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm2 = [0,0,0,5,11,128,128,128,0,0,0,5,11,128,128,128]
; AVX1-ONLY-NEXT:    # xmm2 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm8 = [0,0,0,128,128,1,7,13,0,0,0,128,128,1,7,13]
; AVX1-ONLY-NEXT:    # xmm8 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm11, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm8, %xmm10, %xmm2
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vpblendvb %xmm6, %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm14, %xmm13, %xmm0
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm5, %xmm1
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm15, %xmm1
; AVX1-ONLY-NEXT:    vpshufb %xmm8, %xmm7, %xmm2
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vpblendvb %xmm6, %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm1 = [4,10,128,128,128,0,0,0,4,10,128,128,128,0,0,0]
; AVX1-ONLY-NEXT:    # xmm1 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm10, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, %xmm5
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm2 = [128,128,0,6,12,0,0,0,128,128,0,6,12,0,0,0]
; AVX1-ONLY-NEXT:    # xmm2 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm4, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm9
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm1
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm2 = [4,10,0,0,4,10,0,0,4,10,0,0,4,10,0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm12, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm6
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm3 = [0,2,8,14,0,2,8,14,0,2,8,14,0,2,8,14]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm14, %xmm2
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm2 = xmm2[1],xmm0[1]
; AVX1-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <255,255,255,255,255,0,0,0,0,0,u,u,u,u,u,u>
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm15, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm9, %xmm8, %xmm2
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm7, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm6, %xmm3
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm2 = xmm3[1],xmm2[1]
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm2 = [5,11,128,128,128,0,0,0,5,11,128,128,128,0,0,0]
; AVX1-ONLY-NEXT:    # xmm2 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm10, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm9
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm3 = [128,128,1,7,13,0,0,0,128,128,1,7,13,0,0,0]
; AVX1-ONLY-NEXT:    # xmm3 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm4, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm10
; AVX1-ONLY-NEXT:    vmovdqa %xmm4, %xmm13
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm3 = [5,11,0,0,5,11,0,0,5,11,0,0,5,11,0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm12, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm11
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm4 = [0,3,9,15,0,3,9,15,0,3,9,15,0,3,9,15]
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm14, %xmm3
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm2 = xmm3[1],xmm2[1]
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm9, %xmm15, %xmm1
; AVX1-ONLY-NEXT:    vpshufb %xmm10, %xmm8, %xmm2
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vpshufb %xmm11, %xmm7, %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm6, %xmm3
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm2 = xmm3[1],xmm2[1]
; AVX1-ONLY-NEXT:    vpblendvb %xmm0, %xmm1, %xmm2, %xmm0
; AVX1-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm1 = [128,128,128,2,8,14,0,0,128,128,128,2,8,14,0,0]
; AVX1-ONLY-NEXT:    # xmm1 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa 112(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, %xmm3
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm2 = [0,6,12,128,128,128,0,0,0,6,12,128,128,128,0,0]
; AVX1-ONLY-NEXT:    # xmm2 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa 96(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm5
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa 80(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm10 = [0,0,0,128,128,128,4,10,0,0,0,128,128,128,4,10]
; AVX1-ONLY-NEXT:    # xmm10 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm10, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm9
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 64(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm11 = [0,0,0,2,8,14,128,128,0,0,0,2,8,14,128,128]
; AVX1-ONLY-NEXT:    # xmm11 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm11, %xmm2, %xmm4
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm4, %xmm1
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm14[u,u,4,10,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm4 = xmm12[u,u,u,u,0,6,12,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm13, %xmm4
; AVX1-ONLY-NEXT:    vmovdqa %xmm3, %xmm14
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX1-ONLY-NEXT:    vmovdqa %xmm5, %xmm12
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm2, %xmm5
; AVX1-ONLY-NEXT:    vpor %xmm4, %xmm5, %xmm4
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm4 = xmm4[0,1,2],xmm0[3,4,5],xmm4[6,7]
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm13 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255]
; AVX1-ONLY-NEXT:    vandnps %ymm1, %ymm13, %ymm1
; AVX1-ONLY-NEXT:    vandps %ymm4, %ymm13, %ymm4
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm4, %ymm4
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm3 = [65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,0,0,0,0,0]
; AVX1-ONLY-NEXT:    vandps %ymm3, %ymm4, %ymm4
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm5 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm5, %ymm3, %ymm5
; AVX1-ONLY-NEXT:    vorps %ymm5, %ymm4, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa 304(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm14, %xmm0, %xmm4
; AVX1-ONLY-NEXT:    vmovdqa 288(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm0, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, %xmm12
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpor %xmm4, %xmm5, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa 272(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm10, %xmm0, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa 256(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vpshufb %xmm11, %xmm2, %xmm10
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm10, %xmm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm6[u,u,4,10,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm10 = xmm7[u,u,u,u,0,6,12,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm10[0],xmm1[0],xmm10[1],xmm1[1]
; AVX1-ONLY-NEXT:    vpshufb %xmm14, %xmm8, %xmm7
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = xmm15[0,6,12],zero,zero,zero,xmm15[u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm6, %xmm6
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm1 = xmm6[0,1,2],xmm1[3,4,5],xmm6[6,7]
; AVX1-ONLY-NEXT:    vandnps %ymm0, %ymm13, %ymm0
; AVX1-ONLY-NEXT:    vandps %ymm1, %ymm13, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm0, %ymm1, %ymm0
; AVX1-ONLY-NEXT:    vandps %ymm3, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm1 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm1, %ymm3, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm0 = [128,128,128,3,9,15,0,0,128,128,128,3,9,15,0,0]
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa (%rsp), %xmm14 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm14, %xmm1
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm6 = [1,7,13,128,128,128,0,0,1,7,13,128,128,128,0,0]
; AVX1-ONLY-NEXT:    # xmm6 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm15, %xmm7
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm7, %xmm1
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm8 = [0,0,0,128,128,128,5,11,0,0,0,128,128,128,5,11]
; AVX1-ONLY-NEXT:    # xmm8 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm8, %xmm9, %xmm7
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm9 = [0,0,0,3,9,15,128,128,0,0,0,3,9,15,128,128]
; AVX1-ONLY-NEXT:    # xmm9 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm9, %xmm2, %xmm10
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm10, %xmm7
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm7, %ymm1
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm4 = [0,0,5,11,0,0,5,11,0,0,5,11,0,0,5,11]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm5, %xmm7
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm5 = [1,7,13,0,1,7,13,0,1,7,13,0,1,7,13,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm10, %xmm10
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm7 = xmm10[0],xmm7[0],xmm10[1],xmm7[1]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm10, %xmm10
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm11, %xmm11
; AVX1-ONLY-NEXT:    vpor %xmm10, %xmm11, %xmm10
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm7 = xmm10[0,1,2],xmm7[3,4,5],xmm10[6,7]
; AVX1-ONLY-NEXT:    vandnps %ymm1, %ymm13, %ymm1
; AVX1-ONLY-NEXT:    vandps %ymm7, %ymm13, %ymm7
