llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-5.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,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,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,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,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,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,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,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,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12

; These patterns are produced by LoopVectorizer for interleaved loads.

define void @load_i32_stride5_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4) nounwind {
; SSE-LABEL: load_i32_stride5_vf2:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa (%rdi), %xmm0
; SSE-NEXT:    movdqa 16(%rdi), %xmm1
; SSE-NEXT:    movdqa 32(%rdi), %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,1,1,1]
; SSE-NEXT:    movdqa %xmm0, %xmm4
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm3[0],xmm5[1],xmm3[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,2,3,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm0[3,3,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm3[2],xmm0[3],xmm3[3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
; SSE-NEXT:    movq %xmm4, (%rsi)
; SSE-NEXT:    movq %xmm5, (%rdx)
; SSE-NEXT:    movq %xmm0, (%rcx)
; SSE-NEXT:    movq %xmm6, (%r8)
; SSE-NEXT:    movq %xmm1, (%r9)
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i32_stride5_vf2:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    vmovdqa (%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovdqa 32(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm3 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm4 = xmm0[0,1,2,3],xmm1[4,5,6,7]
; AVX1-ONLY-NEXT:    vpshufd {{.*#+}} xmm4 = xmm4[1,2,2,3]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm5 = xmm0[0,1,2,3,4,5],xmm1[6,7]
; AVX1-ONLY-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
; AVX1-ONLY-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],mem[2,3],xmm1[4,5,6,7]
; AVX1-ONLY-NEXT:    vmovq %xmm3, (%rsi)
; AVX1-ONLY-NEXT:    vmovq %xmm4, (%rdx)
; AVX1-ONLY-NEXT:    vpextrq $1, %xmm5, (%rcx)
; AVX1-ONLY-NEXT:    vmovq %xmm0, (%r8)
; AVX1-ONLY-NEXT:    vmovq %xmm1, (%r9)
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i32_stride5_vf2:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %xmm0
; AVX2-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %xmm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} xmm4 = xmm4[1,2,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} xmm1 = xmm0[0,1,2],xmm1[3]
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
; AVX2-ONLY-NEXT:    vpbroadcastd 16(%rdi), %ymm5
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} xmm2 = xmm5[0],xmm2[1],xmm5[2,3]
; AVX2-ONLY-NEXT:    vmovq %xmm3, (%rsi)
; AVX2-ONLY-NEXT:    vmovq %xmm4, (%rdx)
; AVX2-ONLY-NEXT:    vpextrq $1, %xmm1, (%rcx)
; AVX2-ONLY-NEXT:    vmovq %xmm0, (%r8)
; AVX2-ONLY-NEXT:    vmovq %xmm2, (%r9)
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-SLOW-LABEL: load_i32_stride5_vf2:
; AVX512F-SLOW:       # %bb.0:
; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX512F-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm2
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm4 = xmm4[1,2,2,3]
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm1 = xmm0[0,1,2],xmm1[3]
; AVX512F-SLOW-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
; AVX512F-SLOW-NEXT:    vpbroadcastd 16(%rdi), %ymm5
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm2 = xmm5[0],xmm2[1],xmm5[2,3]
; AVX512F-SLOW-NEXT:    vmovq %xmm3, (%rsi)
; AVX512F-SLOW-NEXT:    vmovq %xmm4, (%rdx)
; AVX512F-SLOW-NEXT:    vpextrq $1, %xmm1, (%rcx)
; AVX512F-SLOW-NEXT:    vmovq %xmm0, (%r8)
; AVX512F-SLOW-NEXT:    vmovq %xmm2, (%r9)
; AVX512F-SLOW-NEXT:    vzeroupper
; AVX512F-SLOW-NEXT:    retq
;
; AVX512F-FAST-LABEL: load_i32_stride5_vf2:
; AVX512F-FAST:       # %bb.0:
; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX512F-FAST-NEXT:    vmovdqa 32(%rdi), %xmm2
; AVX512F-FAST-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm4 = [1,6,1,6]
; AVX512F-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm4
; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [2,7,2,7]
; AVX512F-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm5
; AVX512F-FAST-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
; AVX512F-FAST-NEXT:    vpbroadcastd 16(%rdi), %ymm1
; AVX512F-FAST-NEXT:    vpblendd {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2,3]
; AVX512F-FAST-NEXT:    vmovq %xmm3, (%rsi)
; AVX512F-FAST-NEXT:    vmovq %xmm4, (%rdx)
; AVX512F-FAST-NEXT:    vmovq %xmm5, (%rcx)
; AVX512F-FAST-NEXT:    vmovq %xmm0, (%r8)
; AVX512F-FAST-NEXT:    vmovq %xmm1, (%r9)
; AVX512F-FAST-NEXT:    vzeroupper
; AVX512F-FAST-NEXT:    retq
;
; AVX512BW-SLOW-LABEL: load_i32_stride5_vf2:
; AVX512BW-SLOW:       # %bb.0:
; AVX512BW-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX512BW-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm2
; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm4 = xmm4[1,2,2,3]
; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm1 = xmm0[0,1,2],xmm1[3]
; AVX512BW-SLOW-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
; AVX512BW-SLOW-NEXT:    vpbroadcastd 16(%rdi), %ymm5
; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm2 = xmm5[0],xmm2[1],xmm5[2,3]
; AVX512BW-SLOW-NEXT:    vmovq %xmm3, (%rsi)
; AVX512BW-SLOW-NEXT:    vmovq %xmm4, (%rdx)
; AVX512BW-SLOW-NEXT:    vpextrq $1, %xmm1, (%rcx)
; AVX512BW-SLOW-NEXT:    vmovq %xmm0, (%r8)
; AVX512BW-SLOW-NEXT:    vmovq %xmm2, (%r9)
; AVX512BW-SLOW-NEXT:    vzeroupper
; AVX512BW-SLOW-NEXT:    retq
;
; AVX512BW-FAST-LABEL: load_i32_stride5_vf2:
; AVX512BW-FAST:       # %bb.0:
; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX512BW-FAST-NEXT:    vmovdqa 32(%rdi), %xmm2
; AVX512BW-FAST-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm4 = [1,6,1,6]
; AVX512BW-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm4
; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [2,7,2,7]
; AVX512BW-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm5
; AVX512BW-FAST-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
; AVX512BW-FAST-NEXT:    vpbroadcastd 16(%rdi), %ymm1
; AVX512BW-FAST-NEXT:    vpblendd {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2,3]
; AVX512BW-FAST-NEXT:    vmovq %xmm3, (%rsi)
; AVX512BW-FAST-NEXT:    vmovq %xmm4, (%rdx)
; AVX512BW-FAST-NEXT:    vmovq %xmm5, (%rcx)
; AVX512BW-FAST-NEXT:    vmovq %xmm0, (%r8)
; AVX512BW-FAST-NEXT:    vmovq %xmm1, (%r9)
; AVX512BW-FAST-NEXT:    vzeroupper
; AVX512BW-FAST-NEXT:    retq
  %wide.vec = load <10 x i32>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <10 x i32> %wide.vec, <10 x i32> poison, <2 x i32> <i32 0, i32 5>
  %strided.vec1 = shufflevector <10 x i32> %wide.vec, <10 x i32> poison, <2 x i32> <i32 1, i32 6>
  %strided.vec2 = shufflevector <10 x i32> %wide.vec, <10 x i32> poison, <2 x i32> <i32 2, i32 7>
  %strided.vec3 = shufflevector <10 x i32> %wide.vec, <10 x i32> poison, <2 x i32> <i32 3, i32 8>
  %strided.vec4 = shufflevector <10 x i32> %wide.vec, <10 x i32> poison, <2 x i32> <i32 4, i32 9>
  store <2 x i32> %strided.vec0, ptr %out.vec0, align 64
  store <2 x i32> %strided.vec1, ptr %out.vec1, align 64
  store <2 x i32> %strided.vec2, ptr %out.vec2, align 64
  store <2 x i32> %strided.vec3, ptr %out.vec3, align 64
  store <2 x i32> %strided.vec4, ptr %out.vec4, align 64
  ret void
}

define void @load_i32_stride5_vf4(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4) nounwind {
; SSE-LABEL: load_i32_stride5_vf4:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa 64(%rdi), %xmm1
; SSE-NEXT:    movdqa (%rdi), %xmm5
; SSE-NEXT:    movdqa 16(%rdi), %xmm2
; SSE-NEXT:    movdqa 32(%rdi), %xmm4
; SSE-NEXT:    movdqa 48(%rdi), %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,1,1]
; SSE-NEXT:    movdqa %xmm5, %xmm6
; SSE-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm3 = xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm6[0],xmm3[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm2[2,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm7 = xmm5[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm6[0],xmm7[1],xmm6[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm4[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm8 = xmm8[0],xmm6[0],xmm8[1],xmm6[1]
; SSE-NEXT:    movsd {{.*#+}} xmm8 = xmm7[0],xmm8[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm2[2,2,3,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm7 = xmm5[3,3,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm5 = xmm5[2],xmm6[2],xmm5[3],xmm6[3]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm0[0,0,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm1[0],xmm6[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm6 = xmm5[0],xmm6[1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm4[0],xmm7[1],xmm4[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[2,2,3,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm9 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm7[0],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm9 = xmm9[2],xmm1[2],xmm9[3],xmm1[3]
; SSE-NEXT:    movsd {{.*#+}} xmm9 = xmm2[0],xmm9[1]
; SSE-NEXT:    movapd %xmm3, (%rsi)
; SSE-NEXT:    movapd %xmm8, (%rdx)
; SSE-NEXT:    movapd %xmm6, (%rcx)
; SSE-NEXT:    movapd %xmm0, (%r8)
; SSE-NEXT:    movapd %xmm9, (%r9)
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i32_stride5_vf4:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %ymm0
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %ymm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3]
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %xmm3
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm5 = xmm3[0,1],xmm4[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm5 = xmm5[0,1],mem[2],xmm5[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm5 = xmm5[1,2,3,3]
; AVX1-ONLY-NEXT:    vmovaps 64(%rdi), %xmm6
; AVX1-ONLY-NEXT:    vinsertps {{.*#+}} xmm5 = xmm5[0,1,2],xmm6[0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm7 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm7, %xmm7
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm7 = xmm7[0,1],mem[2],xmm7[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm7 = xmm7[1,0]
; AVX1-ONLY-NEXT:    vinsertps {{.*#+}} xmm7 = xmm7[0,1,2],xmm6[1]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm4[0,1],xmm3[2,3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm3 = xmm3[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vinsertps {{.*#+}} xmm3 = xmm3[0,1,2],xmm6[2]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],mem[1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0,1,2],xmm6[3]
; AVX1-ONLY-NEXT:    vmovaps %xmm2, (%rsi)
; AVX1-ONLY-NEXT:    vmovaps %xmm5, (%rdx)
; AVX1-ONLY-NEXT:    vmovaps %xmm7, (%rcx)
; AVX1-ONLY-NEXT:    vmovaps %xmm3, (%r8)
; AVX1-ONLY-NEXT:    vmovaps %xmm0, (%r9)
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i32_stride5_vf4:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,5,2,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm2, %ymm2
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm3 = <1,6,3,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm4, %ymm3, %ymm3
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %xmm4
; AVX2-ONLY-NEXT:    vpbroadcastd %xmm4, %xmm5
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} xmm3 = xmm3[0,1,2],xmm5[3]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm5 = <2,7,4,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm6 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm6, %ymm5, %ymm5
; AVX2-ONLY-NEXT:    vpbroadcastd 68(%rdi), %xmm6
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} xmm5 = xmm5[0,1,2],xmm6[3]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} xmm6 = xmm4[2,2,2,2]
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm7 = ymm0[12,13,14,15],ymm1[0,1,2,3,4,5,6,7,8,9,10,11],ymm0[28,29,30,31],ymm1[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm7 = ymm7[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} xmm6 = xmm7[0,1,2],xmm6[3]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm7 = <4,1,6,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm7, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0,1,2],xmm4[3]
; AVX2-ONLY-NEXT:    vmovdqa %xmm2, (%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %xmm3, (%rdx)
; AVX2-ONLY-NEXT:    vmovdqa %xmm5, (%rcx)
; AVX2-ONLY-NEXT:    vmovdqa %xmm6, (%r8)
; AVX2-ONLY-NEXT:    vmovdqa %xmm0, (%r9)
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512-LABEL: load_i32_stride5_vf4:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmovdqa64 (%rdi), %zmm0
; AVX512-NEXT:    vmovdqa64 64(%rdi), %zmm1
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,5,10,15]
; AVX512-NEXT:    vmovdqa (%rdi), %ymm3
; AVX512-NEXT:    vpermt2d 32(%rdi), %ymm2, %ymm3
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm2 = [1,6,11,16]
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm2
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm4 = [2,7,12,17]
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm4
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm5 = [3,8,13,18]
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm5
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm6 = [4,9,14,19]
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm6
; AVX512-NEXT:    vmovdqa %xmm3, (%rsi)
; AVX512-NEXT:    vmovdqa %xmm2, (%rdx)
; AVX512-NEXT:    vmovdqa %xmm4, (%rcx)
; AVX512-NEXT:    vmovdqa %xmm5, (%r8)
; AVX512-NEXT:    vmovdqa %xmm6, (%r9)
; AVX512-NEXT:    vzeroupper
; AVX512-NEXT:    retq
  %wide.vec = load <20 x i32>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <20 x i32> %wide.vec, <20 x i32> poison, <4 x i32> <i32 0, i32 5, i32 10, i32 15>
  %strided.vec1 = shufflevector <20 x i32> %wide.vec, <20 x i32> poison, <4 x i32> <i32 1, i32 6, i32 11, i32 16>
  %strided.vec2 = shufflevector <20 x i32> %wide.vec, <20 x i32> poison, <4 x i32> <i32 2, i32 7, i32 12, i32 17>
  %strided.vec3 = shufflevector <20 x i32> %wide.vec, <20 x i32> poison, <4 x i32> <i32 3, i32 8, i32 13, i32 18>
  %strided.vec4 = shufflevector <20 x i32> %wide.vec, <20 x i32> poison, <4 x i32> <i32 4, i32 9, i32 14, i32 19>
  store <4 x i32> %strided.vec0, ptr %out.vec0, align 64
  store <4 x i32> %strided.vec1, ptr %out.vec1, align 64
  store <4 x i32> %strided.vec2, ptr %out.vec2, align 64
  store <4 x i32> %strided.vec3, ptr %out.vec3, align 64
  store <4 x i32> %strided.vec4, ptr %out.vec4, align 64
  ret void
}

define void @load_i32_stride5_vf8(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4) nounwind {
; SSE-LABEL: load_i32_stride5_vf8:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa 144(%rdi), %xmm3
; SSE-NEXT:    movdqa 64(%rdi), %xmm0
; SSE-NEXT:    movdqa 128(%rdi), %xmm4
; SSE-NEXT:    movdqa 112(%rdi), %xmm5
; SSE-NEXT:    movdqa 80(%rdi), %xmm11
; SSE-NEXT:    movdqa 96(%rdi), %xmm1
; SSE-NEXT:    movdqa (%rdi), %xmm14
; SSE-NEXT:    movdqa 16(%rdi), %xmm7
; SSE-NEXT:    movdqa 32(%rdi), %xmm9
; SSE-NEXT:    movdqa 48(%rdi), %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm7[1,1,1,1]
; SSE-NEXT:    movdqa %xmm14, %xmm8
; SSE-NEXT:    punpckldq {{.*#+}} xmm8 = xmm8[0],xmm6[0],xmm8[1],xmm6[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm6 = xmm9[2,2,2,2]
; SSE-NEXT:    movdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm6 = xmm6[2],xmm2[2],xmm6[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm6 = xmm8[0],xmm6[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm1[1,1,1,1]
; SSE-NEXT:    movdqa %xmm11, %xmm10
; SSE-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm8[0],xmm10[1],xmm8[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm5[2,2,2,2]
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm8 = xmm8[2],xmm4[2],xmm8[3],xmm4[3]
; SSE-NEXT:    movsd {{.*#+}} xmm8 = xmm10[0],xmm8[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm7[2,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm12 = xmm14[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm12 = xmm12[0],xmm10[0],xmm12[1],xmm10[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm13 = xmm0[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm10 = xmm9[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm13[0],xmm10[1],xmm13[1]
; SSE-NEXT:    movsd {{.*#+}} xmm10 = xmm12[0],xmm10[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm12 = xmm1[2,3,2,3]
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm13 = xmm11[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm13 = xmm13[0],xmm12[0],xmm13[1],xmm12[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm15 = xmm3[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm12 = xmm5[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm12 = xmm12[0],xmm15[0],xmm12[1],xmm15[1]
; SSE-NEXT:    movsd {{.*#+}} xmm12 = xmm13[0],xmm12[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm13 = xmm7[2,2,3,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm15 = xmm14[3,3,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm14 = xmm14[2],xmm13[2],xmm14[3],xmm13[3]
; SSE-NEXT:    pshufd {{.*#+}} xmm13 = xmm2[0,0,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm13 = xmm13[0],xmm0[0],xmm13[1],xmm0[1]
; SSE-NEXT:    movdqa %xmm0, %xmm9
; SSE-NEXT:    movsd {{.*#+}} xmm13 = xmm14[0],xmm13[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm14 = xmm1[2,2,3,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[3,3,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm11 = xmm11[2],xmm14[2],xmm11[3],xmm14[3]
; SSE-NEXT:    pshufd {{.*#+}} xmm14 = xmm4[0,0,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm14 = xmm14[0],xmm3[0],xmm14[1],xmm3[1]
; SSE-NEXT:    movdqa %xmm3, %xmm5
; SSE-NEXT:    movsd {{.*#+}} xmm14 = xmm11[0],xmm14[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,2,3,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm11 = xmm2[2,2,2,2]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm15[0],xmm2[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm5[2,2,3,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm15 = xmm4[2,2,2,2]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm1[0],xmm4[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm0[0],xmm7[1],xmm0[1]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm11 = xmm11[2],xmm9[2],xmm11[3],xmm9[3]
; SSE-NEXT:    movsd {{.*#+}} xmm11 = xmm7[0],xmm11[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm15 = xmm15[2],xmm5[2],xmm15[3],xmm5[3]
; SSE-NEXT:    movsd {{.*#+}} xmm15 = xmm1[0],xmm15[1]
; SSE-NEXT:    movapd %xmm8, 16(%rsi)
; SSE-NEXT:    movapd %xmm6, (%rsi)
; SSE-NEXT:    movapd %xmm12, 16(%rdx)
; SSE-NEXT:    movapd %xmm10, (%rdx)
; SSE-NEXT:    movapd %xmm14, 16(%rcx)
; SSE-NEXT:    movapd %xmm13, (%rcx)
; SSE-NEXT:    movapd %xmm4, 16(%r8)
; SSE-NEXT:    movapd %xmm2, (%r8)
; SSE-NEXT:    movapd %xmm15, 16(%r9)
; SSE-NEXT:    movapd %xmm11, (%r9)
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i32_stride5_vf8:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    vmovaps 128(%rdi), %ymm0
; AVX1-ONLY-NEXT:    vmovaps 64(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vmovaps 96(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm5 = ymm4[0,1],ymm3[2,3],ymm4[4,5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm5, %xmm6
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm5 = xmm5[0],xmm6[1],xmm5[2],xmm6[3]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 128(%rdi), %ymm5, %ymm7
; AVX1-ONLY-NEXT:    vmovaps 96(%rdi), %xmm6
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm6, %ymm0, %ymm8
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm9 = ymm8[0,1,2,3,4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm9 = ymm9[0,1,2,3],ymm1[4],ymm9[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm5 = ymm5[0,1,2,3],ymm9[4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm5 = ymm5[0,1,2],ymm7[3],ymm5[4,5,6],ymm7[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm7 = ymm1[0,1],ymm2[2,3],ymm1[4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm9 = ymm2[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm9 = ymm9[2,1],ymm1[1,3],ymm9[6,5],ymm1[5,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm7 = ymm9[2,0],ymm7[3,0],ymm9[6,4],ymm7[7,4]
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %xmm9
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %xmm10
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm11 = xmm9[0,1],xmm10[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm11 = xmm11[0,1],mem[2],xmm11[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm11 = xmm11[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm7 = ymm11[0,1,2],ymm7[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 144(%rdi), %ymm11
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4,5,6],ymm11[7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm8 = ymm8[3,0],ymm1[2,0],ymm8[7,4],ymm1[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm8 = ymm8[2,0],ymm1[2,1],ymm8[6,4],ymm1[6,5]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm11 = ymm4[0,1,2,3],ymm3[4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm11, %xmm11