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm7, %ymm1
; AVX1-ONLY-NEXT:    vandps %ymm3, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm7 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm7, %ymm3, %ymm7
; AVX1-ONLY-NEXT:    vorps %ymm7, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm11, %xmm1
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm12, %xmm7
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm7, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm8, %xmm12, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm9, %xmm8, %xmm10
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm10, %xmm7
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm7, %ymm1
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm7, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm4, %xmm10
; AVX1-ONLY-NEXT:    vpunpckldq {{.*#+}} xmm7 = xmm10[0],xmm7[0],xmm10[1],xmm7[1]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm4, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm4, %xmm6
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm6, %xmm0
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm7[3,4,5],xmm0[6,7]
; AVX1-ONLY-NEXT:    vandnps %ymm1, %ymm13, %ymm1
; AVX1-ONLY-NEXT:    vandps %ymm0, %ymm13, %ymm0
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vandps %ymm3, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm1 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm1, %ymm3, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm3 = [128,128,128,4,10,0,0,0,128,128,128,4,10,0,0,0]
; AVX1-ONLY-NEXT:    # xmm3 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa %xmm14, %xmm13
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm14, %xmm0
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm4 = [2,8,14,128,128,0,0,0,2,8,14,128,128,0,0,0]
; AVX1-ONLY-NEXT:    # xmm4 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm15, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm15, %xmm4
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm15 = [0,0,0,4,10,128,128,128,0,0,0,4,10,128,128,128]
; AVX1-ONLY-NEXT:    # xmm15 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm15, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm2, %xmm5
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm6 = [0,0,0,128,128,0,6,12,0,0,0,128,128,0,6,12]
; AVX1-ONLY-NEXT:    # xmm6 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm7, %xmm2
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm0 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255]
; AVX1-ONLY-NEXT:    vandnps %ymm1, %ymm0, %ymm1
; AVX1-ONLY-NEXT:    vandps {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm2 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm2, %ymm1
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm14 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX1-ONLY-NEXT:    vandps %ymm1, %ymm14, %ymm1
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm2 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm2, %ymm14, %ymm2
; AVX1-ONLY-NEXT:    vorps %ymm2, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa %xmm11, %xmm9
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm11, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = xmm10[2,8,14],zero,zero,xmm10[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm15, %xmm8, %xmm3
; AVX1-ONLY-NEXT:    vmovdqa %xmm8, %xmm15
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm12, %xmm6
; AVX1-ONLY-NEXT:    vmovdqa %xmm12, %xmm11
; AVX1-ONLY-NEXT:    vpor %xmm3, %xmm6, %xmm3
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm2, %ymm3, %ymm2
; AVX1-ONLY-NEXT:    vandnps %ymm2, %ymm0, %ymm2
; AVX1-ONLY-NEXT:    vandps {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm3 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vorps %ymm2, %ymm3, %ymm2
; AVX1-ONLY-NEXT:    vandps %ymm2, %ymm14, %ymm2
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm3 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm3, %ymm14, %ymm3
; AVX1-ONLY-NEXT:    vorps %ymm3, %ymm2, %ymm1
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm1 = [128,128,128,5,11,0,0,0,128,128,128,5,11,0,0,0]
; AVX1-ONLY-NEXT:    # xmm1 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm13, %xmm3
; AVX1-ONLY-NEXT:    vmovdqa %xmm13, %xmm1
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm2 = [3,9,15,128,128,0,0,0,3,9,15,128,128,0,0,0]
; AVX1-ONLY-NEXT:    # xmm2 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa %xmm4, %xmm13
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm4, %xmm6
; AVX1-ONLY-NEXT:    vpor %xmm3, %xmm6, %xmm3
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm4 = [0,0,0,5,11,128,128,128,0,0,0,5,11,128,128,128]
; AVX1-ONLY-NEXT:    # xmm4 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa %xmm5, %xmm12
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm5, %xmm6
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm5 = [0,0,0,128,128,1,7,13,0,0,0,128,128,1,7,13]
; AVX1-ONLY-NEXT:    # xmm5 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa %xmm7, %xmm8
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm7, %xmm7
; AVX1-ONLY-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm6, %ymm3
; AVX1-ONLY-NEXT:    vandnps %ymm3, %ymm0, %ymm3
; AVX1-ONLY-NEXT:    vandps {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm6 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vorps %ymm3, %ymm6, %ymm3
; AVX1-ONLY-NEXT:    vandps %ymm3, %ymm14, %ymm3
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm6 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm6, %ymm14, %ymm6
; AVX1-ONLY-NEXT:    vorps %ymm6, %ymm3, %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,zero,xmm9[5,11,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb %xmm2, %xmm10, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa %xmm10, %xmm9
; AVX1-ONLY-NEXT:    vpor %xmm6, %xmm7, %xmm6
; AVX1-ONLY-NEXT:    vpshufb %xmm4, %xmm15, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa %xmm15, %xmm4
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm11, %xmm10
; AVX1-ONLY-NEXT:    vmovdqa %xmm11, %xmm2
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm10, %xmm7
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm6, %ymm7, %ymm6
; AVX1-ONLY-NEXT:    vandnps %ymm6, %ymm0, %ymm6
; AVX1-ONLY-NEXT:    vandps {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vorps %ymm6, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vandps %ymm0, %ymm14, %ymm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm6 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vandnps %ymm6, %ymm14, %ymm6
; AVX1-ONLY-NEXT:    vmovaps %ymm14, %ymm11
; AVX1-ONLY-NEXT:    vorps %ymm6, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm14 = [4,10,128,128,128,0,0,0,4,10,128,128,128,0,0,0]
; AVX1-ONLY-NEXT:    # xmm14 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm14, %xmm13, %xmm0
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm15 = [128,128,0,6,12,0,0,0,128,128,0,6,12,0,0,0]
; AVX1-ONLY-NEXT:    # xmm15 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm15, %xmm1, %xmm7
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm7, %xmm0
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm1 = [0,0,128,128,128,2,8,14,0,0,128,128,128,2,8,14]
; AVX1-ONLY-NEXT:    # xmm1 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm8, %xmm7
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm5 = [0,0,0,6,12,128,128,128,0,0,0,6,12,128,128,128]
; AVX1-ONLY-NEXT:    # xmm5 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm12, %xmm10
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm10, %xmm7
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm0, %ymm7, %ymm0
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm10 = [0,0,0,0,0,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535]
; AVX1-ONLY-NEXT:    vandnps {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vandps %ymm0, %ymm10, %ymm0
; AVX1-ONLY-NEXT:    vorps %ymm7, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm15, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm8, %xmm3
; AVX1-ONLY-NEXT:    vpor %xmm7, %xmm3, %xmm3
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm12 = [4,10,0,0,4,10,0,0,4,10,0,0,4,10,0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm14, %xmm7
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm13 = [0,2,8,14,0,2,8,14,0,2,8,14,0,2,8,14]
; AVX1-ONLY-NEXT:    vpshufb %xmm13, %xmm12, %xmm6
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm6 = xmm6[1],xmm7[1]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm3 = xmm6[0,1,2,3,4],xmm3[5,6,7]
; AVX1-ONLY-NEXT:    vandps %ymm0, %ymm11, %ymm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX1-ONLY-NEXT:    vandnps %ymm3, %ymm11, %ymm3
; AVX1-ONLY-NEXT:    vorps %ymm3, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm0 = xmm9[4,10],zero,zero,zero,xmm9[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm3 = zero,zero,xmm7[0,6,12,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm0, %xmm3, %xmm0
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm2, %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm4, %xmm6
; AVX1-ONLY-NEXT:    vpor %xmm3, %xmm6, %xmm3
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm0, %ymm3, %ymm0
; AVX1-ONLY-NEXT:    vandnps {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm3 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vandps %ymm0, %ymm10, %ymm0
; AVX1-ONLY-NEXT:    vorps %ymm3, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm1, %xmm11, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm9, %xmm1
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm1, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[u,u,u,u,u,u,u,u,4,10,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm13, %xmm4, %xmm3
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm2 = xmm3[1],xmm2[1]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3,4],xmm1[5,6,7]
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX1-ONLY-NEXT:    vandps %ymm2, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1
; AVX1-ONLY-NEXT:    vandnps %ymm1, %ymm2, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm1, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[5,11],zero,zero,zero,xmm0[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = zero,zero,xmm0[1,7,13,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm13 = [0,0,128,128,128,3,9,15,0,0,128,128,128,3,9,15]
; AVX1-ONLY-NEXT:    # xmm13 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm13, %xmm0, %xmm2
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm0 = [0,0,1,7,13,128,128,128,0,0,1,7,13,128,128,128]
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,0]
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm3, %xmm3
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
; AVX1-ONLY-NEXT:    vandnps {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm2 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vandps %ymm1, %ymm10, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm2, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vpshufb %xmm13, %xmm15, %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm8, %xmm3
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm8 = [5,11,0,0,5,11,0,0,5,11,0,0,5,11,0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm8, %xmm14, %xmm3
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm5 = [0,3,9,15,0,3,9,15,0,3,9,15,0,3,9,15]
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm12, %xmm6
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm3 = xmm6[1],xmm3[1]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3,4],xmm2[5,6,7]
; AVX1-ONLY-NEXT:    vmovaps {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX1-ONLY-NEXT:    vandps %ymm3, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm2
; AVX1-ONLY-NEXT:    vandnps %ymm2, %ymm3, %ymm2
; AVX1-ONLY-NEXT:    vmovaps %ymm3, %ymm12
; AVX1-ONLY-NEXT:    vorps %ymm2, %ymm1, %ymm6
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[5,11],zero,zero,zero,xmm1[u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpshufb {{.*#+}} xmm2 = zero,zero,xmm7[1,7,13,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm13, %xmm2, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm3, %xmm3
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
; AVX1-ONLY-NEXT:    vandnps {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm2 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    vandps %ymm1, %ymm10, %ymm1
; AVX1-ONLY-NEXT:    vorps %ymm2, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vpshufb %xmm13, %xmm11, %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm0, %xmm9, %xmm3
; AVX1-ONLY-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-ONLY-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vpshufb %xmm8, %xmm0, %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm5, %xmm4, %xmm4
; AVX1-ONLY-NEXT:    vpunpckhqdq {{.*#+}} xmm3 = xmm4[1],xmm3[1]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3,4],xmm2[5,6,7]
; AVX1-ONLY-NEXT:    vandps %ymm1, %ymm12, %ymm1
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm2
; AVX1-ONLY-NEXT:    vandnps %ymm2, %ymm12, %ymm2
; AVX1-ONLY-NEXT:    vorps %ymm2, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, 32(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, 32(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, 32(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, 32(%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, 32(%r9)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%r9)
; AVX1-ONLY-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX1-ONLY-NEXT:    vmovaps %ymm1, 32(%rax)
; AVX1-ONLY-NEXT:    vmovaps %ymm6, (%rax)
; AVX1-ONLY-NEXT:    addq $808, %rsp # imm = 0x328
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i8_stride6_vf64:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    subq $328, %rsp # imm = 0x148
; AVX2-ONLY-NEXT:    vmovdqa 192(%rdi), %ymm7
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm5
; AVX2-ONLY-NEXT:    vmovdqu %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 96(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm13 = <u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0>
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm0[0,1],ymm1[0,1]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[2,3]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendvb %ymm13, %ymm2, %ymm0, %ymm4
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm0 = ymm4[u,u,u,u,u,u,u,u,u,u,u,2,8,14,4,10,16,22,28,18,24,30,u,u,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm1 = <255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255>
; AVX2-ONLY-NEXT:    vpblendvb %ymm1, %ymm3, %ymm5, %ymm2
; AVX2-ONLY-NEXT:    vmovdqa %ymm3, %ymm5
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, (%rsp) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm10 = <0,6,12,128,128,128,4,10,128,128,128,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm10, %xmm2, %xmm9
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm2, %xmm3
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm11 = <128,128,128,2,8,14,128,128,0,6,12,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm11, %xmm3, %xmm12
; AVX2-ONLY-NEXT:    vpor %xmm9, %xmm12, %xmm9
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm12 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0]
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm9, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 224(%rdi), %ymm8
; AVX2-ONLY-NEXT:    vpblendvb %ymm1, %ymm7, %ymm8, %ymm14
; AVX2-ONLY-NEXT:    vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpshufb %xmm10, %xmm14, %xmm0
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm14, %xmm15
; AVX2-ONLY-NEXT:    vpshufb %xmm11, %xmm15, %xmm10
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm10, %xmm1
; AVX2-ONLY-NEXT:    vmovdqa 288(%rdi), %ymm11
; AVX2-ONLY-NEXT:    vmovdqa 256(%rdi), %ymm0
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm9 = ymm0[0,1],ymm11[0,1]
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm11 = ymm0[2,3],ymm11[2,3]
; AVX2-ONLY-NEXT:    vpblendvb %ymm13, %ymm9, %ymm11, %ymm13
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm0 = ymm13[u,u,u,u,u,u,u,u,u,u,u,2,8,14,4,10,16,22,28,18,24,30,u,u,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm1, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <1,7,13,128,128,128,5,11,128,128,128,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm0, %xmm2, %xmm1
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm2 = <128,128,128,3,9,15,128,128,1,7,13,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm2, %xmm3, %xmm3
; AVX2-ONLY-NEXT:    vpor %xmm1, %xmm3, %xmm1
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm3 = [1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11]
; AVX2-ONLY-NEXT:    vpshufb %ymm3, %ymm4, %ymm4
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm1, %ymm4, %ymm1
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpshufb %xmm0, %xmm14, %xmm0
; AVX2-ONLY-NEXT:    vpshufb %xmm2, %xmm15, %xmm1
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX2-ONLY-NEXT:    vpshufb %ymm3, %ymm13, %ymm1
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm0, %ymm1, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm13 = <255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255>
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendvb %ymm13, %ymm10, %ymm5, %ymm1
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm1, %xmm2
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm6 = <128,128,128,4,10,128,128,128,2,8,14,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm6, %xmm2, %xmm3
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm4 = <2,8,14,128,128,0,6,12,128,128,128,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm4, %xmm1, %xmm5
; AVX2-ONLY-NEXT:    vpor %xmm3, %xmm5, %xmm5
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm14 = [2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm0 = <255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0>
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendvb %ymm0, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpshufb %ymm14, %ymm3, %ymm15
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm5, %ymm15, %ymm5
; AVX2-ONLY-NEXT:    vmovdqu %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendvb %ymm13, %ymm8, %ymm7, %ymm5
; AVX2-ONLY-NEXT:    vmovdqa %ymm8, %ymm7
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm5, %xmm15