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm11 = xmm11[0,1],mem[2],xmm11[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm11 = xmm11[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm8 = ymm11[0,1,2],ymm8[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm11 = ymm0[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm12 = ymm0[1,0],ymm11[0,0],ymm0[5,4],ymm11[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm12 = ymm12[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm8 = ymm8[0,1,2,3,4,5],ymm12[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm12 = ymm2[0,0],ymm1[3,0],ymm2[4,4],ymm1[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm12 = ymm12[2,0],ymm1[2,2],ymm12[6,4],ymm1[6,6]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm9 = xmm10[0,1],xmm9[2,3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm9 = xmm9[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm9 = ymm9[0,1,2],ymm12[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm10 = ymm0[2,0],ymm11[1,0],ymm0[6,4],ymm11[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm10 = ymm10[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm9 = ymm9[0,1,2,3,4,5],ymm10[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm6, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm2[5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0],mem[1],xmm2[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm11[0,1,2,3,4,5,6],ymm0[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3,4,5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vmovaps %ymm5, (%rsi)
; AVX1-ONLY-NEXT:    vmovaps %ymm7, (%rdx)
; AVX1-ONLY-NEXT:    vmovaps %ymm8, (%rcx)
; AVX1-ONLY-NEXT:    vmovaps %ymm9, (%r8)
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%r9)
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i32_stride5_vf8:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    vmovdqa 128(%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqa 96(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm5 = ymm4[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm5 = ymm5[0,1,2,3],ymm3[4],ymm5[5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm6 = [0,5,2,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm1[0,1],ymm2[2,3],ymm1[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm7, %ymm6, %ymm6
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm5 = ymm6[0,1,2,3],ymm5[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 128(%rdi), %ymm6, %ymm6
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm5 = ymm5[0,1,2],ymm6[3],ymm5[4,5,6],ymm6[7]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm6 = <1,6,3,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm1[0,1],ymm2[2,3],ymm1[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm7, %ymm6, %ymm6
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm3[0,1],ymm4[2,3],ymm3[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vbroadcasti128 {{.*#+}} ymm8 = [5,2,7,0,5,2,7,0]
; AVX2-ONLY-NEXT:    # ymm8 = mem[0,1,0,1]
; AVX2-ONLY-NEXT:    vpermd %ymm7, %ymm8, %ymm7
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm6 = ymm6[0,1,2],ymm7[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 144(%rdi), %ymm7
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm6 = ymm6[0,1,2,3,4,5,6],ymm7[7]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm7 = <2,7,4,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm8, %ymm7, %ymm7
; AVX2-ONLY-NEXT:    vinserti128 $1, 96(%rdi), %ymm0, %ymm8
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm3[0,1,2,3,4,5,6],ymm8[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm8 = ymm8[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2],ymm8[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm8 = [0,5,0,5,0,5,0,5]
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm8, %ymm8
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2,3,4,5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm8 = [1,6,1,6,1,6,1,6]
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm8, %ymm8
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm9 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm9 = ymm9[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm10 = ymm1[12,13,14,15],ymm2[0,1,2,3,4,5,6,7,8,9,10,11],ymm1[28,29,30,31],ymm2[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm10 = ymm10[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm9 = ymm10[0,1,2],ymm9[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm9[0,1,2,3,4,5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm3 = ymm3[0,1],ymm4[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3,4],ymm4[5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm4 = <4,1,6,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2,3],ymm1[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm4, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [2,7,2,7,2,7,2,7]
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm2, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1,2,3,4,5],ymm0[6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm5, (%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %ymm6, (%rdx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm7, (%rcx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm8, (%r8)
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, (%r9)
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512-LABEL: load_i32_stride5_vf8:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmovdqa64 (%rdi), %zmm1
; AVX512-NEXT:    vmovdqa64 64(%rdi), %zmm2
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm0 = <0,5,10,15,20,25,30,u>
; AVX512-NEXT:    vpermi2d %zmm2, %zmm1, %zmm0
; AVX512-NEXT:    vinserti128 $1, 128(%rdi), %ymm0, %ymm3
; AVX512-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm3[3],ymm0[4,5,6],ymm3[7]
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm3 = <17,22,27,0,5,10,15,u>
; AVX512-NEXT:    vpermi2d %zmm1, %zmm2, %zmm3
; AVX512-NEXT:    vpbroadcastd 144(%rdi), %ymm4
; AVX512-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm4 = <2,7,12,17,22,27,u,u>
; AVX512-NEXT:    vpermi2d %zmm2, %zmm1, %zmm4
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm5 = [0,1,2,3,4,5,8,13]
; AVX512-NEXT:    vmovdqa 128(%rdi), %ymm6
; AVX512-NEXT:    vpermi2d %ymm6, %ymm4, %ymm5
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm4 = <3,8,13,18,23,28,u,u>
; AVX512-NEXT:    vpermi2d %zmm2, %zmm1, %zmm4
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm7 = [0,1,2,3,4,5,9,14]
; AVX512-NEXT:    vpermi2d %ymm6, %ymm4, %ymm7
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm4 = <4,9,14,19,24,29,u,u>
; AVX512-NEXT:    vpermi2d %zmm2, %zmm1, %zmm4
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,1,2,3,4,5,10,15]
; AVX512-NEXT:    vpermi2d %ymm6, %ymm4, %ymm1
; AVX512-NEXT:    vmovdqa %ymm0, (%rsi)
; AVX512-NEXT:    vmovdqa %ymm3, (%rdx)
; AVX512-NEXT:    vmovdqa %ymm5, (%rcx)
; AVX512-NEXT:    vmovdqa %ymm7, (%r8)
; AVX512-NEXT:    vmovdqa %ymm1, (%r9)
; AVX512-NEXT:    vzeroupper
; AVX512-NEXT:    retq
  %wide.vec = load <40 x i32>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <40 x i32> %wide.vec, <40 x i32> poison, <8 x i32> <i32 0, i32 5, i32 10, i32 15, i32 20, i32 25, i32 30, i32 35>
  %strided.vec1 = shufflevector <40 x i32> %wide.vec, <40 x i32> poison, <8 x i32> <i32 1, i32 6, i32 11, i32 16, i32 21, i32 26, i32 31, i32 36>
  %strided.vec2 = shufflevector <40 x i32> %wide.vec, <40 x i32> poison, <8 x i32> <i32 2, i32 7, i32 12, i32 17, i32 22, i32 27, i32 32, i32 37>
  %strided.vec3 = shufflevector <40 x i32> %wide.vec, <40 x i32> poison, <8 x i32> <i32 3, i32 8, i32 13, i32 18, i32 23, i32 28, i32 33, i32 38>
  %strided.vec4 = shufflevector <40 x i32> %wide.vec, <40 x i32> poison, <8 x i32> <i32 4, i32 9, i32 14, i32 19, i32 24, i32 29, i32 34, i32 39>
  store <8 x i32> %strided.vec0, ptr %out.vec0, align 64
  store <8 x i32> %strided.vec1, ptr %out.vec1, align 64
  store <8 x i32> %strided.vec2, ptr %out.vec2, align 64
  store <8 x i32> %strided.vec3, ptr %out.vec3, align 64
  store <8 x i32> %strided.vec4, ptr %out.vec4, align 64
  ret void
}

define void @load_i32_stride5_vf16(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4) nounwind {
; SSE-LABEL: load_i32_stride5_vf16:
; SSE:       # %bb.0:
; SSE-NEXT:    subq $296, %rsp # imm = 0x128
; SSE-NEXT:    movdqa 208(%rdi), %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 192(%rdi), %xmm4
; SSE-NEXT:    movdqa 160(%rdi), %xmm9
; SSE-NEXT:    movdqa 176(%rdi), %xmm2
; SSE-NEXT:    movdqa 288(%rdi), %xmm10
; SSE-NEXT:    movdqa %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 272(%rdi), %xmm6
; SSE-NEXT:    movdqa 240(%rdi), %xmm11
; SSE-NEXT:    movdqa 256(%rdi), %xmm15
; SSE-NEXT:    movdqa (%rdi), %xmm5
; SSE-NEXT:    movdqa 16(%rdi), %xmm12
; SSE-NEXT:    movdqa 32(%rdi), %xmm8
; SSE-NEXT:    movdqa 48(%rdi), %xmm7
; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm12[1,1,1,1]
; SSE-NEXT:    movdqa %xmm5, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm8[2,2,2,2]
; SSE-NEXT:    movdqa %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm7[2],xmm0[3],xmm7[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm15[1,1,1,1]
; SSE-NEXT:    movdqa %xmm11, %xmm1
; SSE-NEXT:    movdqa %xmm11, %xmm14
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[2,2,2,2]
; SSE-NEXT:    movdqa %xmm6, %xmm11
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm10[2],xmm0[3],xmm10[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[1,1,1,1]
; SSE-NEXT:    movdqa %xmm2, %xmm6
; SSE-NEXT:    movdqa %xmm9, %xmm7
; SSE-NEXT:    movdqa %xmm9, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa %xmm4, %xmm2
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm3[2],xmm0[3],xmm3[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 80(%rdi), %xmm10
; SSE-NEXT:    movdqa 96(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa %xmm10, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 128(%rdi), %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 112(%rdi), %xmm9
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm9[2,2,2,2]
; SSE-NEXT:    movdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm3[2],xmm0[3],xmm3[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm5[1,1,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm12[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 64(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm8[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm14[1,1,1,1]
; SSE-NEXT:    movdqa %xmm14, %xmm13
; SSE-NEXT:    movdqa %xmm14, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm15[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 304(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm11[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm7[1,1,1,1]
; SSE-NEXT:    movdqa %xmm7, %xmm4
; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 224(%rdi), %xmm14
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm14[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm10, %xmm7
; SSE-NEXT:    movdqa %xmm10, (%rsp) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm10[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 144(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm9[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
; SSE-NEXT:    movapd %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm5, %xmm9
; SSE-NEXT:    movdqa %xmm5, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm12[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm8[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm13, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm15[2,2,3,3]
; SSE-NEXT:    movdqa %xmm15, %xmm11
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[2,2,3,3]
; SSE-NEXT:    movdqa %xmm6, %xmm10
; SSE-NEXT:    punpckhdq {{.*#+}} xmm4 = xmm4[2],xmm1[2],xmm4[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[0,0,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm14[0],xmm1[1],xmm14[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm4[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm13 = xmm4[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm13 = xmm13[0],xmm1[0],xmm13[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm13 = xmm0[0],xmm13[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm5[2,2,3,3]
; SSE-NEXT:    movdqa %xmm8, %xmm7
; SSE-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm0[0],xmm7[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm9[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm8[0],xmm0[1],xmm8[1]
; SSE-NEXT:    movsd {{.*#+}} xmm7 = xmm0[0],xmm7[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm15[2,2,3,3]
; SSE-NEXT:    movdqa %xmm2, %xmm5
; SSE-NEXT:    movdqa %xmm2, %xmm15
; SSE-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1]
; SSE-NEXT:    movsd {{.*#+}} xmm5 = xmm0[0],xmm5[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm14[2,2,3,3]
; SSE-NEXT:    movdqa %xmm6, %xmm2
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,2,3,3]
; SSE-NEXT:    movdqa %xmm4, %xmm0
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    pshufd $255, (%rsp), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm8[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm12 = xmm12[0],xmm1[0],xmm12[1],xmm1[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[2],mem[2],xmm1[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm12[0],xmm1[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm12 = xmm9[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm11 = xmm11[0],xmm12[0],xmm11[1],xmm12[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm12 = xmm15[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
; SSE-NEXT:    # xmm12 = xmm12[2],mem[2],xmm12[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm12 = xmm11[0],xmm12[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm15 = xmm6[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm15[0],xmm10[1],xmm15[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Folded Reload
; SSE-NEXT:    # xmm15 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm15 = xmm15[2],xmm14[2],xmm15[3],xmm14[3]
; SSE-NEXT:    movsd {{.*#+}} xmm15 = xmm10[0],xmm15[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm14 = xmm3[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm14[0],xmm10[1],xmm14[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm14 = xmm4[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
; SSE-NEXT:    # xmm14 = xmm14[2],mem[2],xmm14[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm14 = xmm10[0],xmm14[1]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    movaps %xmm10, 16(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    movaps %xmm11, 32(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    movaps %xmm11, 48(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    movaps %xmm11, (%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    movaps %xmm3, 16(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    movaps %xmm3, 32(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    movaps %xmm3, 48(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    movaps %xmm3, (%rdx)
; SSE-NEXT:    movapd %xmm13, 16(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    movaps %xmm3, 32(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    movaps %xmm3, 48(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    movaps %xmm3, (%rcx)
; SSE-NEXT:    movapd %xmm0, 16(%r8)
; SSE-NEXT:    movapd %xmm2, 32(%r8)
; SSE-NEXT:    movapd %xmm5, 48(%r8)
; SSE-NEXT:    movapd %xmm7, (%r8)
; SSE-NEXT:    movapd %xmm14, 16(%r9)
; SSE-NEXT:    movapd %xmm15, 32(%r9)
; SSE-NEXT:    movapd %xmm12, 48(%r9)
; SSE-NEXT:    movapd %xmm1, (%r9)
; SSE-NEXT:    addq $296, %rsp # imm = 0x128
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i32_stride5_vf16:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    subq $168, %rsp
; AVX1-ONLY-NEXT:    vmovaps 224(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovaps 256(%rdi), %ymm11
; AVX1-ONLY-NEXT:    vmovaps 192(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vmovups %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 160(%rdi), %ymm12
; AVX1-ONLY-NEXT:    vmovaps 64(%rdi), %ymm5
; AVX1-ONLY-NEXT:    vmovaps 96(%rdi), %ymm6
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %ymm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1],ymm1[2,3],ymm3[4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vmovaps %ymm3, %ymm7
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps %ymm1, %ymm13
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 128(%rdi), %ymm0, %ymm1
; AVX1-ONLY-NEXT:    vmovaps 96(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm14
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm14[0,1,2,3,4,5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm5[4],ymm3[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3],ymm0[4,5,6],ymm1[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm12[0,1],ymm4[2,3],ymm12[4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 288(%rdi), %ymm1, %ymm3
; AVX1-ONLY-NEXT:    vmovaps 256(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm0[0,1,2,3,4,5],ymm11[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3],ymm2[4],ymm4[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm4[4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2],ymm3[3],ymm1[4,5,6],ymm3[7]
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm5[0,1],ymm6[2,3],ymm5[4,5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm6[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm5[1,3],ymm3[6,5],ymm5[5,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm3[2,0],ymm1[3,0],ymm3[6,4],ymm1[7,4]
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %xmm15
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %xmm10
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm15[0,1],xmm10[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm3[0,1],mem[2],xmm3[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm3 = xmm3[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm3[0,1,2],ymm1[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 144(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5,6],ymm3[7]
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm11[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[2,1],ymm2[1,3],ymm1[6,5],ymm2[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm2[0,1],ymm11[2,3],ymm2[4,5],ymm11[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[2,0],ymm3[3,0],ymm1[6,4],ymm3[7,4]
; AVX1-ONLY-NEXT:    vmovaps 160(%rdi), %xmm9
; AVX1-ONLY-NEXT:    vmovaps 192(%rdi), %xmm8
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm9[0,1],xmm8[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm3[0,1],mem[2],xmm3[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm3 = xmm3[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm3[0,1,2],ymm1[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 304(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5,6],ymm3[7]
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm14[3,0],ymm5[2,0],ymm14[7,4],ymm5[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[2,0],ymm5[2,1],ymm1[6,4],ymm5[6,5]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm7[0,1,2,3],ymm13[4,5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm3, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm3[0,1],mem[2],xmm3[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm3 = xmm3[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2],ymm1[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 128(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm4 = ymm1[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm13 = ymm1[1,0],ymm4[0,0],ymm1[5,4],ymm4[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm13 = ymm13[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm13[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm3, (%rsp) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm2[2,0],ymm0[7,4],ymm2[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm2[2,1],ymm0[6,4],ymm2[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm12[0,1,2,3],ymm7[4,5],ymm12[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm3, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm3[0,1],mem[2],xmm3[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm3 = xmm3[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm13 = ymm3[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 288(%rdi), %ymm0
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm0[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm14 = ymm0[1,0],ymm3[0,0],ymm0[5,4],ymm3[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm14 = ymm14[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm13 = ymm13[0,1,2,3,4,5],ymm14[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm10 = xmm10[0,1],xmm15[2,3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm14 = ymm6[0,0],ymm5[3,0],ymm6[4,4],ymm5[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm14 = ymm14[2,0],ymm5[2,2],ymm14[6,4],ymm5[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm10 = xmm10[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm10 = ymm10[0,1,2],ymm14[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm14 = ymm1[2,0],ymm4[1,0],ymm1[6,4],ymm4[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm14 = ymm14[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm10 = ymm10[0,1,2,3,4,5],ymm14[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm8 = xmm8[0,1],xmm9[2,3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm9 = ymm11[0,0],ymm2[3,0],ymm11[4,4],ymm2[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm9 = ymm9[2,0],ymm2[2,2],ymm9[6,4],ymm2[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm8 = xmm8[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm8 = ymm8[0,1,2],ymm9[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm9 = ymm0[2,0],ymm3[1,0],ymm0[6,4],ymm3[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm9 = ymm9[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm8 = ymm8[0,1,2,3,4,5],ymm9[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm5 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm5 = ymm5[0,1,2,3,4],ymm6[5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm6 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm6 = ymm6[0,1,2,3],mem[4,5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm6, %xmm6