; AVX2-ONLY-NEXT:    vpshufb %xmm6, %xmm15, %xmm6
; AVX2-ONLY-NEXT:    vpshufb %xmm4, %xmm5, %xmm4
; AVX2-ONLY-NEXT:    vpor %xmm6, %xmm4, %xmm4
; AVX2-ONLY-NEXT:    vpblendvb %ymm0, %ymm11, %ymm9, %ymm0
; AVX2-ONLY-NEXT:    vpshufb %ymm14, %ymm0, %ymm6
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm4, %ymm6, %ymm4
; AVX2-ONLY-NEXT:    vmovdqu %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm4 = <128,128,128,5,11,128,128,128,3,9,15,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm6 = <3,9,15,128,128,1,7,13,128,128,128,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm6, %xmm1, %xmm1
; AVX2-ONLY-NEXT:    vpor %xmm2, %xmm1, %xmm1
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13]
; AVX2-ONLY-NEXT:    vpshufb %ymm2, %ymm3, %ymm3
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm1, %ymm3, %ymm1
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpshufb %xmm4, %xmm15, %xmm1
; AVX2-ONLY-NEXT:    vpshufb %xmm6, %xmm5, %xmm3
; AVX2-ONLY-NEXT:    vpor %xmm1, %xmm3, %xmm1
; AVX2-ONLY-NEXT:    vpshufb %ymm2, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendvb %ymm12, %ymm1, %ymm0, %ymm14
; AVX2-ONLY-NEXT:    vmovdqa 160(%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 128(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm2 = <u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u,255,255,0,0,u,u>
; AVX2-ONLY-NEXT:    vpblendvb %ymm2, %ymm0, %ymm3, %ymm1
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm1 = <255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255>
; AVX2-ONLY-NEXT:    vpblendvb %ymm1, %ymm3, %ymm0, %ymm5
; AVX2-ONLY-NEXT:    vpblendvb %ymm13, %ymm0, %ymm3, %ymm15
; AVX2-ONLY-NEXT:    vmovdqa 352(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vmovdqa 320(%rdi), %ymm6
; AVX2-ONLY-NEXT:    vpblendvb %ymm1, %ymm6, %ymm4, %ymm1
; AVX2-ONLY-NEXT:    vpblendvb %ymm13, %ymm4, %ymm6, %ymm12
; AVX2-ONLY-NEXT:    vpblendvb %ymm2, %ymm4, %ymm6, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu (%rsp), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendvb %ymm2, %ymm10, %ymm0, %ymm8
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendvb %ymm2, %ymm7, %ymm0, %ymm10
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,0,255,255,u,u,0,0,255,255,u,u,0,0,255,255,255,255,u,u,0,0,255,255,u,u,0,0,255,255,u,u>
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendvb %ymm2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm13 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendvb %ymm2, %ymm11, %ymm9, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm2 = <u,u,u,u,u,0,6,12,128,128,128,4,10,128,128,128>
; AVX2-ONLY-NEXT:    vpshufb %xmm2, %xmm5, %xmm6
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm5, %xmm11
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,u,128,128,128,2,8,14,128,128,0,6,12>
; AVX2-ONLY-NEXT:    vpshufb %xmm0, %xmm11, %xmm9
; AVX2-ONLY-NEXT:    vpor %xmm6, %xmm9, %xmm6
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm6, %ymm0, %ymm6
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} ymm9 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX2-ONLY-NEXT:    vpblendvb %ymm9, {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpshufb %xmm2, %xmm1, %xmm2
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm1, %xmm6
; AVX2-ONLY-NEXT:    vpshufb %xmm0, %xmm6, %xmm0
; AVX2-ONLY-NEXT:    vpor %xmm2, %xmm0, %xmm0
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendvb %ymm9, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,u,1,7,13,128,128,128,5,11,128,128,128>
; AVX2-ONLY-NEXT:    vpshufb %xmm0, %xmm5, %xmm2
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm5 = <u,u,u,u,u,128,128,128,3,9,15,128,128,1,7,13>
; AVX2-ONLY-NEXT:    vpshufb %xmm5, %xmm11, %xmm11
; AVX2-ONLY-NEXT:    vpor %xmm2, %xmm11, %xmm2
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vpblendvb %ymm9, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm2 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpshufb %xmm0, %xmm1, %xmm0
; AVX2-ONLY-NEXT:    vpshufb %xmm5, %xmm6, %xmm1
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendvb %ymm9, %ymm14, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, (%rsp) # 32-byte Spill
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm15, %xmm14
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm7 = <u,u,u,u,u,128,128,128,4,10,128,128,128,2,8,14>
; AVX2-ONLY-NEXT:    vpshufb %xmm7, %xmm14, %xmm0
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm2 = <u,u,u,u,u,2,8,14,128,128,0,6,12,128,128,128>
; AVX2-ONLY-NEXT:    vpshufb %xmm2, %xmm15, %xmm1
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm1
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm8, %xmm3
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm5 = <128,128,0,6,12,128,128,128,4,10,u,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm5, %xmm3, %xmm6
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm4 = <4,10,128,128,128,2,8,14,128,128,u,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm4, %xmm8, %xmm11
; AVX2-ONLY-NEXT:    vpor %xmm6, %xmm11, %xmm6
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm11 = [4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14]
; AVX2-ONLY-NEXT:    vpshufb %ymm11, %ymm13, %ymm0
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} xmm6 = xmm6[0,1,2,3,4],xmm0[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm6[0,1,2,3],ymm0[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vpblendvb %ymm9, %ymm0, %ymm1, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm12, %xmm1
; AVX2-ONLY-NEXT:    vpshufb %xmm7, %xmm1, %xmm0
; AVX2-ONLY-NEXT:    vpshufb %xmm2, %xmm12, %xmm2
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm10, %xmm2
; AVX2-ONLY-NEXT:    vpshufb %xmm5, %xmm2, %xmm5
; AVX2-ONLY-NEXT:    vpshufb %xmm4, %xmm10, %xmm4
; AVX2-ONLY-NEXT:    vpor %xmm5, %xmm4, %xmm4
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpshufb %ymm11, %ymm6, %ymm5
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} xmm4 = xmm4[0,1,2,3,4],xmm5[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm4[0,1,2,3],ymm5[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendvb %ymm9, %ymm4, %ymm0, %ymm5
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,u,128,128,128,5,11,128,128,128,3,9,15>
; AVX2-ONLY-NEXT:    vpshufb %xmm0, %xmm14, %xmm4
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm7 = <u,u,u,u,u,3,9,15,128,128,1,7,13,128,128,128>
; AVX2-ONLY-NEXT:    vpshufb %xmm7, %xmm15, %xmm11
; AVX2-ONLY-NEXT:    vpor %xmm4, %xmm11, %xmm4
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm11 = <128,128,1,7,13,128,128,128,5,11,u,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm11, %xmm3, %xmm3
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm14 = <5,11,128,128,128,3,9,15,128,128,u,u,u,u,u,u>
; AVX2-ONLY-NEXT:    vpshufb %xmm14, %xmm8, %xmm8
; AVX2-ONLY-NEXT:    vpor %xmm3, %xmm8, %xmm3
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm8 = [5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15]
; AVX2-ONLY-NEXT:    vpshufb %ymm8, %ymm13, %ymm13
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} xmm3 = xmm3[0,1,2,3,4],xmm13[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm13[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm4, %ymm0, %ymm4
; AVX2-ONLY-NEXT:    vpblendvb %ymm9, %ymm3, %ymm4, %ymm4
; AVX2-ONLY-NEXT:    vpshufb %xmm0, %xmm1, %xmm0
; AVX2-ONLY-NEXT:    vpshufb %xmm7, %xmm12, %xmm1
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX2-ONLY-NEXT:    vpshufb %xmm11, %xmm2, %xmm1
; AVX2-ONLY-NEXT:    vpshufb %xmm14, %xmm10, %xmm2
; AVX2-ONLY-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX2-ONLY-NEXT:    vpshufb %ymm8, %ymm6, %ymm2
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3,4],xmm2[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendvb %ymm9, %ymm1, %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm9 # 32-byte Reload
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm9, %xmm0
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm1 = <u,u,u,u,u,u,128,128,0,6,12,128,128,128,4,10>
; AVX2-ONLY-NEXT:    vpshufb %xmm1, %xmm0, %xmm2
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm7 = <u,u,u,u,u,u,4,10,128,128,128,2,8,14,128,128>
; AVX2-ONLY-NEXT:    vpshufb %xmm7, %xmm9, %xmm8
; AVX2-ONLY-NEXT:    vmovdqa %ymm9, %ymm10
; AVX2-ONLY-NEXT:    vpor %xmm2, %xmm8, %xmm2
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} ymm2 = ymm8[0,1,2],ymm2[3,4,5,6,7],ymm8[8,9,10],ymm2[11,12,13,14,15]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm8[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm9 # 32-byte Reload
; AVX2-ONLY-NEXT:    vextracti128 $1, %ymm9, %xmm8
; AVX2-ONLY-NEXT:    vpshufb %xmm1, %xmm8, %xmm1
; AVX2-ONLY-NEXT:    vpshufb %xmm7, %xmm9, %xmm7
; AVX2-ONLY-NEXT:    vmovdqa %ymm9, %ymm11
; AVX2-ONLY-NEXT:    vpor %xmm1, %xmm7, %xmm1
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} ymm1 = ymm7[0,1,2],ymm1[3,4,5,6,7],ymm7[8,9,10],ymm1[11,12,13,14,15]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm7[0,1,2,3],ymm1[4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm7 = <u,u,u,u,u,u,128,128,1,7,13,128,128,128,5,11>
; AVX2-ONLY-NEXT:    vpshufb %xmm7, %xmm0, %xmm0
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm9 = <u,u,u,u,u,u,5,11,128,128,128,3,9,15,128,128>
; AVX2-ONLY-NEXT:    vpshufb %xmm9, %xmm10, %xmm10
; AVX2-ONLY-NEXT:    vpor %xmm0, %xmm10, %xmm0
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} ymm0 = ymm10[0,1,2],ymm0[3,4,5,6,7],ymm10[8,9,10],ymm0[11,12,13,14,15]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm10[0,1,2,3],ymm0[4,5,6,7]
; AVX2-ONLY-NEXT:    vpshufb %xmm7, %xmm8, %xmm7
; AVX2-ONLY-NEXT:    vpshufb %xmm9, %xmm11, %xmm8
; AVX2-ONLY-NEXT:    vpor %xmm7, %xmm8, %xmm7
; AVX2-ONLY-NEXT:    vinserti128 $1, %xmm7, %ymm0, %ymm7
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendw {{.*#+}} ymm7 = ymm8[0,1,2],ymm7[3,4,5,6,7],ymm8[8,9,10],ymm7[11,12,13,14,15]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm8[0,1,2,3],ymm7[4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm1, 32(%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, (%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %ymm7, 32(%rdx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, (%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm0, 32(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm0, (%rcx)
; AVX2-ONLY-NEXT:    vmovups (%rsp), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm0, 32(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm0, (%r8)
; AVX2-ONLY-NEXT:    vmovdqa %ymm5, 32(%r9)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm0, (%r9)
; AVX2-ONLY-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX2-ONLY-NEXT:    vmovdqa %ymm3, 32(%rax)
; AVX2-ONLY-NEXT:    vmovdqa %ymm4, (%rax)
; AVX2-ONLY-NEXT:    addq $328, %rsp # imm = 0x148
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-LABEL: load_i8_stride6_vf64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    subq $40, %rsp
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = <0,6,12,128,128,128,4,10,128,128,128,u,u,u,u,u>
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm11 = [65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535]
; AVX512F-NEXT:    vmovdqa64 224(%rdi), %ymm25
; AVX512F-NEXT:    vmovdqa64 192(%rdi), %ymm26
; AVX512F-NEXT:    vmovdqa %ymm11, %ymm0
; AVX512F-NEXT:    vpternlogq $202, %ymm25, %ymm26, %ymm0
; AVX512F-NEXT:    vpshufb %xmm1, %xmm0, %xmm3
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm5 = <128,128,128,2,8,14,128,128,0,6,12,u,u,u,u,u>
; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm4
; AVX512F-NEXT:    vpshufb %xmm5, %xmm4, %xmm6
; AVX512F-NEXT:    vpor %xmm3, %xmm6, %xmm9
; AVX512F-NEXT:    vmovdqa64 (%rdi), %ymm30
; AVX512F-NEXT:    vmovdqa64 32(%rdi), %ymm31
; AVX512F-NEXT:    vmovdqa64 128(%rdi), %ymm29
; AVX512F-NEXT:    vmovdqa64 160(%rdi), %ymm18
; AVX512F-NEXT:    vmovdqa %ymm11, %ymm6
; AVX512F-NEXT:    vpternlogq $202, %ymm29, %ymm18, %ymm6
; AVX512F-NEXT:    vextracti128 $1, %ymm6, %xmm7
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm3 = <u,u,u,u,u,u,128,128,0,6,12,128,128,128,4,10>
; AVX512F-NEXT:    vpshufb %xmm3, %xmm7, %xmm12
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm8 = <u,u,u,u,u,u,4,10,128,128,128,2,8,14,128,128>
; AVX512F-NEXT:    vpshufb %xmm8, %xmm6, %xmm13
; AVX512F-NEXT:    vpor %xmm12, %xmm13, %xmm12
; AVX512F-NEXT:    vinserti128 $1, %xmm12, %ymm0, %ymm12
; AVX512F-NEXT:    vinserti32x4 $2, %xmm9, %zmm12, %zmm2
; AVX512F-NEXT:    vmovdqu64 %zmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa %ymm11, %ymm9
; AVX512F-NEXT:    vpternlogq $202, %ymm31, %ymm30, %ymm9
; AVX512F-NEXT:    vpshufb %xmm1, %xmm9, %xmm1
; AVX512F-NEXT:    vextracti128 $1, %ymm9, %xmm13
; AVX512F-NEXT:    vpshufb %xmm5, %xmm13, %xmm5
; AVX512F-NEXT:    vporq %xmm1, %xmm5, %xmm17
; AVX512F-NEXT:    vmovdqa64 320(%rdi), %ymm24
; AVX512F-NEXT:    vmovdqa64 352(%rdi), %ymm22
; AVX512F-NEXT:    vmovdqa %ymm11, %ymm1
; AVX512F-NEXT:    vpternlogq $202, %ymm24, %ymm22, %ymm1
; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm5
; AVX512F-NEXT:    vpshufb %xmm3, %xmm5, %xmm3
; AVX512F-NEXT:    vpshufb %xmm8, %xmm1, %xmm8
; AVX512F-NEXT:    vpor %xmm3, %xmm8, %xmm3
; AVX512F-NEXT:    vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm8 = <1,7,13,128,128,128,5,11,128,128,128,u,u,u,u,u>
; AVX512F-NEXT:    vpshufb %xmm8, %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm12 = <128,128,128,3,9,15,128,128,1,7,13,u,u,u,u,u>
; AVX512F-NEXT:    vpshufb %xmm12, %xmm4, %xmm4
; AVX512F-NEXT:    vpor %xmm0, %xmm4, %xmm0
; AVX512F-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,u,u,128,128,1,7,13,128,128,128,5,11>
; AVX512F-NEXT:    vpshufb %xmm0, %xmm7, %xmm4
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm7 = <u,u,u,u,u,u,5,11,128,128,128,3,9,15,128,128>
; AVX512F-NEXT:    vpshufb %xmm7, %xmm6, %xmm6
; AVX512F-NEXT:    vporq %xmm4, %xmm6, %xmm28
; AVX512F-NEXT:    vpshufb %xmm8, %xmm9, %xmm4
; AVX512F-NEXT:    vpshufb %xmm12, %xmm13, %xmm6
; AVX512F-NEXT:    vporq %xmm4, %xmm6, %xmm21
; AVX512F-NEXT:    vpshufb %xmm0, %xmm5, %xmm0
; AVX512F-NEXT:    vpshufb %xmm7, %xmm1, %xmm1
; AVX512F-NEXT:    vporq %xmm0, %xmm1, %xmm27
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm0 = <128,128,128,4,10,128,128,128,2,8,14,u,u,u,u,u>
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm9 = [65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535,0,65535,65535]
; AVX512F-NEXT:    vmovdqa %ymm9, %ymm4
; AVX512F-NEXT:    vpternlogq $202, %ymm26, %ymm25, %ymm4
; AVX512F-NEXT:    vextracti128 $1, %ymm4, %xmm15
; AVX512F-NEXT:    vpshufb %xmm0, %xmm15, %xmm1
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm6 = <2,8,14,128,128,0,6,12,128,128,128,u,u,u,u,u>
; AVX512F-NEXT:    vpshufb %xmm6, %xmm4, %xmm5
; AVX512F-NEXT:    vpor %xmm1, %xmm5, %xmm1
; AVX512F-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX512F-NEXT:    vmovdqa %ymm11, %ymm5
; AVX512F-NEXT:    vpternlogq $202, %ymm18, %ymm29, %ymm5
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm7 = <u,u,u,u,u,0,6,12,128,128,128,4,10,128,128,128>
; AVX512F-NEXT:    vpshufb %xmm7, %xmm5, %xmm8
; AVX512F-NEXT:    vextracti128 $1, %ymm5, %xmm1
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm12 = <u,u,u,u,u,128,128,128,2,8,14,128,128,0,6,12>
; AVX512F-NEXT:    vpshufb %xmm12, %xmm1, %xmm13
; AVX512F-NEXT:    vpor %xmm8, %xmm13, %xmm2
; AVX512F-NEXT:    vmovdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX512F-NEXT:    vmovdqa %ymm9, %ymm13
; AVX512F-NEXT:    vpternlogq $202, %ymm30, %ymm31, %ymm13
; AVX512F-NEXT:    vextracti128 $1, %ymm13, %xmm14
; AVX512F-NEXT:    vpshufb %xmm0, %xmm14, %xmm0
; AVX512F-NEXT:    vpshufb %xmm6, %xmm13, %xmm6
; AVX512F-NEXT:    vporq %xmm0, %xmm6, %xmm16
; AVX512F-NEXT:    vmovdqa %ymm11, %ymm10
; AVX512F-NEXT:    vpternlogq $202, %ymm22, %ymm24, %ymm10
; AVX512F-NEXT:    vpshufb %xmm7, %xmm10, %xmm8
; AVX512F-NEXT:    vextracti128 $1, %ymm10, %xmm7
; AVX512F-NEXT:    vpshufb %xmm12, %xmm7, %xmm12
; AVX512F-NEXT:    vpor %xmm8, %xmm12, %xmm0
; AVX512F-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm12 = <128,128,128,5,11,128,128,128,3,9,15,u,u,u,u,u>
; AVX512F-NEXT:    vpshufb %xmm12, %xmm15, %xmm15
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm8 = <3,9,15,128,128,1,7,13,128,128,128,u,u,u,u,u>
; AVX512F-NEXT:    vpshufb %xmm8, %xmm4, %xmm4
; AVX512F-NEXT:    vpor %xmm4, %xmm15, %xmm0
; AVX512F-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm15 = xmm5[u,u,u,u,u,1,7,13],zero,zero,zero,xmm5[5,11],zero,zero,zero
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[u,u,u,u,u],zero,zero,zero,xmm1[3,9,15],zero,zero,xmm1[1,7,13]
; AVX512F-NEXT:    vpor %xmm1, %xmm15, %xmm0
; AVX512F-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX512F-NEXT:    vmovdqa 256(%rdi), %ymm1
; AVX512F-NEXT:    vshufi64x2 {{.*#+}} ymm19 = ymm1[2,3],mem[2,3]
; AVX512F-NEXT:    vinserti32x4 $1, 288(%rdi), %ymm1, %ymm20
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = [65535,65535,0,65535,65535,0,65535,65535,65535,0,65535,65535,0,65535,65535,0]
; AVX512F-NEXT:    vmovdqa %ymm5, %ymm1
; AVX512F-NEXT:    vpternlogq $202, %ymm19, %ymm20, %ymm1
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm15 = ymm1[u,u,u,u,u,u,u,u,u,u,u,2,8,14,4,10,16,22,28,18,24,30,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpblendw {{.*#+}} ymm3 = ymm15[0,1,2],ymm3[3,4,5,6,7],ymm15[8,9,10],ymm3[11,12,13,14,15]
; AVX512F-NEXT:    vpblendd {{.*#+}} ymm15 = ymm15[0,1,2,3],ymm3[4,5,6,7]
; AVX512F-NEXT:    vmovdqa64 64(%rdi), %ymm23
; AVX512F-NEXT:    vshufi64x2 {{.*#+}} ymm6 = ymm23[2,3],mem[2,3]
; AVX512F-NEXT:    vinserti32x4 $1, 96(%rdi), %ymm23, %ymm23
; AVX512F-NEXT:    vinserti64x4 $1, %ymm15, %zmm0, %zmm15
; AVX512F-NEXT:    vmovdqa %ymm5, %ymm2
; AVX512F-NEXT:    vpternlogq $202, %ymm6, %ymm23, %ymm2
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm4 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255]
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm2[2,8,14,4,10,16,22,28,18,24,30],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; AVX512F-NEXT:    vpternlogq $248, %ymm4, %ymm17, %ymm0
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm17 = [65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,0,0,0,0,0,0,0,0,0,0,0,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535]
; AVX512F-NEXT:    vpternlogq $226, {{[-0-9]+}}(%r{{[sb]}}p), %zmm17, %zmm0 # 64-byte Folded Reload