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm6 = xmm6[0],mem[1],xmm6[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm5 = ymm6[0,1,2],ymm5[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm4[0,1,2,3,4,5,6],ymm1[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm5[0,1,2,3,4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm2 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3,4],ymm11[5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm7[0,1,2,3],ymm12[4,5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm4, %xmm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0],mem[1],xmm4[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm4[0,1,2],ymm2[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5,6],ymm0[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3,4,5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm2, 32(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm2, (%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm2, 32(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm2, (%rdx)
; AVX1-ONLY-NEXT:    vmovaps %ymm13, 32(%rcx)
; AVX1-ONLY-NEXT:    vmovups (%rsp), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm2, (%rcx)
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 32(%r8)
; AVX1-ONLY-NEXT:    vmovaps %ymm10, (%r8)
; AVX1-ONLY-NEXT:    vmovaps %ymm0, 32(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm1, (%r9)
; AVX1-ONLY-NEXT:    addq $168, %rsp
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i32_stride5_vf16:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    pushq %rax
; AVX2-ONLY-NEXT:    vmovdqa 128(%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm5
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %ymm8
; AVX2-ONLY-NEXT:    vmovdqa 96(%rdi), %ymm9
; AVX2-ONLY-NEXT:    vmovdqa 224(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqa 256(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqa 192(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vmovdqa 160(%rdi), %ymm6
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm10 = [0,5,2,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm6[0,1],ymm4[2,3],ymm6[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm7, %ymm10, %ymm7
; AVX2-ONLY-NEXT:    vinserti128 $1, 288(%rdi), %ymm7, %ymm11
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm12 = ymm2[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm12 = ymm12[0,1,2,3],ymm1[4],ymm12[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2,3],ymm12[4,5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2],ymm11[3],ymm7[4,5,6],ymm11[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm11 = ymm9[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm11 = ymm11[0,1,2,3],ymm8[4],ymm11[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm12 = ymm3[0,1],ymm5[2,3],ymm3[4,5],ymm5[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm12, %ymm10, %ymm10
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm11 = ymm10[0,1,2,3],ymm11[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 128(%rdi), %ymm10, %ymm10
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm11[0,1,2],ymm10[3],ymm11[4,5,6],ymm10[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm12 = <1,6,3,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm11 = ymm3[0,1],ymm5[2,3],ymm3[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm11, %ymm12, %ymm11
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm8[0,1],ymm9[2,3],ymm8[4,5],ymm9[6,7]
; AVX2-ONLY-NEXT:    vbroadcasti128 {{.*#+}} ymm14 = [5,2,7,0,5,2,7,0]
; AVX2-ONLY-NEXT:    # ymm14 = mem[0,1,0,1]
; AVX2-ONLY-NEXT:    vpermd %ymm13, %ymm14, %ymm13
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm11 = ymm11[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 144(%rdi), %ymm13
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm11[0,1,2,3,4,5,6],ymm13[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm6[0,1],ymm4[2,3],ymm6[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm13, %ymm12, %ymm12
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm1[0,1],ymm2[2,3],ymm1[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm13, %ymm14, %ymm13
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm12 = ymm12[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 304(%rdi), %ymm13
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm12[0,1,2,3,4,5,6],ymm13[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm14 = <2,7,4,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm3[0,1,2,3],ymm5[4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm13, %ymm14, %ymm13
; AVX2-ONLY-NEXT:    vinserti128 $1, 96(%rdi), %ymm0, %ymm15
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm15 = ymm8[0,1,2,3,4,5,6],ymm15[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm15 = ymm15[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm13[0,1,2],ymm15[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm15 = [0,5,0,5,0,5,0,5]
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm15, %ymm7
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm13[0,1,2,3,4,5],ymm7[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm6[0,1,2,3],ymm4[4,5],ymm6[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm7, %ymm14, %ymm7
; AVX2-ONLY-NEXT:    vinserti128 $1, 256(%rdi), %ymm0, %ymm14
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm14 = ymm1[0,1,2,3,4,5,6],ymm14[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm14 = ymm14[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2],ymm14[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 288(%rdi), %ymm14
; AVX2-ONLY-NEXT:    vpermd %ymm14, %ymm15, %ymm15
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm15 = ymm8[0,1,2,3],ymm9[4,5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm15 = ymm15[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm10 = ymm3[12,13,14,15],ymm5[0,1,2,3,4,5,6,7,8,9,10,11],ymm3[28,29,30,31],ymm5[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm10 = ymm10[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm10 = ymm10[0,1,2],ymm15[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm15 = [1,6,1,6,1,6,1,6]
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm15, %ymm11
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm10 = ymm10[0,1,2,3,4,5],ymm11[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm11 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm11 = ymm11[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm12 = ymm6[12,13,14,15],ymm4[0,1,2,3,4,5,6,7,8,9,10,11],ymm6[28,29,30,31],ymm4[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm12 = ymm12[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm11 = ymm12[0,1,2],ymm11[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm14, %ymm15, %ymm12
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm11 = ymm11[0,1,2,3,4,5],ymm12[6,7]
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm8 = ymm8[0,1],ymm9[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm8[0,1,2,3,4],ymm9[5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm5[0,1,2,3],ymm3[4,5],ymm5[6,7]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm5 = <4,1,6,u>
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm5, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm8[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm8 = [2,7,2,7,2,7,2,7]
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm8, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5],ymm0[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm6[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm5, %ymm3
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm1[0,1],ymm2[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm2[5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm3[0,1,2],ymm1[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm14, %ymm8, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm2, 32(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm2, (%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm2, 32(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm2, (%rdx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm7, 32(%rcx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm13, (%rcx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm11, 32(%r8)
; AVX2-ONLY-NEXT:    vmovdqa %ymm10, (%r8)
; AVX2-ONLY-NEXT:    vmovdqa %ymm1, 32(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, (%r9)
; AVX2-ONLY-NEXT:    popq %rax
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-LABEL: load_i32_stride5_vf16:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovdqa64 256(%rdi), %zmm0
; AVX512F-NEXT:    vmovdqa64 (%rdi), %zmm1
; AVX512F-NEXT:    vmovdqa64 64(%rdi), %zmm2
; AVX512F-NEXT:    vmovdqa64 128(%rdi), %zmm3
; AVX512F-NEXT:    vmovdqa64 192(%rdi), %zmm4
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm5 = [8,13,18,23,28,0,0,3,8,13,18,23,28,0,0,3]
; AVX512F-NEXT:    # zmm5 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpermi2d %zmm4, %zmm3, %zmm5
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm6 = <0,5,10,15,20,25,30,u>
; AVX512F-NEXT:    vpermi2d %zmm2, %zmm1, %zmm6
; AVX512F-NEXT:    movw $8064, %ax # imm = 0x1F80
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqa32 %zmm5, %zmm6 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm5 = [0,1,2,3,4,5,6,7,8,9,10,11,12,17,22,27]
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm6, %zmm5
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm6 = [9,14,19,24,29,0,0,4,9,14,19,24,29,0,0,4]
; AVX512F-NEXT:    # zmm6 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpermi2d %zmm4, %zmm3, %zmm6
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm7 = <17,22,27,0,5,10,15,u>
; AVX512F-NEXT:    vpermi2d %zmm1, %zmm2, %zmm7
; AVX512F-NEXT:    vmovdqa32 %zmm6, %zmm7 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm6 = [0,1,2,3,4,5,6,7,8,9,10,11,12,18,23,28]
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm7, %zmm6
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm7 = <2,7,12,17,22,27,u,u>
; AVX512F-NEXT:    vpermi2d %zmm2, %zmm1, %zmm7
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm8 = [10,15,20,25,30,0,0,5,10,15,20,25,30,0,0,5]
; AVX512F-NEXT:    # zmm8 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpermi2d %zmm4, %zmm3, %zmm8
; AVX512F-NEXT:    movb $7, %al
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqa64 %zmm7, %zmm8 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm7 = [0,1,2,3,4,5,6,7,8,9,10,11,12,19,24,29]
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm8, %zmm7
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm8 = <3,8,13,18,23,28,u,u>
; AVX512F-NEXT:    vpermi2d %zmm2, %zmm1, %zmm8
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm9 = [27,0,5,10,15,0,17,22,27,0,5,10,15,0,17,22]
; AVX512F-NEXT:    # zmm9 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpermi2d %zmm3, %zmm4, %zmm9
; AVX512F-NEXT:    vmovdqa64 %zmm8, %zmm9 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm8 = [0,1,2,3,4,5,6,7,8,9,10,11,12,20,25,30]
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm9, %zmm8
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm9 = [12,17,22,27,0,0,2,7,12,17,22,27,0,0,2,7]
; AVX512F-NEXT:    # zmm9 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpermi2d %zmm4, %zmm3, %zmm9
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = <4,9,14,19,24,29,u,u>
; AVX512F-NEXT:    vpermi2d %zmm2, %zmm1, %zmm3
; AVX512F-NEXT:    movb $56, %al
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqa64 %zmm9, %zmm3 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm1 = [0,1,2,3,4,5,6,7,8,9,10,11,16,21,26,31]
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm3, %zmm1
; AVX512F-NEXT:    vmovdqa64 %zmm5, (%rsi)
; AVX512F-NEXT:    vmovdqa64 %zmm6, (%rdx)
; AVX512F-NEXT:    vmovdqa64 %zmm7, (%rcx)
; AVX512F-NEXT:    vmovdqa64 %zmm8, (%r8)
; AVX512F-NEXT:    vmovdqa64 %zmm1, (%r9)
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i32_stride5_vf16:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovdqa64 256(%rdi), %zmm0
; AVX512BW-NEXT:    vmovdqa64 (%rdi), %zmm1
; AVX512BW-NEXT:    vmovdqa64 64(%rdi), %zmm2
; AVX512BW-NEXT:    vmovdqa64 128(%rdi), %zmm3
; AVX512BW-NEXT:    vmovdqa64 192(%rdi), %zmm4
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm5 = [8,13,18,23,28,0,0,3,8,13,18,23,28,0,0,3]
; AVX512BW-NEXT:    # zmm5 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vpermi2d %zmm4, %zmm3, %zmm5
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm6 = <0,5,10,15,20,25,30,u>
; AVX512BW-NEXT:    vpermi2d %zmm2, %zmm1, %zmm6
; AVX512BW-NEXT:    movw $8064, %ax # imm = 0x1F80
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqa32 %zmm5, %zmm6 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm5 = [0,1,2,3,4,5,6,7,8,9,10,11,12,17,22,27]
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm6, %zmm5
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm6 = [9,14,19,24,29,0,0,4,9,14,19,24,29,0,0,4]
; AVX512BW-NEXT:    # zmm6 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vpermi2d %zmm4, %zmm3, %zmm6
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm7 = <17,22,27,0,5,10,15,u>
; AVX512BW-NEXT:    vpermi2d %zmm1, %zmm2, %zmm7
; AVX512BW-NEXT:    vmovdqa32 %zmm6, %zmm7 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm6 = [0,1,2,3,4,5,6,7,8,9,10,11,12,18,23,28]
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm7, %zmm6
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm7 = <2,7,12,17,22,27,u,u>
; AVX512BW-NEXT:    vpermi2d %zmm2, %zmm1, %zmm7
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm8 = [10,15,20,25,30,0,0,5,10,15,20,25,30,0,0,5]
; AVX512BW-NEXT:    # zmm8 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vpermi2d %zmm4, %zmm3, %zmm8
; AVX512BW-NEXT:    movb $7, %al
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqa64 %zmm7, %zmm8 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm7 = [0,1,2,3,4,5,6,7,8,9,10,11,12,19,24,29]
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm8, %zmm7
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm8 = <3,8,13,18,23,28,u,u>
; AVX512BW-NEXT:    vpermi2d %zmm2, %zmm1, %zmm8
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm9 = [27,0,5,10,15,0,17,22,27,0,5,10,15,0,17,22]
; AVX512BW-NEXT:    # zmm9 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vpermi2d %zmm3, %zmm4, %zmm9
; AVX512BW-NEXT:    vmovdqa64 %zmm8, %zmm9 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm8 = [0,1,2,3,4,5,6,7,8,9,10,11,12,20,25,30]
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm9, %zmm8
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm9 = [12,17,22,27,0,0,2,7,12,17,22,27,0,0,2,7]
; AVX512BW-NEXT:    # zmm9 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vpermi2d %zmm4, %zmm3, %zmm9
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm3 = <4,9,14,19,24,29,u,u>
; AVX512BW-NEXT:    vpermi2d %zmm2, %zmm1, %zmm3
; AVX512BW-NEXT:    movb $56, %al
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqa64 %zmm9, %zmm3 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm1 = [0,1,2,3,4,5,6,7,8,9,10,11,16,21,26,31]
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm3, %zmm1
; AVX512BW-NEXT:    vmovdqa64 %zmm5, (%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm6, (%rdx)
; AVX512BW-NEXT:    vmovdqa64 %zmm7, (%rcx)
; AVX512BW-NEXT:    vmovdqa64 %zmm8, (%r8)
; AVX512BW-NEXT:    vmovdqa64 %zmm1, (%r9)
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <80 x i32>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <80 x i32> %wide.vec, <80 x i32> poison, <16 x i32> <i32 0, i32 5, i32 10, i32 15, i32 20, i32 25, i32 30, i32 35, i32 40, i32 45, i32 50, i32 55, i32 60, i32 65, i32 70, i32 75>
  %strided.vec1 = shufflevector <80 x i32> %wide.vec, <80 x i32> poison, <16 x i32> <i32 1, i32 6, i32 11, i32 16, i32 21, i32 26, i32 31, i32 36, i32 41, i32 46, i32 51, i32 56, i32 61, i32 66, i32 71, i32 76>
  %strided.vec2 = shufflevector <80 x i32> %wide.vec, <80 x i32> poison, <16 x i32> <i32 2, i32 7, i32 12, i32 17, i32 22, i32 27, i32 32, i32 37, i32 42, i32 47, i32 52, i32 57, i32 62, i32 67, i32 72, i32 77>
  %strided.vec3 = shufflevector <80 x i32> %wide.vec, <80 x i32> poison, <16 x i32> <i32 3, i32 8, i32 13, i32 18, i32 23, i32 28, i32 33, i32 38, i32 43, i32 48, i32 53, i32 58, i32 63, i32 68, i32 73, i32 78>
  %strided.vec4 = shufflevector <80 x i32> %wide.vec, <80 x i32> poison, <16 x i32> <i32 4, i32 9, i32 14, i32 19, i32 24, i32 29, i32 34, i32 39, i32 44, i32 49, i32 54, i32 59, i32 64, i32 69, i32 74, i32 79>
  store <16 x i32> %strided.vec0, ptr %out.vec0, align 64
  store <16 x i32> %strided.vec1, ptr %out.vec1, align 64
  store <16 x i32> %strided.vec2, ptr %out.vec2, align 64
  store <16 x i32> %strided.vec3, ptr %out.vec3, align 64
  store <16 x i32> %strided.vec4, ptr %out.vec4, align 64
  ret void
}

define void @load_i32_stride5_vf32(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4) nounwind {
; SSE-LABEL: load_i32_stride5_vf32:
; SSE:       # %bb.0:
; SSE-NEXT:    subq $904, %rsp # imm = 0x388
; SSE-NEXT:    movdqa 288(%rdi), %xmm5
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 272(%rdi), %xmm9
; SSE-NEXT:    movdqa 240(%rdi), %xmm12
; SSE-NEXT:    movdqa 256(%rdi), %xmm10
; SSE-NEXT:    movdqa 528(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 512(%rdi), %xmm3
; SSE-NEXT:    movdqa 480(%rdi), %xmm13
; SSE-NEXT:    movdqa 496(%rdi), %xmm6
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 208(%rdi), %xmm7
; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 192(%rdi), %xmm8
; SSE-NEXT:    movdqa 160(%rdi), %xmm11
; SSE-NEXT:    movdqa 176(%rdi), %xmm4
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[1,1,1,1]
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm11, %xmm1
; SSE-NEXT:    movdqa %xmm11, %xmm15
; SSE-NEXT:    movdqa %xmm11, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm8[2,2,2,2]
; SSE-NEXT:    movdqa %xmm8, %xmm11
; SSE-NEXT:    movdqa %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm7[2],xmm0[3],xmm7[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,1,1]
; SSE-NEXT:    movdqa %xmm13, %xmm1
; SSE-NEXT:    movdqa %xmm13, %xmm7
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[2,2,2,2]
; SSE-NEXT:    movdqa %xmm3, %xmm13
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm10[1,1,1,1]
; SSE-NEXT:    movdqa %xmm10, %xmm6
; SSE-NEXT:    movdqa %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm12, %xmm1
; SSE-NEXT:    movdqa %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm9[2,2,2,2]
; SSE-NEXT:    movdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm5[2],xmm0[3],xmm5[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 560(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 576(%rdi), %xmm5
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm5[1,1,1,1]
; SSE-NEXT:    movdqa %xmm2, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 608(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 592(%rdi), %xmm3
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[2,2,2,2]
; SSE-NEXT:    movdqa %xmm3, %xmm8
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa (%rdi), %xmm3
; SSE-NEXT:    movdqa 16(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa %xmm3, %xmm2
; SSE-NEXT:    movdqa %xmm3, %xmm10
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
; SSE-NEXT:    movdqa 32(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 48(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 320(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 336(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
; SSE-NEXT:    movdqa 368(%rdi), %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 352(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm3[2],xmm0[3],xmm3[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 80(%rdi), %xmm14
; SSE-NEXT:    movdqa 96(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa %xmm14, %xmm3
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; SSE-NEXT:    movdqa 128(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 112(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm3[0],xmm2[1]
; SSE-NEXT:    movapd %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 400(%rdi), %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 416(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; SSE-NEXT:    movdqa 448(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 432(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm3[0],xmm2[1]
; SSE-NEXT:    movapd %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm15[1,1,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movdqa 224(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm11[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm2[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm7, %xmm1
; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm7[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm7[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movdqa 544(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm13[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm2[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm12[1,1,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm6[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movdqa 304(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm9[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm13[1,1,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm5[2,3,2,3]
; SSE-NEXT:    movdqa %xmm5, %xmm9
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movdqa 624(%rdi), %xmm15
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm15[0,0,1,1]
; SSE-NEXT:    movdqa %xmm15, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm8[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = mem[2,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm10[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; SSE-NEXT:    movdqa 64(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm3[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm12[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movdqa 384(%rdi), %xmm6
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm6[0,0,1,1]
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa (%rsp), %xmm8 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm8[2,3,2,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm14[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; SSE-NEXT:    movdqa 144(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    # xmm10 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm2[0],xmm10[1],xmm2[1]
; SSE-NEXT:    movsd {{.*#+}} xmm10 = xmm3[0],xmm10[1]
; SSE-NEXT:    movapd %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm11[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movdqa 464(%rdi), %xmm5
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm5[0,0,1,1]
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    # xmm10 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm3[0],xmm10[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm10 = xmm2[0],xmm10[1]
; SSE-NEXT:    movapd %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm7[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm1 = xmm1[2],xmm3[2],xmm1[3],xmm3[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = xmm3[0],mem[0],xmm3[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm1[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm13, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm9[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm2[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm12, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm7[0,0,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm2[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm0, %xmm2
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm11[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm9, %xmm2
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm11[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm12[0],xmm0[1],xmm12[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm6, %xmm2
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm15[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm13[0],xmm0[1],xmm13[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[3,3,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm4[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm8[2,2,3,3]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm14[3,3,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm14 = xmm14[2],xmm1[2],xmm14[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm8[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm14[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm5[2,2,3,3]
; SSE-NEXT:    movdqa %xmm4, %xmm5
; SSE-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm10[0],xmm3[1],xmm10[1]
; SSE-NEXT:    movsd {{.*#+}} xmm5 = xmm3[0],xmm5[1]
; SSE-NEXT:    movapd %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,2,3,3]
; SSE-NEXT:    movdqa %xmm8, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm2[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm12[2,2,3,3]
; SSE-NEXT:    movdqa %xmm11, %xmm14
; SSE-NEXT:    punpckldq {{.*#+}} xmm14 = xmm14[0],xmm0[0],xmm14[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm9[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm14 = xmm0[0],xmm14[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm13[2,2,3,3]
; SSE-NEXT:    movdqa %xmm15, %xmm13
; SSE-NEXT:    punpckldq {{.*#+}} xmm13 = xmm13[0],xmm0[0],xmm13[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm13 = xmm0[0],xmm13[1]
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa %xmm7, %xmm12
; SSE-NEXT:    punpckldq {{.*#+}} xmm12 = xmm12[0],xmm0[0],xmm12[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
; SSE-NEXT:    movsd {{.*#+}} xmm12 = xmm0[0],xmm12[1]
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm11 = xmm11[0],xmm0[0],xmm11[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm11 = xmm0[0],xmm11[1]
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm9 = xmm9[0],xmm0[0],xmm9[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
; SSE-NEXT:    movsd {{.*#+}} xmm9 = xmm0[0],xmm9[1]
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm0[0],xmm7[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm15[0],xmm0[1],xmm15[1]
; SSE-NEXT:    movsd {{.*#+}} xmm7 = xmm0[0],xmm7[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm10[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    # xmm10 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    # xmm10 = xmm10[2],mem[2],xmm10[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm10 = xmm6[0],xmm10[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm5[1,1,1,1]
; SSE-NEXT:    movdqa (%rsp), %xmm6 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm8 = xmm8[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
; SSE-NEXT:    # xmm8 = xmm8[2],mem[2],xmm8[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm8 = xmm6[0],xmm8[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
; SSE-NEXT:    # xmm6 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
; SSE-NEXT:    # xmm6 = xmm6[2],mem[2],xmm6[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm6 = xmm3[0],xmm6[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Folded Reload
; SSE-NEXT:    # xmm5 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Folded Reload
; SSE-NEXT:    # xmm5 = xmm5[2],mem[2],xmm5[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm5 = xmm1[0],xmm5[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = xmm4[2],mem[2],xmm4[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm1[0],xmm4[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = xmm3[2],mem[2],xmm3[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm1[0],xmm3[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = xmm2[2],mem[2],xmm2[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm1[0],xmm2[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm15[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[2],mem[2],xmm0[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm15[0],xmm0[1]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 80(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 16(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 64(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, (%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 112(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 48(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 96(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 32(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 80(%rdx)
; 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, 64(%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, 112(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 48(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 96(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 32(%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, 48(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 32(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 80(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 64(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 112(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 96(%rcx)
; SSE-NEXT:    movapd %xmm7, 112(%r8)
; SSE-NEXT:    movapd %xmm9, 96(%r8)
; SSE-NEXT:    movapd %xmm11, 80(%r8)
; SSE-NEXT:    movapd %xmm12, 64(%r8)
; SSE-NEXT:    movapd %xmm13, 48(%r8)
; SSE-NEXT:    movapd %xmm14, 32(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, 16(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movaps %xmm1, (%r8)
; SSE-NEXT:    movapd %xmm0, 112(%r9)
; SSE-NEXT:    movapd %xmm2, 96(%r9)
; SSE-NEXT:    movapd %xmm3, 80(%r9)
; SSE-NEXT:    movapd %xmm4, 64(%r9)
; SSE-NEXT:    movapd %xmm5, 48(%r9)
; SSE-NEXT:    movapd %xmm6, 32(%r9)
; SSE-NEXT:    movapd %xmm8, 16(%r9)
; SSE-NEXT:    movapd %xmm10, (%r9)
; SSE-NEXT:    addq $904, %rsp # imm = 0x388
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i32_stride5_vf32:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    subq $920, %rsp # imm = 0x398
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 544(%rdi), %ymm7
; AVX1-ONLY-NEXT:    vmovaps 576(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vmovaps 512(%rdi), %ymm5
; AVX1-ONLY-NEXT:    vmovups %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 480(%rdi), %ymm6
; AVX1-ONLY-NEXT:    vmovups %ymm6, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 224(%rdi), %ymm8
; AVX1-ONLY-NEXT:    vmovaps 256(%rdi), %ymm9
; AVX1-ONLY-NEXT:    vmovaps 192(%rdi), %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 160(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1],ymm0[2,3],ymm2[4,5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vmovaps %ymm2, %ymm12
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 288(%rdi), %ymm0, %ymm1
; AVX1-ONLY-NEXT:    vmovaps 256(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm11
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm11[0,1,2,3,4,5],ymm9[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm8[4],ymm2[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm2[4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3],ymm0[4,5,6],ymm1[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm6[0,1],ymm5[2,3],ymm6[4,5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX1-ONLY-NEXT:    vmovaps 576(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm13
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm13[0,1,2,3,4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm7[4],ymm1[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 608(%rdi), %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3],ymm1[4,5,6],ymm0[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %ymm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm3[2,3],ymm1[4,5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vmovaps %ymm1, %ymm15
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX1-ONLY-NEXT:    vmovaps 64(%rdi), %ymm14
; AVX1-ONLY-NEXT:    vmovaps 96(%rdi), %ymm5
; AVX1-ONLY-NEXT:    vmovaps 96(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm10
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm10[0,1,2,3,4,5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm14[4],ymm1[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 128(%rdi), %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3],ymm1[4,5,6],ymm0[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 352(%rdi), %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 320(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX1-ONLY-NEXT:    vmovaps 384(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vmovaps 416(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vmovaps 416(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4],ymm2[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm0[0,1,2,3],ymm2[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 448(%rdi), %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2],ymm0[3],ymm2[4,5,6],ymm0[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm9[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,1],ymm8[1,3],ymm0[6,5],ymm8[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm8[0,1],ymm9[2,3],ymm8[4,5],ymm9[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm2[3,0],ymm0[6,4],ymm2[7,4]
; AVX1-ONLY-NEXT:    vmovaps 160(%rdi), %xmm6
; AVX1-ONLY-NEXT:    vmovaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 192(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm6[0,1],xmm2[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],mem[2],xmm2[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm2 = xmm2[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 304(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm4[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,1],ymm7[1,3],ymm0[6,5],ymm7[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm7[0,1],ymm4[2,3],ymm7[4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm2[3,0],ymm0[6,4],ymm2[7,4]
; AVX1-ONLY-NEXT:    vmovaps 480(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vmovaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 512(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm4[0,1],xmm2[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],mem[2],xmm2[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm2 = xmm2[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 624(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm5[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,1],ymm14[1,3],ymm0[6,5],ymm14[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm14[0,1],ymm5[2,3],ymm14[4,5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm2[3,0],ymm0[6,4],ymm2[7,4]
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %xmm4
; AVX1-ONLY-NEXT:    vmovaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm4[0,1],xmm2[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],mem[2],xmm2[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm2 = xmm2[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 144(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm1[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,1],ymm3[1,3],ymm0[6,5],ymm3[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1],ymm1[2,3],ymm3[4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm2[3,0],ymm0[6,4],ymm2[7,4]
; AVX1-ONLY-NEXT:    vmovaps 320(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 352(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm1[0,1],xmm2[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],mem[2],xmm2[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm2 = xmm2[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 464(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm11[3,0],ymm8[2,0],ymm11[7,4],ymm8[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm8[2,1],ymm0[6,4],ymm8[6,5]
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm12, %ymm2 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm2 = ymm12[0,1,2,3],mem[4,5],ymm12[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],mem[2],xmm2[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm2 = xmm2[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 288(%rdi), %ymm5
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm5[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm2 = ymm5[1,0],ymm1[0,0],ymm5[5,4],ymm1[4,4]
; AVX1-ONLY-NEXT:    vmovaps %ymm1, %ymm11
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm2 = ymm2[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm13[3,0],ymm7[2,0],ymm13[7,4],ymm7[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm7[2,1],ymm0[6,4],ymm7[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm2 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm2 = mem[0,1,2,3],ymm1[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],mem[2],xmm2[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm2 = xmm2[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 608(%rdi), %ymm6
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm12 = ymm6[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm4 = ymm6[1,0],ymm12[0,0],ymm6[5,4],ymm12[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm4 = ymm4[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm10[3,0],ymm14[2,0],ymm10[7,4],ymm14[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm14[2,1],ymm0[6,4],ymm14[6,5]
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm15, %ymm4 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm4 = ymm15[0,1,2,3],mem[4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm4, %xmm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm4 = xmm4[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm4[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 128(%rdi), %ymm9
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm13 = ymm9[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm4 = ymm9[1,0],ymm13[0,0],ymm9[5,4],ymm13[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm4 = ymm4[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm3[2,0],ymm0[7,4],ymm3[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm3[2,1],ymm0[6,4],ymm3[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm4 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm4 = ymm1[0,1,2,3],mem[4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm4, %xmm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm4 = xmm4[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm4[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 448(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm10 = ymm4[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm4[1,0],ymm10[0,0],ymm4[5,4],ymm10[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm1[0,0],ymm8[3,0],ymm1[4,4],ymm8[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm8[2,2],ymm15[6,4],ymm8[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm5[2,0],ymm11[1,0],ymm5[6,4],ymm11[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm2[0,0],ymm7[3,0],ymm2[4,4],ymm7[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm7[2,2],ymm15[6,4],ymm7[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm6[2,0],ymm12[1,0],ymm6[6,4],ymm12[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm11 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm11[0,0],ymm14[3,0],ymm11[4,4],ymm14[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm14[2,2],ymm15[6,4],ymm14[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm9[2,0],ymm13[1,0],ymm9[6,4],ymm13[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm1[0,0],ymm3[3,0],ymm1[4,4],ymm3[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm3[2,2],ymm15[6,4],ymm3[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm4[2,0],ymm10[1,0],ymm4[6,4],ymm10[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm8, %ymm8 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps $32, {{[-0-9]+}}(%r{{[sb]}}p), %ymm8, %ymm8 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm8 = ymm8[0,1,2,3,4],mem[5],ymm8[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm15 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm15, %ymm15 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm15 = mem[0,1,2,3],ymm15[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm15, %xmm15
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm15 = xmm15[0],mem[1],xmm15[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm8 = ymm15[0,1,2],ymm8[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps $127, {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm5 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm5 = mem[0,1,2,3,4,5,6],ymm5[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm5 = ymm8[0,1,2,3,4,5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4],ymm2[5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm8, %ymm8 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm8 = ymm8[0,1,2,3],mem[4,5],ymm8[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm8, %xmm8
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm8 = xmm8[0],mem[1],xmm8[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm7 = ymm8[0,1,2],ymm7[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm12[0,1,2,3,4,5,6],ymm6[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm7[0,1,2,3,4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm3 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4],ymm1[5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm7 = mem[0,1,2,3],ymm7[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm7, %xmm7