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
; AVX512F-NEXT:    vpternlogq $184, %zmm0, %zmm3, %zmm15
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm0 = ymm1[u,u,u,u,u,u,u,u,u,u,u,3,9,15,5,11,17,23,29,19,25,31,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vinserti32x4 $1, %xmm27, %ymm0, %ymm1
; AVX512F-NEXT:    vpblendw {{.*#+}} ymm1 = ymm0[0,1,2],ymm1[3,4,5,6,7],ymm0[8,9,10],ymm1[11,12,13,14,15]
; AVX512F-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm2[3,9,15,5,11,17,23,29,19,25,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; AVX512F-NEXT:    vpternlogq $248, %ymm4, %ymm21, %ymm1
; AVX512F-NEXT:    vinserti32x4 $1, %xmm28, %ymm0, %ymm2
; AVX512F-NEXT:    vinserti32x4 $2, {{[-0-9]+}}(%r{{[sb]}}p), %zmm2, %zmm2 # 16-byte Folded Reload
; AVX512F-NEXT:    vpternlogq $226, %zmm2, %zmm17, %zmm1
; AVX512F-NEXT:    vinserti64x4 $1, %ymm0, %zmm0, %zmm17
; AVX512F-NEXT:    vpternlogq $184, %zmm1, %zmm3, %zmm17
; AVX512F-NEXT:    vpshufb %xmm12, %xmm14, %xmm0
; AVX512F-NEXT:    vpshufb %xmm8, %xmm13, %xmm1
; AVX512F-NEXT:    vporq %xmm0, %xmm1, %xmm21
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm10[u,u,u,u,u,1,7,13],zero,zero,zero,xmm10[5,11],zero,zero,zero
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm1 = xmm7[u,u,u,u,u],zero,zero,zero,xmm7[3,9,15],zero,zero,xmm7[1,7,13]
; AVX512F-NEXT:    vporq %xmm0, %xmm1, %xmm28
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm2 = <128,128,0,6,12,128,128,128,4,10,u,u,u,u,u,u>
; AVX512F-NEXT:    vmovdqa64 %ymm25, %ymm7
; AVX512F-NEXT:    vpternlogq $226, %ymm26, %ymm11, %ymm7
; AVX512F-NEXT:    vextracti128 $1, %ymm7, %xmm0
; AVX512F-NEXT:    vpshufb %xmm2, %xmm0, %xmm1
; AVX512F-NEXT:    vmovdqa64 %xmm2, %xmm25
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm14 = <4,10,128,128,128,2,8,14,128,128,u,u,u,u,u,u>
; AVX512F-NEXT:    vpshufb %xmm14, %xmm7, %xmm2
; AVX512F-NEXT:    vporq %xmm1, %xmm2, %xmm26
; AVX512F-NEXT:    vmovdqa64 %ymm18, %ymm10
; AVX512F-NEXT:    vpternlogq $226, %ymm29, %ymm9, %ymm10
; AVX512F-NEXT:    vextracti128 $1, %ymm10, %xmm8
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = <u,u,u,u,u,128,128,128,4,10,128,128,128,2,8,14>
; AVX512F-NEXT:    vpshufb %xmm1, %xmm8, %xmm2
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm3 = <u,u,u,u,u,2,8,14,128,128,0,6,12,128,128,128>
; AVX512F-NEXT:    vpshufb %xmm3, %xmm10, %xmm4
; AVX512F-NEXT:    vporq %xmm2, %xmm4, %xmm27
; AVX512F-NEXT:    vpternlogq $202, %ymm30, %ymm31, %ymm11
; AVX512F-NEXT:    vmovdqa %ymm5, %ymm4
; AVX512F-NEXT:    vpternlogq $202, %ymm23, %ymm6, %ymm4
; AVX512F-NEXT:    vpternlogq $202, %ymm24, %ymm22, %ymm9
; AVX512F-NEXT:    vextracti128 $1, %ymm9, %xmm12
; AVX512F-NEXT:    vpshufb %xmm1, %xmm12, %xmm1
; AVX512F-NEXT:    vpshufb %xmm3, %xmm9, %xmm2
; AVX512F-NEXT:    vporq %xmm1, %xmm2, %xmm24
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = <128,128,1,7,13,128,128,128,5,11,u,u,u,u,u,u>
; AVX512F-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqa64 %xmm1, %xmm29
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm13 = <5,11,128,128,128,3,9,15,128,128,u,u,u,u,u,u>
; AVX512F-NEXT:    vpshufb %xmm13, %xmm7, %xmm1
; AVX512F-NEXT:    vporq %xmm0, %xmm1, %xmm18
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm2 = <u,u,u,u,u,128,128,128,5,11,128,128,128,3,9,15>
; AVX512F-NEXT:    vpshufb %xmm2, %xmm8, %xmm1
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,u,3,9,15,128,128,1,7,13,128,128,128>
; AVX512F-NEXT:    vpshufb %xmm0, %xmm10, %xmm8
; AVX512F-NEXT:    vpor %xmm1, %xmm8, %xmm8
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm1 = [128,128,128,128,128,128,128,128,128,128,128,4,10,0,6,12,18,24,30,20,26,128,128,128,128,128,128,128,128,128,128,128]
; AVX512F-NEXT:    vpshufb %ymm1, %ymm4, %ymm3
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm22 = [255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255]
; AVX512F-NEXT:    vpternlogq $236, %ymm22, %ymm3, %ymm16
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [128,128,128,128,128,128,128,128,128,128,128,5,11,1,7,13,19,25,31,21,27,128,128,128,128,128,128,128,128,128,128,128]
; AVX512F-NEXT:    vpshufb %ymm3, %ymm4, %ymm4
; AVX512F-NEXT:    vpternlogq $236, %ymm22, %ymm4, %ymm21
; AVX512F-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm4 # 16-byte Folded Reload
; AVX512F-NEXT:    vpternlogq $202, %ymm20, %ymm19, %ymm5
; AVX512F-NEXT:    vpshufb %ymm1, %ymm5, %ymm1
; AVX512F-NEXT:    vpternlogq $248, %ymm22, %ymm4, %ymm1
; AVX512F-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm4
; AVX512F-NEXT:    vpshufb %ymm3, %ymm5, %ymm3
; AVX512F-NEXT:    vextracti128 $1, %ymm11, %xmm1
; AVX512F-NEXT:    vmovdqa64 %xmm25, %xmm5
; AVX512F-NEXT:    vpshufb %xmm5, %xmm1, %xmm5
; AVX512F-NEXT:    vpshufb %xmm14, %xmm11, %xmm14
; AVX512F-NEXT:    vpor %xmm5, %xmm14, %xmm5
; AVX512F-NEXT:    vpshufb %xmm2, %xmm12, %xmm2
; AVX512F-NEXT:    vpshufb %xmm0, %xmm9, %xmm0
; AVX512F-NEXT:    vpor %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,65535,65535,0,65535,65535,0,65535,65535,65535,0,65535,65535,0,65535,65535]
; AVX512F-NEXT:    vpternlogq $226, %ymm23, %ymm2, %ymm6
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm9 = ymm6[u,u,u,u,u,u,u,u,u,u,0,6,12,2,8,14,20,26,16,22,28,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpblendw {{.*#+}} xmm5 = xmm5[0,1,2,3,4],xmm9[5,6,7]
; AVX512F-NEXT:    vpblendd {{.*#+}} ymm5 = ymm5[0,1,2,3],ymm9[4,5,6,7]
; AVX512F-NEXT:    vinserti32x4 $1, %xmm28, %ymm0, %ymm9
; AVX512F-NEXT:    vpternlogq $248, %ymm22, %ymm9, %ymm3
; AVX512F-NEXT:    vinserti64x4 $1, %ymm3, %zmm0, %zmm3
; AVX512F-NEXT:    vinserti32x4 $1, %xmm24, %ymm0, %ymm9
; AVX512F-NEXT:    vpternlogq $202, %ymm20, %ymm19, %ymm2
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm10 = ymm2[u,u,u,u,u,u,u,u,u,u,0,6,12,2,8,14,20,26,16,22,28],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm12 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
; AVX512F-NEXT:    vpternlogq $242, %ymm9, %ymm12, %ymm10
; AVX512F-NEXT:    vinserti64x4 $1, %ymm10, %zmm0, %zmm9
; AVX512F-NEXT:    vinserti32x4 $1, %xmm27, %ymm0, %ymm10
; AVX512F-NEXT:    vinserti32x4 $2, %xmm26, %zmm10, %zmm10
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm14 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX512F-NEXT:    vpternlogq $226, %zmm10, %zmm14, %zmm5
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm10 = [65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,65535,0,0,0,0,0,0,0,0,0,0,0]
; AVX512F-NEXT:    vpternlogq $184, %zmm5, %zmm10, %zmm9
; AVX512F-NEXT:    vmovdqa64 %xmm29, %xmm5
; AVX512F-NEXT:    vpshufb %xmm5, %xmm1, %xmm1
; AVX512F-NEXT:    vpshufb %xmm13, %xmm11, %xmm5
; AVX512F-NEXT:    vpor %xmm1, %xmm5, %xmm1
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm5 = ymm6[u,u,u,u,u,u,u,u,u,u,1,7,13,3,9,15,21,27,17,23,29,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3,4],xmm5[5,6,7]
; AVX512F-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm5[4,5,6,7]
; AVX512F-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX512F-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[u,u,u,u,u,u,u,u,u,u,1,7,13,3,9,15,21,27,17,23,29],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; AVX512F-NEXT:    vpternlogq $242, %ymm0, %ymm12, %ymm2
; AVX512F-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
; AVX512F-NEXT:    vinserti128 $1, %xmm8, %ymm0, %ymm2
; AVX512F-NEXT:    vinserti32x4 $2, %xmm18, %zmm2, %zmm2
; AVX512F-NEXT:    vpternlogq $226, %zmm2, %zmm14, %zmm1
; AVX512F-NEXT:    vpternlogq $184, %zmm1, %zmm10, %zmm0
; AVX512F-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm1 # 16-byte Folded Reload
; AVX512F-NEXT:    vinserti32x4 $2, {{[-0-9]+}}(%r{{[sb]}}p), %zmm1, %zmm1 # 16-byte Folded Reload
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
; AVX512F-NEXT:    vpternlogq $184, %zmm1, %zmm2, %zmm16
; AVX512F-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm1 # 16-byte Folded Reload
; AVX512F-NEXT:    vinserti32x4 $2, {{[-0-9]+}}(%r{{[sb]}}p), %zmm1, %zmm1 # 16-byte Folded Reload
; AVX512F-NEXT:    vpternlogq $184, %zmm1, %zmm2, %zmm21
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm1 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
; AVX512F-NEXT:    vpternlogq $184, %zmm16, %zmm1, %zmm4
; AVX512F-NEXT:    vpternlogq $184, %zmm21, %zmm1, %zmm3
; AVX512F-NEXT:    vmovdqa64 %zmm15, (%rsi)
; AVX512F-NEXT:    vmovdqa64 %zmm17, (%rdx)
; AVX512F-NEXT:    vmovdqa64 %zmm4, (%rcx)
; AVX512F-NEXT:    vmovdqa64 %zmm3, (%r8)
; AVX512F-NEXT:    vmovdqa64 %zmm9, (%r9)
; AVX512F-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX512F-NEXT:    vmovdqa64 %zmm0, (%rax)
; AVX512F-NEXT:    addq $40, %rsp
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride6_vf64:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,6,12,128,128,128,4,10,128,128,128,u,u,u,u,u>
; AVX512BW-NEXT:    vmovdqa 224(%rdi), %ymm0
; AVX512BW-NEXT:    vmovdqa64 192(%rdi), %ymm23
; AVX512BW-NEXT:    movw $18724, %r10w # imm = 0x4924
; AVX512BW-NEXT:    kmovd %r10d, %k1
; AVX512BW-NEXT:    vpblendmw %ymm0, %ymm23, %ymm9 {%k1}
; AVX512BW-NEXT:    vpshufb %xmm2, %xmm9, %xmm1
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm4 = <128,128,128,2,8,14,128,128,0,6,12,u,u,u,u,u>