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm7 = xmm7[0],mem[1],xmm7[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm7[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm10[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm3[0,1,2,3,4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm14, %ymm3 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4],ymm11[5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm4 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm4 = mem[0,1,2,3],ymm4[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm4, %xmm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0],mem[1],xmm4[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm13[0,1,2,3,4,5,6],ymm9[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 64(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, (%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 96(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 32(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 64(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, (%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 96(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 32(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 64(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, (%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 96(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 32(%rcx)
; AVX1-ONLY-NEXT:    vmovaps %ymm0, 64(%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, 96(%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm0, 32(%r8)
; AVX1-ONLY-NEXT:    vmovaps %ymm3, (%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm1, 64(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm2, 96(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm5, 32(%r9)
; AVX1-ONLY-NEXT:    addq $920, %rsp # imm = 0x398
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i32_stride5_vf32:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    subq $1032, %rsp # imm = 0x408
; AVX2-ONLY-NEXT:    vmovdqa 384(%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 416(%rdi), %ymm13
; AVX2-ONLY-NEXT:    vmovdqu %ymm13, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 352(%rdi), %ymm6
; AVX2-ONLY-NEXT:    vmovdqu %ymm6, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 320(%rdi), %ymm7
; AVX2-ONLY-NEXT:    vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 544(%rdi), %ymm10
; AVX2-ONLY-NEXT:    vmovdqu %ymm10, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 576(%rdi), %ymm11
; AVX2-ONLY-NEXT:    vmovdqa 512(%rdi), %ymm15
; AVX2-ONLY-NEXT:    vmovdqa 480(%rdi), %ymm8
; AVX2-ONLY-NEXT:    vmovdqu %ymm8, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 224(%rdi), %ymm9
; AVX2-ONLY-NEXT:    vmovdqu %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 256(%rdi), %ymm12
; AVX2-ONLY-NEXT:    vmovdqa 192(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vmovdqa 160(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm5 = [0,5,2,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm4[2,3],ymm1[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm5, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 288(%rdi), %ymm1, %ymm2
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm3 = ymm12[0,1,0,3]
; AVX2-ONLY-NEXT:    vmovdqu %ymm12, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm9[4],ymm3[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3],ymm1[4,5,6],ymm2[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm8[0,1],ymm15[2,3],ymm8[4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm15, %ymm9
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm5, %ymm1
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm11[0,1,0,3]
; AVX2-ONLY-NEXT:    vmovdqu %ymm11, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm10[4],ymm2[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 608(%rdi), %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm7[0,1],ymm6[2,3],ymm7[4,5],ymm6[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm5, %ymm1
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm13[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm0[4],ymm2[5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, %ymm10
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, (%rsp) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 448(%rdi), %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm7
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm7[0,1],ymm3[2,3],ymm7[4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm3, %ymm13
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm5, %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqa 96(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm1 = ymm2[0,1,0,3]
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, %ymm14
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4],ymm1[5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm3, %ymm15
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 128(%rdi), %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3],ymm1[4,5,6],ymm0[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <1,6,3,u>
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm8[0,1],ymm4[2,3],ymm8[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm5[0,1],ymm12[2,3],ymm5[4,5],ymm12[6,7]
; AVX2-ONLY-NEXT:    vbroadcasti128 {{.*#+}} ymm3 = [5,2,7,0,5,2,7,0]
; AVX2-ONLY-NEXT:    # ymm3 = mem[0,1,0,1]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm3, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 304(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5,6],ymm2[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm4[0,1],ymm9[2,3],ymm4[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm6[0,1],ymm11[2,3],ymm6[4,5],ymm11[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm3, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 624(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5,6],ymm2[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm7[0,1],ymm13[2,3],ymm7[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm15[0,1],ymm14[2,3],ymm15[4,5],ymm14[6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm15, %ymm14
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm3, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 144(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5,6],ymm2[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm11 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $243, {{[-0-9]+}}(%r{{[sb]}}p), %ymm11, %ymm1 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm1 = mem[0,1],ymm11[2,3],mem[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendd $204, {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm1 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm1 = ymm10[0,1],mem[2,3],ymm10[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm3, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 464(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm13 = <2,7,4,u>
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm12 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm8[0,1,2,3],ymm12[4,5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm13, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 256(%rdi), %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm5[0,1,2,3,4,5,6],ymm2[7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm5, %ymm7
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 288(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm0 = [0,5,0,5,0,5,0,5]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm4[0,1,2,3],ymm9[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm13, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 576(%rdi), %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm6[0,1,2,3,4,5,6],ymm2[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm2 = ymm2[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 608(%rdi), %ymm9
; AVX2-ONLY-NEXT:    vpermd %ymm9, %ymm0, %ymm4
; AVX2-ONLY-NEXT:    vmovdqu %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm15 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm15[0,1,2,3],ymm5[4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm13, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 96(%rdi), %ymm0, %ymm4
; AVX2-ONLY-NEXT:    vmovdqa %ymm14, %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm14, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm14[0,1,2,3,4,5,6],ymm4[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm4 = ymm4[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm4[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 128(%rdi), %ymm14
; AVX2-ONLY-NEXT:    vpermd %ymm14, %ymm0, %ymm10
; AVX2-ONLY-NEXT:    vmovdqu %ymm14, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm10[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm3[0,1,2,3],ymm11[4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm13, %ymm4
; AVX2-ONLY-NEXT:    vinserti128 $1, 416(%rdi), %ymm0, %ymm8
; AVX2-ONLY-NEXT:    vmovdqu (%rsp), %ymm1 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm1[0,1,2,3,4,5,6],ymm8[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm8 = ymm8[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm4[0,1,2],ymm8[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 448(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vmovdqu %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm4, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm8[0,1,2,3,4,5],ymm0[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm13 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm7[0,1,2,3],ymm13[4,5],ymm7[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vpalignr $12, {{[-0-9]+}}(%r{{[sb]}}p), %ymm12, %ymm8 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm8 = mem[12,13,14,15],ymm12[0,1,2,3,4,5,6,7,8,9,10,11],mem[28,29,30,31],ymm12[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm8 = ymm8[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm8[0,1,2],ymm0[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm8 = [1,6,1,6,1,6,1,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm11 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpermd %ymm11, %ymm8, %ymm10
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm10[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm12 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm6[0,1,2,3],ymm12[4,5],ymm6[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm10 = ymm6[12,13,14,15],ymm4[0,1,2,3,4,5,6,7,8,9,10,11],ymm6[28,29,30,31],ymm4[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm10 = ymm10[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm10[0,1,2],ymm0[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm9, %ymm8, %ymm10
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm10[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm0 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm0 = ymm2[0,1,2,3],mem[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm10 = ymm15[12,13,14,15],ymm5[0,1,2,3,4,5,6,7,8,9,10,11],ymm15[28,29,30,31],ymm5[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm10 = ymm10[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm10[0,1,2],ymm0[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm14, %ymm8, %ymm10
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm10[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm5[4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm15 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm10 = ymm3[12,13,14,15],ymm15[0,1,2,3,4,5,6,7,8,9,10,11],ymm3[28,29,30,31],ymm15[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vmovdqa %ymm3, %ymm14
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm10 = ymm10[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm10[0,1,2],ymm0[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm8, %ymm8
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm0 = ymm0[0,1,2,3],mem[4,5],ymm0[6,7]
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm7 = ymm7[0,1],ymm13[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2,3,4],ymm13[5],ymm7[6,7]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm8 = <4,1,6,u>
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm8, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm7[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm7 = [2,7,2,7,2,7,2,7]
; AVX2-ONLY-NEXT:    vpermd %ymm11, %ymm7, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm6[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vperm2i128 $2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm12, %ymm9 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm9 = mem[0,1],ymm12[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm9 = ymm9[0,1,2,3,4],ymm12[5],ymm9[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm8, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm9[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm2 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm3[0,1,2,3,4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm15[0,1,2,3],ymm14[4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vperm2i128 $2, (%rsp), %ymm5, %ymm9 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm9 = mem[0,1],ymm5[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm9 = ymm9[0,1,2,3,4],ymm5[5],ymm9[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm8, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm9[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm7, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm3[0,1,2,3,4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vperm2i128 $2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = mem[0,1],ymm4[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3,4],ymm4[5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm5 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm5 = ymm4[0,1,2,3],mem[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm5, %ymm8, %ymm5
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm5[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm4 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 64(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, (%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 96(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 32(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 64(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, (%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 96(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 32(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 64(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, (%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 96(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 32(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 64(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, (%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 96(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm4, 32(%r8)
; AVX2-ONLY-NEXT:    vmovdqa %ymm3, (%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm1, 64(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, 96(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, 32(%r9)
; AVX2-ONLY-NEXT:    addq $1032, %rsp # imm = 0x408
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-LABEL: load_i32_stride5_vf32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovdqa64 576(%rdi), %zmm2
; AVX512F-NEXT:    vmovdqa64 384(%rdi), %zmm3
; AVX512F-NEXT:    vmovdqa64 320(%rdi), %zmm0
; AVX512F-NEXT:    vmovdqa64 512(%rdi), %zmm5
; AVX512F-NEXT:    vmovdqa64 448(%rdi), %zmm4
; AVX512F-NEXT:    vmovdqa64 256(%rdi), %zmm7
; AVX512F-NEXT:    vmovdqa64 (%rdi), %zmm1
; AVX512F-NEXT:    vmovdqa64 64(%rdi), %zmm10
; AVX512F-NEXT:    vmovdqa64 128(%rdi), %zmm9
; AVX512F-NEXT:    vmovdqa64 192(%rdi), %zmm11
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm12 = [8,13,18,23,28,0,0,3,8,13,18,23,28,0,0,3]
; AVX512F-NEXT:    # zmm12 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vmovdqa64 %zmm9, %zmm13
; AVX512F-NEXT:    vpermt2d %zmm11, %zmm12, %zmm13
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm6 = <0,5,10,15,20,25,30,u>
; AVX512F-NEXT:    vmovdqa64 %zmm1, %zmm8
; AVX512F-NEXT:    vpermt2d %zmm10, %zmm6, %zmm8
; AVX512F-NEXT:    movw $8064, %ax # imm = 0x1F80
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqa32 %zmm13, %zmm8 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm13 = [0,1,2,3,4,5,6,7,8,9,10,11,12,17,22,27]
; AVX512F-NEXT:    vpermt2d %zmm7, %zmm13, %zmm8
; AVX512F-NEXT:    vpermi2d %zmm5, %zmm4, %zmm12
; AVX512F-NEXT:    vpermi2d %zmm3, %zmm0, %zmm6
; AVX512F-NEXT:    vmovdqa32 %zmm12, %zmm6 {%k1}
; AVX512F-NEXT:    vpermt2d %zmm2, %zmm13, %zmm6
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm14 = [9,14,19,24,29,0,0,4,9,14,19,24,29,0,0,4]
; AVX512F-NEXT:    # zmm14 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vmovdqa64 %zmm9, %zmm15
; AVX512F-NEXT:    vpermt2d %zmm11, %zmm14, %zmm15
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm12 = <17,22,27,0,5,10,15,u>
; AVX512F-NEXT:    vmovdqa64 %zmm10, %zmm13
; AVX512F-NEXT:    vpermt2d %zmm1, %zmm12, %zmm13
; AVX512F-NEXT:    vmovdqa32 %zmm15, %zmm13 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm15 = [0,1,2,3,4,5,6,7,8,9,10,11,12,18,23,28]
; AVX512F-NEXT:    vpermt2d %zmm7, %zmm15, %zmm13
; AVX512F-NEXT:    vpermi2d %zmm5, %zmm4, %zmm14
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm3, %zmm12
; AVX512F-NEXT:    vmovdqa32 %zmm14, %zmm12 {%k1}
; AVX512F-NEXT:    vpermt2d %zmm2, %zmm15, %zmm12
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm16 = <2,7,12,17,22,27,u,u>
; AVX512F-NEXT:    vmovdqa64 %zmm1, %zmm17
; AVX512F-NEXT:    vpermt2d %zmm10, %zmm16, %zmm17
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm15 = [10,15,20,25,30,0,0,5,10,15,20,25,30,0,0,5]
; AVX512F-NEXT:    # zmm15 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vmovdqa64 %zmm9, %zmm14
; AVX512F-NEXT:    vpermt2d %zmm11, %zmm15, %zmm14
; AVX512F-NEXT:    movb $7, %al
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqa64 %zmm17, %zmm14 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm17 = [0,1,2,3,4,5,6,7,8,9,10,11,12,19,24,29]
; AVX512F-NEXT:    vpermt2d %zmm7, %zmm17, %zmm14
; AVX512F-NEXT:    vpermi2d %zmm5, %zmm4, %zmm15
; AVX512F-NEXT:    vpermi2d %zmm3, %zmm0, %zmm16
; AVX512F-NEXT:    vmovdqa64 %zmm16, %zmm15 {%k1}
; AVX512F-NEXT:    vpermt2d %zmm2, %zmm17, %zmm15
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm16 = <3,8,13,18,23,28,u,u>
; AVX512F-NEXT:    vmovdqa64 %zmm1, %zmm17
; AVX512F-NEXT:    vpermt2d %zmm10, %zmm16, %zmm17
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm18 = [27,0,5,10,15,0,17,22,27,0,5,10,15,0,17,22]
; AVX512F-NEXT:    # zmm18 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vmovdqa64 %zmm11, %zmm19
; AVX512F-NEXT:    vpermt2d %zmm9, %zmm18, %zmm19
; AVX512F-NEXT:    vmovdqa64 %zmm17, %zmm19 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm17 = [0,1,2,3,4,5,6,7,8,9,10,11,12,20,25,30]
; AVX512F-NEXT:    vpermt2d %zmm7, %zmm17, %zmm19
; AVX512F-NEXT:    vpermi2d %zmm4, %zmm5, %zmm18
; AVX512F-NEXT:    vpermi2d %zmm3, %zmm0, %zmm16
; AVX512F-NEXT:    vmovdqa64 %zmm16, %zmm18 {%k1}
; AVX512F-NEXT:    vpermt2d %zmm2, %zmm17, %zmm18
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm16 = [12,17,22,27,0,0,2,7,12,17,22,27,0,0,2,7]
; AVX512F-NEXT:    # zmm16 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpermt2d %zmm11, %zmm16, %zmm9
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm11 = <4,9,14,19,24,29,u,u>
; AVX512F-NEXT:    vpermt2d %zmm10, %zmm11, %zmm1
; AVX512F-NEXT:    movb $56, %al
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqa64 %zmm9, %zmm1 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm9 = [0,1,2,3,4,5,6,7,8,9,10,11,16,21,26,31]
; AVX512F-NEXT:    vpermt2d %zmm7, %zmm9, %zmm1
; AVX512F-NEXT:    vpermt2d %zmm5, %zmm16, %zmm4
; AVX512F-NEXT:    vpermt2d %zmm3, %zmm11, %zmm0
; AVX512F-NEXT:    vmovdqa64 %zmm4, %zmm0 {%k1}
; AVX512F-NEXT:    vpermt2d %zmm2, %zmm9, %zmm0
; AVX512F-NEXT:    vmovdqa64 %zmm6, 64(%rsi)
; AVX512F-NEXT:    vmovdqa64 %zmm8, (%rsi)
; AVX512F-NEXT:    vmovdqa64 %zmm12, 64(%rdx)
; AVX512F-NEXT:    vmovdqa64 %zmm13, (%rdx)
; AVX512F-NEXT:    vmovdqa64 %zmm15, 64(%rcx)
; AVX512F-NEXT:    vmovdqa64 %zmm14, (%rcx)
; AVX512F-NEXT:    vmovdqa64 %zmm18, 64(%r8)
; AVX512F-NEXT:    vmovdqa64 %zmm19, (%r8)
; AVX512F-NEXT:    vmovdqa64 %zmm0, 64(%r9)
; AVX512F-NEXT:    vmovdqa64 %zmm1, (%r9)
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i32_stride5_vf32:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovdqa64 576(%rdi), %zmm2
; AVX512BW-NEXT:    vmovdqa64 384(%rdi), %zmm3
; AVX512BW-NEXT:    vmovdqa64 320(%rdi), %zmm0
; AVX512BW-NEXT:    vmovdqa64 512(%rdi), %zmm5
; AVX512BW-NEXT:    vmovdqa64 448(%rdi), %zmm4
; AVX512BW-NEXT:    vmovdqa64 256(%rdi), %zmm7
; AVX512BW-NEXT:    vmovdqa64 (%rdi), %zmm1
; AVX512BW-NEXT:    vmovdqa64 64(%rdi), %zmm10
; AVX512BW-NEXT:    vmovdqa64 128(%rdi), %zmm9
; AVX512BW-NEXT:    vmovdqa64 192(%rdi), %zmm11
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm12 = [8,13,18,23,28,0,0,3,8,13,18,23,28,0,0,3]
; AVX512BW-NEXT:    # zmm12 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vmovdqa64 %zmm9, %zmm13
; AVX512BW-NEXT:    vpermt2d %zmm11, %zmm12, %zmm13
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm6 = <0,5,10,15,20,25,30,u>
; AVX512BW-NEXT:    vmovdqa64 %zmm1, %zmm8
; AVX512BW-NEXT:    vpermt2d %zmm10, %zmm6, %zmm8
; AVX512BW-NEXT:    movw $8064, %ax # imm = 0x1F80
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqa32 %zmm13, %zmm8 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm13 = [0,1,2,3,4,5,6,7,8,9,10,11,12,17,22,27]
; AVX512BW-NEXT:    vpermt2d %zmm7, %zmm13, %zmm8
; AVX512BW-NEXT:    vpermi2d %zmm5, %zmm4, %zmm12
; AVX512BW-NEXT:    vpermi2d %zmm3, %zmm0, %zmm6
; AVX512BW-NEXT:    vmovdqa32 %zmm12, %zmm6 {%k1}
; AVX512BW-NEXT:    vpermt2d %zmm2, %zmm13, %zmm6
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm14 = [9,14,19,24,29,0,0,4,9,14,19,24,29,0,0,4]
; AVX512BW-NEXT:    # zmm14 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vmovdqa64 %zmm9, %zmm15
; AVX512BW-NEXT:    vpermt2d %zmm11, %zmm14, %zmm15
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm12 = <17,22,27,0,5,10,15,u>
; AVX512BW-NEXT:    vmovdqa64 %zmm10, %zmm13
; AVX512BW-NEXT:    vpermt2d %zmm1, %zmm12, %zmm13
; AVX512BW-NEXT:    vmovdqa32 %zmm15, %zmm13 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm15 = [0,1,2,3,4,5,6,7,8,9,10,11,12,18,23,28]
; AVX512BW-NEXT:    vpermt2d %zmm7, %zmm15, %zmm13
; AVX512BW-NEXT:    vpermi2d %zmm5, %zmm4, %zmm14
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm3, %zmm12
; AVX512BW-NEXT:    vmovdqa32 %zmm14, %zmm12 {%k1}
; AVX512BW-NEXT:    vpermt2d %zmm2, %zmm15, %zmm12
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} ymm16 = <2,7,12,17,22,27,u,u>
; AVX512BW-NEXT:    vmovdqa64 %zmm1, %zmm17
; AVX512BW-NEXT:    vpermt2d %zmm10, %zmm16, %zmm17
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm15 = [10,15,20,25,30,0,0,5,10,15,20,25,30,0,0,5]
; AVX512BW-NEXT:    # zmm15 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vmovdqa64 %zmm9, %zmm14
; AVX512BW-NEXT:    vpermt2d %zmm11, %zmm15, %zmm14
; AVX512BW-NEXT:    movb $7, %al
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqa64 %zmm17, %zmm14 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm17 = [0,1,2,3,4,5,6,7,8,9,10,11,12,19,24,29]
; AVX512BW-NEXT:    vpermt2d %zmm7, %zmm17, %zmm14
; AVX512BW-NEXT:    vpermi2d %zmm5, %zmm4, %zmm15
; AVX512BW-NEXT:    vpermi2d %zmm3, %zmm0, %zmm16
; AVX512BW-NEXT:    vmovdqa64 %zmm16, %zmm15 {%k1}
; AVX512BW-NEXT:    vpermt2d %zmm2, %zmm17, %zmm15
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} ymm16 = <3,8,13,18,23,28,u,u>
; AVX512BW-NEXT:    vmovdqa64 %zmm1, %zmm17
; AVX512BW-NEXT:    vpermt2d %zmm10, %zmm16, %zmm17
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm18 = [27,0,5,10,15,0,17,22,27,0,5,10,15,0,17,22]
; AVX512BW-NEXT:    # zmm18 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vmovdqa64 %zmm11, %zmm19
; AVX512BW-NEXT:    vpermt2d %zmm9, %zmm18, %zmm19
; AVX512BW-NEXT:    vmovdqa64 %zmm17, %zmm19 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm17 = [0,1,2,3,4,5,6,7,8,9,10,11,12,20,25,30]
; AVX512BW-NEXT:    vpermt2d %zmm7, %zmm17, %zmm19
; AVX512BW-NEXT:    vpermi2d %zmm4, %zmm5, %zmm18
; AVX512BW-NEXT:    vpermi2d %zmm3, %zmm0, %zmm16
; AVX512BW-NEXT:    vmovdqa64 %zmm16, %zmm18 {%k1}
; AVX512BW-NEXT:    vpermt2d %zmm2, %zmm17, %zmm18
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm16 = [12,17,22,27,0,0,2,7,12,17,22,27,0,0,2,7]
; AVX512BW-NEXT:    # zmm16 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vpermt2d %zmm11, %zmm16, %zmm9
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm11 = <4,9,14,19,24,29,u,u>
; AVX512BW-NEXT:    vpermt2d %zmm10, %zmm11, %zmm1
; AVX512BW-NEXT:    movb $56, %al
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqa64 %zmm9, %zmm1 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm9 = [0,1,2,3,4,5,6,7,8,9,10,11,16,21,26,31]
; AVX512BW-NEXT:    vpermt2d %zmm7, %zmm9, %zmm1
; AVX512BW-NEXT:    vpermt2d %zmm5, %zmm16, %zmm4
; AVX512BW-NEXT:    vpermt2d %zmm3, %zmm11, %zmm0
; AVX512BW-NEXT:    vmovdqa64 %zmm4, %zmm0 {%k1}
; AVX512BW-NEXT:    vpermt2d %zmm2, %zmm9, %zmm0
; AVX512BW-NEXT:    vmovdqa64 %zmm6, 64(%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm8, (%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm12, 64(%rdx)
; AVX512BW-NEXT:    vmovdqa64 %zmm13, (%rdx)
; AVX512BW-NEXT:    vmovdqa64 %zmm15, 64(%rcx)
; AVX512BW-NEXT:    vmovdqa64 %zmm14, (%rcx)
; AVX512BW-NEXT:    vmovdqa64 %zmm18, 64(%r8)
; AVX512BW-NEXT:    vmovdqa64 %zmm19, (%r8)
; AVX512BW-NEXT:    vmovdqa64 %zmm0, 64(%r9)
; AVX512BW-NEXT:    vmovdqa64 %zmm1, (%r9)
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <160 x i32>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <160 x i32> %wide.vec, <160 x i32> poison, <32 x i32> <i32 0, i32 5, i32 10, i32 15, i32 20, i32 25, i32 30, i32 35, i32 40, i32 45, i32 50, i32 55, i32 60, i32 65, i32 70, i32 75, i32 80, i32 85, i32 90, i32 95, i32 100, i32 105, i32 110, i32 115, i32 120, i32 125, i32 130, i32 135, i32 140, i32 145, i32 150, i32 155>
  %strided.vec1 = shufflevector <160 x i32> %wide.vec, <160 x i32> poison, <32 x i32> <i32 1, i32 6, i32 11, i32 16, i32 21, i32 26, i32 31, i32 36, i32 41, i32 46, i32 51, i32 56, i32 61, i32 66, i32 71, i32 76, i32 81, i32 86, i32 91, i32 96, i32 101, i32 106, i32 111, i32 116, i32 121, i32 126, i32 131, i32 136, i32 141, i32 146, i32 151, i32 156>
  %strided.vec2 = shufflevector <160 x i32> %wide.vec, <160 x i32> poison, <32 x i32> <i32 2, i32 7, i32 12, i32 17, i32 22, i32 27, i32 32, i32 37, i32 42, i32 47, i32 52, i32 57, i32 62, i32 67, i32 72, i32 77, i32 82, i32 87, i32 92, i32 97, i32 102, i32 107, i32 112, i32 117, i32 122, i32 127, i32 132, i32 137, i32 142, i32 147, i32 152, i32 157>
  %strided.vec3 = shufflevector <160 x i32> %wide.vec, <160 x i32> poison, <32 x i32> <i32 3, i32 8, i32 13, i32 18, i32 23, i32 28, i32 33, i32 38, i32 43, i32 48, i32 53, i32 58, i32 63, i32 68, i32 73, i32 78, i32 83, i32 88, i32 93, i32 98, i32 103, i32 108, i32 113, i32 118, i32 123, i32 128, i32 133, i32 138, i32 143, i32 148, i32 153, i32 158>
  %strided.vec4 = shufflevector <160 x i32> %wide.vec, <160 x i32> poison, <32 x i32> <i32 4, i32 9, i32 14, i32 19, i32 24, i32 29, i32 34, i32 39, i32 44, i32 49, i32 54, i32 59, i32 64, i32 69, i32 74, i32 79, i32 84, i32 89, i32 94, i32 99, i32 104, i32 109, i32 114, i32 119, i32 124, i32 129, i32 134, i32 139, i32 144, i32 149, i32 154, i32 159>
  store <32 x i32> %strided.vec0, ptr %out.vec0, align 64
  store <32 x i32> %strided.vec1, ptr %out.vec1, align 64
  store <32 x i32> %strided.vec2, ptr %out.vec2, align 64
  store <32 x i32> %strided.vec3, ptr %out.vec3, align 64
  store <32 x i32> %strided.vec4, ptr %out.vec4, align 64
  ret void
}

define void @load_i32_stride5_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3, ptr %out.vec4) nounwind {
; SSE-LABEL: load_i32_stride5_vf64:
; SSE:       # %bb.0:
; SSE-NEXT:    subq $1928, %rsp # imm = 0x788
; SSE-NEXT:    movdqa 768(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 752(%rdi), %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 720(%rdi), %xmm11
; SSE-NEXT:    movdqa %xmm11, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 736(%rdi), %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 448(%rdi), %xmm5
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 432(%rdi), %xmm6
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 400(%rdi), %xmm12
; SSE-NEXT:    movdqa %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 416(%rdi), %xmm7
; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 128(%rdi), %xmm8
; SSE-NEXT:    movdqa %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 112(%rdi), %xmm9
; SSE-NEXT:    movdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 80(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 96(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm9[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm8[2],xmm0[3],xmm8[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm7[1,1,1,1]
; SSE-NEXT:    movdqa %xmm12, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm5[2],xmm0[3],xmm5[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[1,1,1,1]
; SSE-NEXT:    movdqa %xmm11, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 1040(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 1056(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 1088(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 1072(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa (%rdi), %xmm10
; SSE-NEXT:    movdqa 16(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa %xmm10, %xmm1
; SSE-NEXT:    movdqa %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 32(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 48(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-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 336(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 368(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 352(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 640(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 656(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 688(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 672(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 960(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 976(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 1008(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 992(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 240(%rdi), %xmm6
; SSE-NEXT:    movdqa 256(%rdi), %xmm12
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm12[1,1,1,1]
; SSE-NEXT:    movdqa %xmm6, %xmm1
; SSE-NEXT:    movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 288(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 272(%rdi), %xmm7
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm7[2,2,2,2]
; SSE-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 560(%rdi), %xmm15
; SSE-NEXT:    movdqa 576(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa %xmm15, %xmm1
; SSE-NEXT:    movdqa %xmm15, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 608(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 592(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 880(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 896(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa %xmm2, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 928(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 912(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 1200(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 1216(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa %xmm2, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 1248(%rdi), %xmm3
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 1232(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm3[2],xmm0[3],xmm3[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 160(%rdi), %xmm14
; SSE-NEXT:    movdqa 176(%rdi), %xmm3
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[1,1,1,1]
; SSE-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm14, %xmm1
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 208(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 192(%rdi), %xmm5
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm5[2,2,2,2]
; SSE-NEXT:    movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 480(%rdi), %xmm8
; SSE-NEXT:    movdqa 496(%rdi), %xmm11
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm11[1,1,1,1]
; SSE-NEXT:    movdqa %xmm11, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm8, %xmm1
; SSE-NEXT:    movdqa %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 528(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 512(%rdi), %xmm9
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm9[2,2,2,2]
; SSE-NEXT:    movdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 800(%rdi), %xmm13
; SSE-NEXT:    movdqa 816(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    movdqa %xmm13, %xmm1
; SSE-NEXT:    movdqa %xmm13, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 848(%rdi), %xmm2
; SSE-NEXT:    movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 832(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 1120(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, (%rsp) # 16-byte Spill
; SSE-NEXT:    movdqa 1136(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movdqa 1168(%rdi), %xmm4
; SSE-NEXT:    movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa 1152(%rdi), %xmm0
; SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,2,2]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm4[2],xmm0[3],xmm4[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 144(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm10[1,1,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 64(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,1,1]
; SSE-NEXT:    movdqa %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm12[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 304(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm7[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm14, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm14[1,1,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 224(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm5[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
; SSE-NEXT:    movapd %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 464(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
; SSE-NEXT:    movapd %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm12[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 384(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm15[1,1,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 624(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm8[1,1,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 544(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm9[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 784(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm10[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm9[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 704(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm7[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 944(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
; SSE-NEXT:    movapd %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm13[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm13[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 864(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
; SSE-NEXT:    movapd %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 1104(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
; SSE-NEXT:    movapd %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm15[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 1024(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 1264(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa (%rsp), %xmm3 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm5[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movdqa 1184(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT:    pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = mem[2,3,2,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm14 = xmm14[2],xmm1[2],xmm14[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm14[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm12 = xmm12[2],xmm1[2],xmm12[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm14[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm12[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa %xmm10, %xmm0
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm9[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm12[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm10[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm13[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm7[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm15[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm7[0],xmm1[1],xmm7[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm2[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm5[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm3 = xmm3[2],xmm1[2],xmm3[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm13[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm3[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,3,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm9[0,0,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm15[0],xmm0[1],xmm15[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT:    movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm14 = xmm14[0],xmm0[0],xmm14[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm14 = xmm0[0],xmm14[1]
; SSE-NEXT:    movapd %xmm14, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1]
; SSE-NEXT:    movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa %xmm12, %xmm4
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm0[0],xmm10[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm10 = xmm0[0],xmm10[1]
; SSE-NEXT:    movapd %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; SSE-NEXT:    movdqa %xmm5, %xmm4
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $250, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm0[0],xmm10[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm10 = xmm0[0],xmm10[1]
; SSE-NEXT:    movapd %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm7[2,2,3,3]
; SSE-NEXT:    movdqa %xmm11, %xmm4
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[2,2,3,3]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm8[2,2,3,3]
; SSE-NEXT:    movdqa %xmm13, %xmm4
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $255, (%rsp), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm0[0],xmm4[1]
; SSE-NEXT:    movapd %xmm4, (%rsp) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[2,2,3,3]
; SSE-NEXT:    movdqa %xmm9, %xmm13
; SSE-NEXT:    punpckldq {{.*#+}} xmm13 = xmm13[0],xmm0[0],xmm13[1],xmm0[1]
; SSE-NEXT:    pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[3,3,3,3]
; SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
; SSE-NEXT:    movsd {{.*#+}} xmm13 = xmm0[0],xmm13[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[2],mem[2],xmm0[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm4[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm15[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[2],mem[2],xmm0[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm4[0],xmm0[1]
; SSE-NEXT:    movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
; SSE-NEXT:    # xmm14 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
; SSE-NEXT:    # xmm14 = xmm14[2],mem[2],xmm14[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm14 = xmm4[0],xmm14[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
; SSE-NEXT:    # xmm12 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
; SSE-NEXT:    # xmm12 = xmm12[2],mem[2],xmm12[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm12 = xmm4[0],xmm12[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
; SSE-NEXT:    # xmm11 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
; SSE-NEXT:    # xmm11 = xmm11[2],mem[2],xmm11[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm11 = xmm4[0],xmm11[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    # xmm10 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    # xmm10 = xmm10[2],mem[2],xmm10[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm10 = xmm1[0],xmm10[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Folded Reload
; SSE-NEXT:    # xmm9 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Folded Reload
; SSE-NEXT:    # xmm9 = xmm9[2],mem[2],xmm9[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm9 = xmm1[0],xmm9[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
; SSE-NEXT:    # xmm8 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
; SSE-NEXT:    # xmm8 = xmm8[2],mem[2],xmm8[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm8 = xmm1[0],xmm8[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Folded Reload
; SSE-NEXT:    # xmm7 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Folded Reload
; SSE-NEXT:    # xmm7 = xmm7[2],mem[2],xmm7[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm7 = xmm1[0],xmm7[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
; SSE-NEXT:    # xmm6 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
; SSE-NEXT:    # xmm6 = xmm6[2],mem[2],xmm6[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm6 = xmm3[0],xmm6[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Folded Reload
; SSE-NEXT:    # xmm5 = xmm5[2],mem[2],xmm5[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm5 = xmm1[0],xmm5[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
; SSE-NEXT:    # xmm4 = xmm4[2],mem[2],xmm4[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm1[0],xmm4[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = xmm3[2],mem[2],xmm3[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm1[0],xmm3[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = xmm2[2],mem[2],xmm2[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm15[0],xmm2[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
; SSE-NEXT:    # xmm1 = xmm1[2],mem[2],xmm1[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm15[0],xmm1[1]
; SSE-NEXT:    pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[1,1,1,1]
; SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    punpckldq {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1]
; SSE-NEXT:    pshufd $170, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = mem[2,2,2,2]
; SSE-NEXT:    punpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[2],mem[2],xmm0[3],mem[3]
; SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm15[0],xmm0[1]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 224(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 160(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 96(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 32(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 240(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 176(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 112(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 48(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 192(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 128(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 64(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, (%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 208(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 144(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 80(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 16(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 224(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 240(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 192(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 208(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 160(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 176(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 128(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 144(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 96(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 112(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 64(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 80(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 32(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 48(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, (%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 16(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 240(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 224(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 208(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 192(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 176(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 160(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 144(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 128(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 112(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 96(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 80(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 64(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 48(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 32(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, 16(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT:    movaps %xmm15, (%rcx)
; SSE-NEXT:    movapd %xmm13, 240(%r8)
; SSE-NEXT:    movaps (%rsp), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 224(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 208(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 192(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 176(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 160(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 144(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 128(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 112(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 96(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 80(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 64(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 48(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 32(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, 16(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; SSE-NEXT:    movaps %xmm13, (%r8)
; SSE-NEXT:    movapd %xmm0, 240(%r9)
; SSE-NEXT:    movapd %xmm1, 224(%r9)
; SSE-NEXT:    movapd %xmm2, 208(%r9)
; SSE-NEXT:    movapd %xmm3, 192(%r9)
; SSE-NEXT:    movapd %xmm4, 176(%r9)
; SSE-NEXT:    movapd %xmm5, 160(%r9)
; SSE-NEXT:    movapd %xmm6, 144(%r9)
; SSE-NEXT:    movapd %xmm7, 128(%r9)
; SSE-NEXT:    movapd %xmm8, 112(%r9)
; SSE-NEXT:    movapd %xmm9, 96(%r9)
; SSE-NEXT:    movapd %xmm10, 80(%r9)
; SSE-NEXT:    movapd %xmm11, 64(%r9)
; SSE-NEXT:    movapd %xmm12, 48(%r9)
; SSE-NEXT:    movapd %xmm14, 32(%r9)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 16(%r9)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, (%r9)
; SSE-NEXT:    addq $1928, %rsp # imm = 0x788
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i32_stride5_vf64:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    subq $2472, %rsp # imm = 0x9A8
; AVX1-ONLY-NEXT:    vmovaps 832(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vmovups %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 544(%rdi), %ymm9
; AVX1-ONLY-NEXT:    vmovups %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 576(%rdi), %ymm10
; AVX1-ONLY-NEXT:    vmovups %ymm10, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 512(%rdi), %ymm7
; AVX1-ONLY-NEXT:    vmovups %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 480(%rdi), %ymm8
; AVX1-ONLY-NEXT:    vmovups %ymm8, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 224(%rdi), %ymm6
; AVX1-ONLY-NEXT:    vmovaps 256(%rdi), %ymm5
; AVX1-ONLY-NEXT:    vmovaps 192(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 160(%rdi), %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 288(%rdi), %ymm1, %ymm2
; AVX1-ONLY-NEXT:    vmovaps 256(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, (%rsp) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm0[0,1,2,3,4,5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm6[4],ymm3[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3],ymm1[4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm8[0,1],ymm7[2,3],ymm8[4,5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm1[0],xmm2[1],xmm1[2],xmm2[3]
; AVX1-ONLY-NEXT:    vmovaps 576(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm0[0,1,2,3,4,5],ymm10[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm9[4],ymm3[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm2[0,1,2,3],ymm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 608(%rdi), %ymm2, %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2],ymm2[3],ymm3[4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 800(%rdi), %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm0[0,1],ymm4[2,3],ymm0[4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3]
; AVX1-ONLY-NEXT:    vmovaps 864(%rdi), %ymm8
; AVX1-ONLY-NEXT:    vmovaps 896(%rdi), %ymm7
; AVX1-ONLY-NEXT:    vmovaps 896(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm8[4],ymm3[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm2[0,1,2,3],ymm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 928(%rdi), %ymm2, %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2],ymm2[3],ymm3[4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 1152(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 1120(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm3[2,3],ymm2[4,5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3]
; AVX1-ONLY-NEXT:    vmovaps 1184(%rdi), %ymm13
; AVX1-ONLY-NEXT:    vmovaps 1216(%rdi), %ymm12
; AVX1-ONLY-NEXT:    vmovaps 1216(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm12[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm13[4],ymm3[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm2[0,1,2,3],ymm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 1248(%rdi), %ymm2, %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2],ymm2[3],ymm3[4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3]
; AVX1-ONLY-NEXT:    vmovaps 64(%rdi), %ymm11
; AVX1-ONLY-NEXT:    vmovaps 96(%rdi), %ymm10
; AVX1-ONLY-NEXT:    vmovaps 96(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm10[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm11[4],ymm3[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm2[0,1,2,3],ymm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 128(%rdi), %ymm2, %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2],ymm2[3],ymm3[4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 352(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 320(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3]
; AVX1-ONLY-NEXT:    vmovaps 384(%rdi), %ymm14
; AVX1-ONLY-NEXT:    vmovaps 416(%rdi), %ymm9
; AVX1-ONLY-NEXT:    vmovaps 416(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm9[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm14[4],ymm3[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm2[0,1,2,3],ymm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 448(%rdi), %ymm2, %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2],ymm2[3],ymm3[4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 672(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 640(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3]
; AVX1-ONLY-NEXT:    vmovaps 704(%rdi), %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 736(%rdi), %ymm15
; AVX1-ONLY-NEXT:    vmovaps 736(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm0[4],ymm3[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm2[0,1,2,3],ymm3[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 768(%rdi), %ymm2, %ymm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2],ymm2[3],ymm3[4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 992(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 960(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm3[2,3],ymm2[4,5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm2[0],xmm3[1],xmm2[2],xmm3[3]
; AVX1-ONLY-NEXT:    vmovaps 1024(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 1056(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovaps 1056(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vmovaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm4, %ymm0, %ymm0
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm0[0,1,2,3,4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3],ymm1[4],ymm4[5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm3[0,1,2,3],ymm4[4,5,6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, 1088(%rdi), %ymm3, %ymm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3],ymm4[4,5,6],ymm3[7]
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm5[2,3,0,1]
; AVX1-ONLY-NEXT:    vmovups %ymm6, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm6[1,3],ymm3[6,5],ymm6[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm6[0,1],ymm5[2,3],ymm6[4,5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,0],ymm4[3,0],ymm3[6,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vmovaps 160(%rdi), %xmm5
; AVX1-ONLY-NEXT:    vmovaps %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 192(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vmovaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm5[0,1],xmm4[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm4 = xmm4[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 304(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm6[2,3,0,1]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm0[1,3],ymm3[6,5],ymm0[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm0[0,1],ymm6[2,3],ymm0[4,5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,0],ymm4[3,0],ymm3[6,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vmovaps 480(%rdi), %xmm5
; AVX1-ONLY-NEXT:    vmovaps %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 512(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vmovaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm5[0,1],xmm4[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm4 = xmm4[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 624(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm7[2,3,0,1]
; AVX1-ONLY-NEXT:    vmovups %ymm8, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm8[1,3],ymm3[6,5],ymm8[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm8[0,1],ymm7[2,3],ymm8[4,5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,0],ymm4[3,0],ymm3[6,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vmovaps 800(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 832(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm4 = xmm4[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 944(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm12, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm12[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm13[1,3],ymm3[6,5],ymm13[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm13[0,1],ymm12[2,3],ymm13[4,5],ymm12[6,7]
; AVX1-ONLY-NEXT:    vmovaps %ymm13, %ymm6
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,0],ymm4[3,0],ymm3[6,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vmovaps 1120(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 1152(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm4 = xmm4[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 1264(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm10, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm10[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm11[1,3],ymm3[6,5],ymm11[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm11[0,1],ymm10[2,3],ymm11[4,5],ymm10[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,0],ymm4[3,0],ymm3[6,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vmovaps (%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 32(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm4 = xmm4[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 144(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm9[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm14[1,3],ymm3[6,5],ymm14[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm14[0,1],ymm9[2,3],ymm14[4,5],ymm9[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,0],ymm4[3,0],ymm3[6,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vmovaps 320(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 352(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm4 = xmm4[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 464(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm15, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm15[2,3,0,1]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm13 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm13[1,3],ymm3[6,5],ymm13[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm13[0,1],ymm15[2,3],ymm13[4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,0],ymm4[3,0],ymm3[6,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vmovaps 640(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 672(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm4 = xmm4[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 784(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm2[2,3,0,1]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm9 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,1],ymm9[1,3],ymm3[6,5],ymm9[5,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm9[0,1],ymm2[2,3],ymm9[4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[2,0],ymm4[3,0],ymm3[6,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vmovaps 960(%rdi), %xmm0
; AVX1-ONLY-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vmovaps 992(%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],mem[2],xmm4[3]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm4 = xmm4[1,2,3,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vbroadcastss 1104(%rdi), %ymm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5,6],ymm4[7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovups (%rsp), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm1[2,0],ymm0[7,4],ymm1[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm1[2,1],ymm0[6,4],ymm1[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm3 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm3 = mem[0,1,2,3],ymm1[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm3, %xmm3
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm3 = xmm3[0,1],mem[2],xmm3[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm3 = xmm3[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 288(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm10 = ymm1[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm1[1,0],ymm10[0,0],ymm1[5,4],ymm10[4,4]
; AVX1-ONLY-NEXT:    vmovups %ymm10, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm3 = ymm3[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm1[2,0],ymm0[7,4],ymm1[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm1[2,1],ymm0[6,4],ymm1[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm1 = mem[0,1,2,3],ymm1[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm1, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],mem[2],xmm1[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 608(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm2[2,3,0,1]
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm2[1,0],ymm1[0,0],ymm2[5,4],ymm1[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm1[2,0],ymm0[7,4],ymm1[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm1[2,1],ymm0[6,4],ymm1[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm1 = ymm1[0,1,2,3],mem[4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm1, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],mem[2],xmm1[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 928(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm2[2,3,0,1]
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm2[1,0],ymm1[0,0],ymm2[5,4],ymm1[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups %ymm6, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm6[2,0],ymm0[7,4],ymm6[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm6[2,1],ymm0[6,4],ymm6[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm1 = mem[0,1,2,3],ymm1[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm1, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],mem[2],xmm1[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 1248(%rdi), %ymm1
; AVX1-ONLY-NEXT:    vmovups %ymm1, (%rsp) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm7 = ymm1[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[1,0],ymm7[0,0],ymm1[5,4],ymm7[4,4]
; AVX1-ONLY-NEXT:    vmovups %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm9[2,0],ymm0[7,4],ymm9[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm9[2,1],ymm0[6,4],ymm9[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm1 = mem[0,1,2,3],ymm1[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm1, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],mem[2],xmm1[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 1088(%rdi), %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm2[2,3,0,1]
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm2[1,0],ymm1[0,0],ymm2[5,4],ymm1[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm11[2,0],ymm0[7,4],ymm11[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm11[2,1],ymm0[6,4],ymm11[6,5]
; AVX1-ONLY-NEXT:    vmovaps %ymm11, %ymm4
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm1 = ymm1[0,1,2,3],mem[4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm1, %xmm1
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],mem[2],xmm1[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 128(%rdi), %ymm3
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm9 = ymm3[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm2 = ymm3[1,0],ymm9[0,0],ymm3[5,4],ymm9[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm2 = ymm2[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3,4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm0[3,0],ymm14[2,0],ymm0[7,4],ymm14[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[2,0],ymm14[2,1],ymm1[6,4],ymm14[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm2 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm2 = ymm0[0,1,2,3],mem[4,5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm2, %xmm2
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],mem[2],xmm2[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm2 = xmm2[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1,2],ymm1[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 448(%rdi), %ymm6
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm12 = ymm6[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm5 = ymm6[1,0],ymm12[0,0],ymm6[5,4],ymm12[4,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm5 = ymm5[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3,4,5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm2 = ymm0[3,0],ymm13[2,0],ymm0[7,4],ymm13[6,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm2 = ymm2[2,0],ymm13[2,1],ymm2[6,4],ymm13[6,5]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm5 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm5 = ymm0[0,1,2,3],mem[4,5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm5, %xmm5
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm5 = xmm5[0,1],mem[2],xmm5[3]
; AVX1-ONLY-NEXT:    vshufpd {{.*#+}} xmm5 = xmm5[1,0]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm5[0,1,2],ymm2[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovaps 768(%rdi), %ymm8
; AVX1-ONLY-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm8[2,3,0,1]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm8[1,0],ymm1[0,0],ymm8[5,4],ymm1[4,4]
; AVX1-ONLY-NEXT:    vmovaps %ymm1, %ymm2