; AVX512BW-NEXT:    vextracti128 $1, %ymm9, %xmm12
; AVX512BW-NEXT:    vpshufb %xmm4, %xmm12, %xmm3
; AVX512BW-NEXT:    vpor %xmm1, %xmm3, %xmm5
; AVX512BW-NEXT:    vmovdqa (%rdi), %ymm10
; AVX512BW-NEXT:    vmovdqa 32(%rdi), %ymm3
; AVX512BW-NEXT:    vmovdqa 64(%rdi), %ymm6
; AVX512BW-NEXT:    vmovdqa64 128(%rdi), %ymm26
; AVX512BW-NEXT:    vmovdqa 160(%rdi), %ymm1
; AVX512BW-NEXT:    vpblendmw %ymm26, %ymm1, %ymm15 {%k1}
; AVX512BW-NEXT:    vextracti32x4 $1, %ymm15, %xmm16
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm17 = <u,u,u,u,u,u,128,128,0,6,12,128,128,128,4,10>
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm16, %xmm11
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm18 = <u,u,u,u,u,u,4,10,128,128,128,2,8,14,128,128>
; AVX512BW-NEXT:    vpshufb %xmm18, %xmm15, %xmm13
; AVX512BW-NEXT:    vpor %xmm11, %xmm13, %xmm11
; AVX512BW-NEXT:    vinserti128 $1, %xmm11, %ymm0, %ymm11
; AVX512BW-NEXT:    vinserti32x4 $2, %xmm5, %zmm11, %zmm11
; AVX512BW-NEXT:    vperm2i128 {{.*#+}} ymm5 = ymm6[2,3],mem[2,3]
; AVX512BW-NEXT:    vinserti128 $1, 96(%rdi), %ymm6, %ymm13
; AVX512BW-NEXT:    movw $-28124, %r10w # imm = 0x9224
; AVX512BW-NEXT:    kmovd %r10d, %k4
; AVX512BW-NEXT:    vpblendmw %ymm5, %ymm13, %ymm19 {%k4}
; AVX512BW-NEXT:    vpblendmw %ymm3, %ymm10, %ymm20 {%k1}
; AVX512BW-NEXT:    vpshufb %xmm2, %xmm20, %xmm2
; AVX512BW-NEXT:    vextracti32x4 $1, %ymm20, %xmm21
; AVX512BW-NEXT:    vpshufb %xmm4, %xmm21, %xmm4
; AVX512BW-NEXT:    vpor %xmm2, %xmm4, %xmm2
; AVX512BW-NEXT:    vpbroadcastq {{.*#+}} ymm6 = [0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10]
; AVX512BW-NEXT:    movl $4192256, %r10d # imm = 0x3FF800
; AVX512BW-NEXT:    kmovd %r10d, %k2
; AVX512BW-NEXT:    vpshufb %ymm6, %ymm19, %ymm2 {%k2}
; AVX512BW-NEXT:    vmovdqu16 %zmm11, %zmm2 {%k2}
; AVX512BW-NEXT:    vmovdqa 256(%rdi), %ymm11
; AVX512BW-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm11[2,3],mem[2,3]
; AVX512BW-NEXT:    vinserti128 $1, 288(%rdi), %ymm11, %ymm14
; AVX512BW-NEXT:    vpblendmw %ymm4, %ymm14, %ymm22 {%k4}
; AVX512BW-NEXT:    vpshufb %ymm6, %ymm22, %ymm7
; AVX512BW-NEXT:    vmovdqa 320(%rdi), %ymm11
; AVX512BW-NEXT:    vmovdqa 352(%rdi), %ymm6
; AVX512BW-NEXT:    vpblendmw %ymm11, %ymm6, %ymm24 {%k1}
; AVX512BW-NEXT:    vextracti32x4 $1, %ymm24, %xmm25
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm25, %xmm17
; AVX512BW-NEXT:    vpshufb %xmm18, %xmm24, %xmm18
; AVX512BW-NEXT:    vporq %xmm17, %xmm18, %xmm17
; AVX512BW-NEXT:    vinserti32x4 $1, %xmm17, %ymm0, %ymm8
; AVX512BW-NEXT:    vpblendw {{.*#+}} ymm8 = ymm7[0,1,2],ymm8[3,4,5,6,7],ymm7[8,9,10],ymm8[11,12,13,14,15]
; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2,3],ymm8[4,5,6,7]
; AVX512BW-NEXT:    vinserti64x4 $1, %ymm7, %zmm0, %zmm7
; AVX512BW-NEXT:    movabsq $-8796093022208, %rdi # imm = 0xFFFFF80000000000
; AVX512BW-NEXT:    kmovq %rdi, %k3
; AVX512BW-NEXT:    vmovdqu8 %zmm7, %zmm2 {%k3}
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm7 = <1,7,13,128,128,128,5,11,128,128,128,u,u,u,u,u>
; AVX512BW-NEXT:    vpshufb %xmm7, %xmm9, %xmm8
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm9 = <128,128,128,3,9,15,128,128,1,7,13,u,u,u,u,u>
; AVX512BW-NEXT:    vpshufb %xmm9, %xmm12, %xmm12
; AVX512BW-NEXT:    vpor %xmm8, %xmm12, %xmm8
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm12 = <u,u,u,u,u,u,128,128,1,7,13,128,128,128,5,11>
; AVX512BW-NEXT:    vpshufb %xmm12, %xmm16, %xmm16
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm17 = <u,u,u,u,u,u,5,11,128,128,128,3,9,15,128,128>
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm15, %xmm15
; AVX512BW-NEXT:    vporq %xmm16, %xmm15, %xmm15
; AVX512BW-NEXT:    vinserti128 $1, %xmm15, %ymm0, %ymm15
; AVX512BW-NEXT:    vinserti32x4 $2, %xmm8, %zmm15, %zmm8
; AVX512BW-NEXT:    vpshufb %xmm7, %xmm20, %xmm7
; AVX512BW-NEXT:    vpshufb %xmm9, %xmm21, %xmm9
; AVX512BW-NEXT:    vpor %xmm7, %xmm9, %xmm9
; AVX512BW-NEXT:    vpbroadcastq {{.*#+}} ymm7 = [1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11]
; AVX512BW-NEXT:    vpshufb %ymm7, %ymm19, %ymm9 {%k2}
; AVX512BW-NEXT:    vmovdqu16 %zmm8, %zmm9 {%k2}
; AVX512BW-NEXT:    vpshufb %ymm7, %ymm22, %ymm7
; AVX512BW-NEXT:    vpshufb %xmm12, %xmm25, %xmm8
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm24, %xmm12
; AVX512BW-NEXT:    vpor %xmm8, %xmm12, %xmm8
; AVX512BW-NEXT:    vinserti128 $1, %xmm8, %ymm0, %ymm8
; AVX512BW-NEXT:    vpblendw {{.*#+}} ymm8 = ymm7[0,1,2],ymm8[3,4,5,6,7],ymm7[8,9,10],ymm8[11,12,13,14,15]
; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2,3],ymm8[4,5,6,7]
; AVX512BW-NEXT:    vinserti64x4 $1, %ymm7, %zmm0, %zmm7
; AVX512BW-NEXT:    vmovdqu8 %zmm7, %zmm9 {%k3}
; AVX512BW-NEXT:    vpblendmw %ymm13, %ymm5, %ymm15 {%k4}
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm7 = <128,128,128,4,10,128,128,128,2,8,14,u,u,u,u,u>
; AVX512BW-NEXT:    movw $9362, %di # imm = 0x2492
; AVX512BW-NEXT:    kmovd %edi, %k2
; AVX512BW-NEXT:    vpblendmw %ymm10, %ymm3, %ymm8 {%k2}
; AVX512BW-NEXT:    vextracti32x4 $1, %ymm8, %xmm16
; AVX512BW-NEXT:    vpshufb %xmm7, %xmm16, %xmm12
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm17 = <2,8,14,128,128,0,6,12,128,128,128,u,u,u,u,u>
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm8, %xmm18
; AVX512BW-NEXT:    vporq %xmm12, %xmm18, %xmm18
; AVX512BW-NEXT:    vpbroadcastq {{.*#+}} ymm19 = [2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12]
; AVX512BW-NEXT:    movl $2095104, %edi # imm = 0x1FF800
; AVX512BW-NEXT:    kmovd %edi, %k5
; AVX512BW-NEXT:    vpshufb %ymm19, %ymm15, %ymm18 {%k5}
; AVX512BW-NEXT:    vpblendmw %ymm23, %ymm0, %ymm20 {%k2}
; AVX512BW-NEXT:    vextracti32x4 $1, %ymm20, %xmm21
; AVX512BW-NEXT:    vpshufb %xmm7, %xmm21, %xmm7
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm20, %xmm12
; AVX512BW-NEXT:    vpor %xmm7, %xmm12, %xmm7
; AVX512BW-NEXT:    vpblendmw %ymm1, %ymm26, %ymm17 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm22 = <u,u,u,u,u,0,6,12,128,128,128,4,10,128,128,128>
; AVX512BW-NEXT:    vpshufb %xmm22, %xmm17, %xmm12
; AVX512BW-NEXT:    vextracti32x4 $1, %ymm17, %xmm24
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm25 = <u,u,u,u,u,128,128,128,2,8,14,128,128,0,6,12>
; AVX512BW-NEXT:    vpshufb %xmm25, %xmm24, %xmm27
; AVX512BW-NEXT:    vporq %xmm12, %xmm27, %xmm12
; AVX512BW-NEXT:    vinserti128 $1, %xmm12, %ymm0, %ymm12
; AVX512BW-NEXT:    vinserti32x4 $2, %xmm7, %zmm12, %zmm12
; AVX512BW-NEXT:    movl $2097151, %edi # imm = 0x1FFFFF
; AVX512BW-NEXT:    kmovq %rdi, %k6
; AVX512BW-NEXT:    vmovdqu8 %zmm18, %zmm12 {%k6}
; AVX512BW-NEXT:    vpblendmw %ymm14, %ymm4, %ymm7 {%k4}
; AVX512BW-NEXT:    vpblendmw %ymm6, %ymm11, %ymm18 {%k1}
; AVX512BW-NEXT:    vpshufb %xmm22, %xmm18, %xmm22
; AVX512BW-NEXT:    vextracti32x4 $1, %ymm18, %xmm27
; AVX512BW-NEXT:    vpshufb %xmm25, %xmm27, %xmm25
; AVX512BW-NEXT:    vporq %xmm22, %xmm25, %xmm22
; AVX512BW-NEXT:    vinserti32x4 $1, %xmm22, %ymm0, %ymm22
; AVX512BW-NEXT:    vpshufb %ymm19, %ymm7, %ymm22 {%k5}
; AVX512BW-NEXT:    vinserti64x4 $1, %ymm22, %zmm0, %zmm19
; AVX512BW-NEXT:    vmovdqu8 %zmm19, %zmm12 {%k3}
; AVX512BW-NEXT:    movw $9289, %di # imm = 0x2449
; AVX512BW-NEXT:    kmovd %edi, %k4
; AVX512BW-NEXT:    vmovdqu16 %ymm14, %ymm4 {%k4}
; AVX512BW-NEXT:    vmovdqu16 %ymm13, %ymm5 {%k4}
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm13 = <128,128,128,5,11,128,128,128,3,9,15,u,u,u,u,u>
; AVX512BW-NEXT:    vpshufb %xmm13, %xmm16, %xmm14
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm16 = <3,9,15,128,128,1,7,13,128,128,128,u,u,u,u,u>
; AVX512BW-NEXT:    vpshufb %xmm16, %xmm8, %xmm8
; AVX512BW-NEXT:    vpor %xmm14, %xmm8, %xmm8
; AVX512BW-NEXT:    vpbroadcastq {{.*#+}} ymm14 = [3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13]
; AVX512BW-NEXT:    vpshufb %ymm14, %ymm15, %ymm8 {%k5}
; AVX512BW-NEXT:    vpshufb %xmm13, %xmm21, %xmm13
; AVX512BW-NEXT:    vpshufb %xmm16, %xmm20, %xmm15
; AVX512BW-NEXT:    vpor %xmm13, %xmm15, %xmm13
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm15 = <u,u,u,u,u,1,7,13,128,128,128,5,11,128,128,128>
; AVX512BW-NEXT:    vpshufb %xmm15, %xmm17, %xmm16
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm17 = <u,u,u,u,u,128,128,128,3,9,15,128,128,1,7,13>
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm24, %xmm19
; AVX512BW-NEXT:    vporq %xmm16, %xmm19, %xmm16
; AVX512BW-NEXT:    vinserti32x4 $1, %xmm16, %ymm0, %ymm16
; AVX512BW-NEXT:    vinserti32x4 $2, %xmm13, %zmm16, %zmm13
; AVX512BW-NEXT:    vmovdqu8 %zmm8, %zmm13 {%k6}
; AVX512BW-NEXT:    vpshufb %xmm15, %xmm18, %xmm8
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm27, %xmm15
; AVX512BW-NEXT:    vpor %xmm8, %xmm15, %xmm8
; AVX512BW-NEXT:    vinserti128 $1, %xmm8, %ymm0, %ymm8
; AVX512BW-NEXT:    vpshufb %ymm14, %ymm7, %ymm8 {%k5}
; AVX512BW-NEXT:    vinserti64x4 $1, %ymm8, %zmm0, %zmm7
; AVX512BW-NEXT:    vmovdqu8 %zmm7, %zmm13 {%k3}
; AVX512BW-NEXT:    vpbroadcastq {{.*#+}} ymm7 = [4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14,4,10,0,6,12,2,8,14]
; AVX512BW-NEXT:    vpshufb %ymm7, %ymm5, %ymm8
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm14 = <128,128,0,6,12,128,128,128,4,10,u,u,u,u,u,u>
; AVX512BW-NEXT:    vmovdqu16 %ymm10, %ymm3 {%k1}
; AVX512BW-NEXT:    vextracti128 $1, %ymm3, %xmm15
; AVX512BW-NEXT:    vpshufb %xmm14, %xmm15, %xmm10
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm16 = <4,10,128,128,128,2,8,14,128,128,u,u,u,u,u,u>
; AVX512BW-NEXT:    vpshufb %xmm16, %xmm3, %xmm17
; AVX512BW-NEXT:    vporq %xmm10, %xmm17, %xmm10
; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm10 = xmm10[0,1,2,3,4],xmm8[5,6,7]
; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm10 = ymm10[0,1,2,3],ymm8[4,5,6,7]
; AVX512BW-NEXT:    vmovdqu16 %ymm23, %ymm0 {%k1}
; AVX512BW-NEXT:    vextracti128 $1, %ymm0, %xmm8
; AVX512BW-NEXT:    vpshufb %xmm14, %xmm8, %xmm14
; AVX512BW-NEXT:    vpshufb %xmm16, %xmm0, %xmm16
; AVX512BW-NEXT:    vporq %xmm14, %xmm16, %xmm14
; AVX512BW-NEXT:    vmovdqu16 %ymm26, %ymm1 {%k2}
; AVX512BW-NEXT:    vextracti32x4 $1, %ymm1, %xmm16
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm17 = <u,u,u,u,u,128,128,128,4,10,128,128,128,2,8,14>
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm16, %xmm18
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm19 = <u,u,u,u,u,2,8,14,128,128,0,6,12,128,128,128>
; AVX512BW-NEXT:    vpshufb %xmm19, %xmm1, %xmm20
; AVX512BW-NEXT:    vporq %xmm18, %xmm20, %xmm18
; AVX512BW-NEXT:    vinserti32x4 $1, %xmm18, %ymm0, %ymm18
; AVX512BW-NEXT:    vinserti32x4 $2, %xmm14, %zmm18, %zmm14
; AVX512BW-NEXT:    movabsq $4398044413952, %rdi # imm = 0x3FFFFE00000
; AVX512BW-NEXT:    kmovq %rdi, %k1
; AVX512BW-NEXT:    vmovdqu8 %zmm14, %zmm10 {%k1}
; AVX512BW-NEXT:    vpshufb %ymm7, %ymm4, %ymm7
; AVX512BW-NEXT:    vmovdqu16 %ymm11, %ymm6 {%k2}
; AVX512BW-NEXT:    vextracti128 $1, %ymm6, %xmm11
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm11, %xmm14
; AVX512BW-NEXT:    vpshufb %xmm19, %xmm6, %xmm17
; AVX512BW-NEXT:    vporq %xmm14, %xmm17, %xmm14
; AVX512BW-NEXT:    vinserti128 $1, %xmm14, %ymm0, %ymm14
; AVX512BW-NEXT:    movl $-2097152, %edi # imm = 0xFFE00000
; AVX512BW-NEXT:    kmovd %edi, %k2
; AVX512BW-NEXT:    vmovdqu8 %ymm14, %ymm7 {%k2}
; AVX512BW-NEXT:    vinserti64x4 $1, %ymm7, %zmm0, %zmm7
; AVX512BW-NEXT:    vmovdqu16 %zmm7, %zmm10 {%k2}
; AVX512BW-NEXT:    vpbroadcastq {{.*#+}} ymm7 = [5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15,5,11,1,7,13,3,9,15]
; AVX512BW-NEXT:    vpshufb %ymm7, %ymm5, %ymm5
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm14 = <128,128,1,7,13,128,128,128,5,11,u,u,u,u,u,u>
; AVX512BW-NEXT:    vpshufb %xmm14, %xmm15, %xmm15
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} xmm17 = <5,11,128,128,128,3,9,15,128,128,u,u,u,u,u,u>
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm3, %xmm3
; AVX512BW-NEXT:    vpor %xmm3, %xmm15, %xmm3
; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm3 = xmm3[0,1,2,3,4],xmm5[5,6,7]
; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm5[4,5,6,7]
; AVX512BW-NEXT:    vpshufb %xmm14, %xmm8, %xmm5
; AVX512BW-NEXT:    vpshufb %xmm17, %xmm0, %xmm0
; AVX512BW-NEXT:    vpor %xmm5, %xmm0, %xmm0
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm5 = <u,u,u,u,u,128,128,128,5,11,128,128,128,3,9,15>
; AVX512BW-NEXT:    vpshufb %xmm5, %xmm16, %xmm8
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm14 = <u,u,u,u,u,3,9,15,128,128,1,7,13,128,128,128>
; AVX512BW-NEXT:    vpshufb %xmm14, %xmm1, %xmm1
; AVX512BW-NEXT:    vpor %xmm1, %xmm8, %xmm1
; AVX512BW-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX512BW-NEXT:    vinserti32x4 $2, %xmm0, %zmm1, %zmm0
; AVX512BW-NEXT:    vmovdqu8 %zmm0, %zmm3 {%k1}
; AVX512BW-NEXT:    vpshufb %ymm7, %ymm4, %ymm0
; AVX512BW-NEXT:    vpshufb %xmm5, %xmm11, %xmm1
; AVX512BW-NEXT:    vpshufb %xmm14, %xmm6, %xmm4
; AVX512BW-NEXT:    vpor %xmm1, %xmm4, %xmm1
; AVX512BW-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX512BW-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k2}
; AVX512BW-NEXT:    vinserti64x4 $1, %ymm0, %zmm0, %zmm0
; AVX512BW-NEXT:    vmovdqu16 %zmm0, %zmm3 {%k2}
; AVX512BW-NEXT:    vmovdqa64 %zmm2, (%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm9, (%rdx)
; AVX512BW-NEXT:    vmovdqa64 %zmm12, (%rcx)
; AVX512BW-NEXT:    vmovdqa64 %zmm13, (%r8)
; AVX512BW-NEXT:    vmovdqa64 %zmm10, (%r9)
; AVX512BW-NEXT:    vmovdqa64 %zmm3, (%rax)
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <384 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <384 x i8> %wide.vec, <384 x i8> poison, <64 x i32> <i32 0, i32 6, i32 12, i32 18, i32 24, i32 30, i32 36, i32 42, i32 48, i32 54, i32 60, i32 66, i32 72, i32 78, i32 84, i32 90, i32 96, i32 102, i32 108, i32 114, i32 120, i32 126, i32 132, i32 138, i32 144, i32 150, i32 156, i32 162, i32 168, i32 174, i32 180, i32 186, i32 192, i32 198, i32 204, i32 210, i32 216, i32 222, i32 228, i32 234, i32 240, i32 246, i32 252, i32 258, i32 264, i32 270, i32 276, i32 282, i32 288, i32 294, i32 300, i32 306, i32 312, i32 318, i32 324, i32 330, i32 336, i32 342, i32 348, i32 354, i32 360, i32 366, i32 372, i32 378>
  %strided.vec1 = shufflevector <384 x i8> %wide.vec, <384 x i8> poison, <64 x i32> <i32 1, i32 7, i32 13, i32 19, i32 25, i32 31, i32 37, i32 43, i32 49, i32 55, i32 61, i32 67, i32 73, i32 79, i32 85, i32 91, i32 97, i32 103, i32 109, i32 115, i32 121, i32 127, i32 133, i32 139, i32 145, i32 151, i32 157, i32 163, i32 169, i32 175, i32 181, i32 187, i32 193, i32 199, i32 205, i32 211, i32 217, i32 223, i32 229, i32 235, i32 241, i32 247, i32 253, i32 259, i32 265, i32 271, i32 277, i32 283, i32 289, i32 295, i32 301, i32 307, i32 313, i32 319, i32 325, i32 331, i32 337, i32 343, i32 349, i32 355, i32 361, i32 367, i32 373, i32 379>
  %strided.vec2 = shufflevector <384 x i8> %wide.vec, <384 x i8> poison, <64 x i32> <i32 2, i32 8, i32 14, i32 20, i32 26, i32 32, i32 38, i32 44, i32 50, i32 56, i32 62, i32 68, i32 74, i32 80, i32 86, i32 92, i32 98, i32 104, i32 110, i32 116, i32 122, i32 128, i32 134, i32 140, i32 146, i32 152, i32 158, i32 164, i32 170, i32 176, i32 182, i32 188, i32 194, i32 200, i32 206, i32 212, i32 218, i32 224, i32 230, i32 236, i32 242, i32 248, i32 254, i32 260, i32 266, i32 272, i32 278, i32 284, i32 290, i32 296, i32 302, i32 308, i32 314, i32 320, i32 326, i32 332, i32 338, i32 344, i32 350, i32 356, i32 362, i32 368, i32 374, i32 380>
  %strided.vec3 = shufflevector <384 x i8> %wide.vec, <384 x i8> poison, <64 x i32> <i32 3, i32 9, i32 15, i32 21, i32 27, i32 33, i32 39, i32 45, i32 51, i32 57, i32 63, i32 69, i32 75, i32 81, i32 87, i32 93, i32 99, i32 105, i32 111, i32 117, i32 123, i32 129, i32 135, i32 141, i32 147, i32 153, i32 159, i32 165, i32 171, i32 177, i32 183, i32 189, i32 195, i32 201, i32 207, i32 213, i32 219, i32 225, i32 231, i32 237, i32 243, i32 249, i32 255, i32 261, i32 267, i32 273, i32 279, i32 285, i32 291, i32 297, i32 303, i32 309, i32 315, i32 321, i32 327, i32 333, i32 339, i32 345, i32 351, i32 357, i32 363, i32 369, i32 375, i32 381>
  %strided.vec4 = shufflevector <384 x i8> %wide.vec, <384 x i8> poison, <64 x i32> <i32 4, i32 10, i32 16, i32 22, i32 28, i32 34, i32 40, i32 46, i32 52, i32 58, i32 64, i32 70, i32 76, i32 82, i32 88, i32 94, i32 100, i32 106, i32 112, i32 118, i32 124, i32 130, i32 136, i32 142, i32 148, i32 154, i32 160, i32 166, i32 172, i32 178, i32 184, i32 190, i32 196, i32 202, i32 208, i32 214, i32 220, i32 226, i32 232, i32 238, i32 244, i32 250, i32 256, i32 262, i32 268, i32 274, i32 280, i32 286, i32 292, i32 298, i32 304, i32 310, i32 316, i32 322, i32 328, i32 334, i32 340, i32 346, i32 352, i32 358, i32 364, i32 370, i32 376, i32 382>
  %strided.vec5 = shufflevector <384 x i8> %wide.vec, <384 x i8> poison, <64 x i32> <i32 5, i32 11, i32 17, i32 23, i32 29, i32 35, i32 41, i32 47, i32 53, i32 59, i32 65, i32 71, i32 77, i32 83, i32 89, i32 95, i32 101, i32 107, i32 113, i32 119, i32 125, i32 131, i32 137, i32 143, i32 149, i32 155, i32 161, i32 167, i32 173, i32 179, i32 185, i32 191, i32 197, i32 203, i32 209, i32 215, i32 221, i32 227, i32 233, i32 239, i32 245, i32 251, i32 257, i32 263, i32 269, i32 275, i32 281, i32 287, i32 293, i32 299, i32 305, i32 311, i32 317, i32 323, i32 329, i32 335, i32 341, i32 347, i32 353, i32 359, i32 365, i32 371, i32 377, i32 383>
  store <64 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <64 x i8> %strided.vec1, ptr %out.vec1, align 64
  store <64 x i8> %strided.vec2, ptr %out.vec2, align 64
  store <64 x i8> %strided.vec3, ptr %out.vec3, align 64
  store <64 x i8> %strided.vec4, ptr %out.vec4, align 64
  store <64 x i8> %strided.vec5, ptr %out.vec5, align 64
  ret void
}
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; AVX1: {{.*}}
; AVX2-FAST: {{.*}}
; AVX2-FAST-PERLANE: {{.*}}
; AVX2-SLOW: {{.*}}
; AVX512: {{.*}}
; AVX512-FAST: {{.*}}
; AVX512-SLOW: {{.*}}
; AVX512BW-FAST: {{.*}}
; AVX512BW-ONLY-FAST: {{.*}}
; AVX512BW-ONLY-SLOW: {{.*}}
; AVX512BW-SLOW: {{.*}}
; AVX512DQ-FAST: {{.*}}
; AVX512DQ-SLOW: {{.*}}
; AVX512DQBW-FAST: {{.*}}
; AVX512DQBW-SLOW: {{.*}}
; AVX512F-FAST: {{.*}}
; AVX512F-ONLY-FAST: {{.*}}
; AVX512F-ONLY-SLOW: {{.*}}
; AVX512F-SLOW: {{.*}}
; FALLBACK0: {{.*}}
; FALLBACK1: {{.*}}
; FALLBACK10: {{.*}}
; FALLBACK11: {{.*}}
; FALLBACK12: {{.*}}
; FALLBACK2: {{.*}}
; FALLBACK3: {{.*}}
; FALLBACK4: {{.*}}
; FALLBACK5: {{.*}}
; FALLBACK6: {{.*}}
; FALLBACK7: {{.*}}
; FALLBACK8: {{.*}}
; FALLBACK9: {{.*}}