; AVX1-ONLY-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm11 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm11[0,0],ymm1[3,0],ymm11[4,4],ymm1[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm1[2,2],ymm15[6,4],ymm1[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm1[2,0],ymm10[1,0],ymm1[6,4],ymm10[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm1[0,0],ymm5[3,0],ymm1[4,4],ymm5[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm5[2,2],ymm15[6,4],ymm5[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps $18, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm15 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm15 = ymm1[2,0],mem[1,0],ymm1[6,4],mem[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm10[0,0],ymm1[3,0],ymm10[4,4],ymm1[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm1[2,2],ymm15[6,4],ymm1[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps $18, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm15 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm15 = ymm1[2,0],mem[1,0],ymm1[6,4],mem[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm15 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,0],ymm1[3,0],ymm15[4,4],ymm1[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm1[2,2],ymm15[6,4],ymm1[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups (%rsp), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm1[2,0],ymm7[1,0],ymm1[6,4],ymm7[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm10[0,0],ymm1[3,0],ymm10[4,4],ymm1[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm1[2,2],ymm15[6,4],ymm1[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps $18, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm15 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm15 = ymm1[2,0],mem[1,0],ymm1[6,4],mem[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm5[0,0],ymm13[3,0],ymm5[4,4],ymm13[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm13[2,2],ymm15[6,4],ymm13[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm8[2,0],ymm2[1,0],ymm8[6,4],ymm2[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm2[0,0],ymm14[3,0],ymm2[4,4],ymm14[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm14[2,2],ymm15[6,4],ymm14[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm6[2,0],ymm12[1,0],ymm6[6,4],ymm12[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-ONLY-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; AVX1-ONLY-NEXT:    vblendps $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    # xmm0 = mem[0,1],xmm0[2,3]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm1[0,0],ymm4[3,0],ymm1[4,4],ymm4[7,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[2,0],ymm4[2,2],ymm15[6,4],ymm4[6,6]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,0],mem[1,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm15[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm3[2,0],ymm9[1,0],ymm3[6,4],ymm9[5,4]
; AVX1-ONLY-NEXT:    vshufps {{.*#+}} ymm15 = ymm15[0,1,2,0,4,5,6,4]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm15 = ymm0[0,1,2,3,4,5],ymm15[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm0 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5],ymm0[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm7 = mem[0,1,2,3],ymm7[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm7, %xmm7
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm7 = xmm7[0],mem[1],xmm7[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm7[0,1,2],ymm0[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm9[0,1,2,3,4,5,6],ymm3[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm3 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4],ymm11[5],ymm3[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm4 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm4 = ymm4[0,1,2,3],mem[4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm4, %xmm4
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0],mem[1],xmm4[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2],ymm3[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $128, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm4 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm4 = ymm1[0,1,2,3,4,5,6],mem[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm14, %ymm4 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3,4],ymm2[5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm7 = mem[0,1,2,3],ymm7[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm7, %xmm7
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm7 = xmm7[0],mem[1],xmm7[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm7[0,1,2],ymm4[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm12[0,1,2,3,4,5,6],ymm6[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm1 = ymm4[0,1,2,3,4,5],ymm1[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm4 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps $32, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm4 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm4 = ymm4[0,1,2,3,4],mem[5],ymm4[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm6 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm6 = ymm6[0,1,2,3],mem[4,5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm6, %xmm6
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm6 = xmm6[0],mem[1],xmm6[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm6[0,1,2],ymm4[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $127, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm6 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm6 = mem[0,1,2,3,4,5,6],ymm2[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3,4,5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm13, %ymm6 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm6 = ymm6[0,1,2,3,4],ymm5[5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm7 = mem[0,1,2,3],ymm7[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm7, %xmm7
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm7 = xmm7[0],mem[1],xmm7[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm6 = ymm7[0,1,2],ymm6[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vblendps $127, {{[-0-9]+}}(%r{{[sb]}}p), %ymm8, %ymm2 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm2 = mem[0,1,2,3,4,5,6],ymm8[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm2 = ymm6[0,1,2,3,4,5],ymm2[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm5 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps $32, {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm5 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm5 = ymm5[0,1,2,3,4],mem[5],ymm5[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm6 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm6 = mem[0,1,2,3],ymm6[4,5],mem[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm6, %xmm6
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm6 = xmm6[0],mem[1],xmm6[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm5 = ymm6[0,1,2],ymm5[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $127, {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm6 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm6 = mem[0,1,2,3,4,5,6],ymm6[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm5 = ymm5[0,1,2,3,4,5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm6 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm6 = ymm6[0,1,2,3,4],ymm10[5],ymm6[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm7 = ymm7[0,1,2,3],mem[4,5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm7, %xmm7
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm7 = xmm7[0],mem[1],xmm7[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm6 = ymm7[0,1,2],ymm6[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $127, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm7 = mem[0,1,2,3,4,5,6],ymm7[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm6 = ymm6[0,1,2,3,4,5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX1-ONLY-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 16-byte Folded Reload
; AVX1-ONLY-NEXT:    vblendps $32, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm7 = ymm7[0,1,2,3,4],mem[5],ymm7[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm8, %ymm8 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm8 = ymm8[0,1,2,3],mem[4,5],ymm8[6,7]
; AVX1-ONLY-NEXT:    vextractf128 $1, %ymm8, %xmm8
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} xmm8 = xmm8[0],mem[1],xmm8[2,3]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm7 = ymm8[0,1,2],ymm7[3,4,5,6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vblendps $128, (%rsp), %ymm8, %ymm8 # 32-byte Folded Reload
; AVX1-ONLY-NEXT:    # ymm8 = ymm8[0,1,2,3,4,5,6],mem[7]
; AVX1-ONLY-NEXT:    vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4,5],ymm8[6,7]
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 192(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 128(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 64(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, (%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 224(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 160(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 96(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 32(%rsi)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 192(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 128(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 64(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, (%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 224(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 160(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 96(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 32(%rdx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 128(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 64(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, (%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 192(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 224(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 160(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 96(%rcx)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 32(%rcx)
; AVX1-ONLY-NEXT:    vmovaps %ymm15, (%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 64(%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 128(%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 192(%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 224(%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 160(%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 96(%r8)
; AVX1-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX1-ONLY-NEXT:    vmovaps %ymm8, 32(%r8)
; AVX1-ONLY-NEXT:    vmovaps %ymm7, 224(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm6, 192(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm5, 160(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm2, 128(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm4, 96(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm1, 64(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm3, 32(%r9)
; AVX1-ONLY-NEXT:    vmovaps %ymm0, (%r9)
; AVX1-ONLY-NEXT:    addq $2472, %rsp # imm = 0x9A8
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i32_stride5_vf64:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    subq $2120, %rsp # imm = 0x848
; AVX2-ONLY-NEXT:    vmovdqa 864(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vmovdqu %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 896(%rdi), %ymm5
; AVX2-ONLY-NEXT:    vmovdqu %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 832(%rdi), %ymm6
; AVX2-ONLY-NEXT:    vmovdqu %ymm6, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 800(%rdi), %ymm7
; AVX2-ONLY-NEXT:    vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 544(%rdi), %ymm8
; AVX2-ONLY-NEXT:    vmovdqu %ymm8, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 576(%rdi), %ymm9
; AVX2-ONLY-NEXT:    vmovdqa 512(%rdi), %ymm10
; AVX2-ONLY-NEXT:    vmovdqa 480(%rdi), %ymm11
; AVX2-ONLY-NEXT:    vmovdqu %ymm11, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 224(%rdi), %ymm12
; AVX2-ONLY-NEXT:    vmovdqu %ymm12, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 256(%rdi), %ymm13
; AVX2-ONLY-NEXT:    vmovdqu %ymm13, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 192(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 160(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = [0,5,2,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm2[2,3],ymm1[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 288(%rdi), %ymm1, %ymm2
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm3 = ymm13[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm12[4],ymm3[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3],ymm1[4,5,6],ymm2[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm11[0,1],ymm10[2,3],ymm11[4,5],ymm10[6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm10, %ymm14
; AVX2-ONLY-NEXT:    vmovdqu %ymm10, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm9[0,1,0,3]
; AVX2-ONLY-NEXT:    vmovdqa %ymm9, %ymm13
; AVX2-ONLY-NEXT:    vmovdqu %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm8[4],ymm2[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 608(%rdi), %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm7[0,1],ymm6[2,3],ymm7[4,5],ymm6[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm5[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm4[4],ymm2[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 928(%rdi), %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 1152(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqa 1120(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm3[0,1],ymm2[2,3],ymm3[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vmovdqa %ymm3, %ymm9
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, %ymm8
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vmovdqa 1184(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 1216(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm4[0,1,0,3]
; AVX2-ONLY-NEXT:    vmovdqa %ymm4, %ymm11
; AVX2-ONLY-NEXT:    vmovdqu %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4],ymm2[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 1248(%rdi), %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 352(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 320(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm2[2,3],ymm1[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vmovdqa 384(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, (%rsp) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 416(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4],ymm2[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 448(%rdi), %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 672(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 640(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm2[2,3],ymm1[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vmovdqa 704(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 736(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4],ymm2[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 768(%rdi), %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 992(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 960(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm2[2,3],ymm1[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vmovdqa 1024(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa 1056(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,1,0,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4],ymm2[5,6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 1088(%rdi), %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm6
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm15
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm6[0,1],ymm15[2,3],ymm6[4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm15, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu %ymm6, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %ymm4
; AVX2-ONLY-NEXT:    vmovdqa 96(%rdi), %ymm5
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm1 = ymm5[0,1,0,3]
; AVX2-ONLY-NEXT:    vmovdqu %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm4[4],ymm1[5,6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX2-ONLY-NEXT:    vinserti128 $1, 128(%rdi), %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3],ymm1[4,5,6],ymm0[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm0 = <1,6,3,u>
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm12 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm12[0,1],ymm7[2,3],ymm12[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $204, {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = ymm10[0,1],mem[2,3],ymm10[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vbroadcasti128 {{.*#+}} ymm1 = [5,2,7,0,5,2,7,0]
; AVX2-ONLY-NEXT:    # ymm1 = mem[0,1,0,1]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm1, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 304(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd $243, {{[-0-9]+}}(%r{{[sb]}}p), %ymm14, %ymm2 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm2 = mem[0,1],ymm14[2,3],mem[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm14 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm14[0,1],ymm13[2,3],ymm14[4,5],ymm13[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm1, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 624(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $243, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm2 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm2 = mem[0,1],ymm2[2,3],mem[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $51, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = mem[0,1],ymm3[2,3],mem[4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm1, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 944(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm9[0,1],ymm8[2,3],ymm9[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm9 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm9[0,1],ymm11[2,3],ymm9[4,5],ymm11[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm1, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 1264(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm6[0,1],ymm15[2,3],ymm6[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm4[0,1],ymm5[2,3],ymm4[4,5],ymm5[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm1, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 144(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm11 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $243, {{[-0-9]+}}(%r{{[sb]}}p), %ymm11, %ymm2 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm2 = mem[0,1],ymm11[2,3],mem[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vmovdqu (%rsp), %ymm3 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $204, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = ymm3[0,1],mem[2,3],ymm3[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm1, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 464(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $243, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm2 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm2 = mem[0,1],ymm2[2,3],mem[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm2
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $51, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = mem[0,1],ymm3[2,3],mem[4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm1, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 784(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm13 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm13[0,1],ymm8[2,3],ymm13[4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm15 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $204, {{[-0-9]+}}(%r{{[sb]}}p), %ymm15, %ymm2 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm2 = ymm15[0,1],mem[2,3],ymm15[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastd 1104(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm2 = <2,7,4,u>
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm12[0,1,2,3],ymm7[4,5],ymm12[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm0, %ymm2, %ymm0
; AVX2-ONLY-NEXT:    vinserti128 $1, 256(%rdi), %ymm0, %ymm1
; AVX2-ONLY-NEXT:    vmovdqa %ymm10, %ymm4
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm10[0,1,2,3,4,5,6],ymm1[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm0[0,1,2],ymm1[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 288(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm0 = [0,5,0,5,0,5,0,5]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm1 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm1 = ymm10[0,1,2,3],mem[4,5],ymm10[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm2, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 576(%rdi), %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm14[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 608(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm1 = mem[0,1,2,3],ymm1[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm2, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 896(%rdi), %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm5[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 928(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm7[0,1,2,3],ymm6[4,5],ymm7[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm2, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 1216(%rdi), %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm9[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 1248(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm13[0,1,2,3],ymm8[4,5],ymm13[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm2, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 1056(%rdi), %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm15[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 1088(%rdi), %ymm8
; AVX2-ONLY-NEXT:    vpermd %ymm8, %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm8, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm1 = ymm1[0,1,2,3],mem[4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm2, %ymm1
; AVX2-ONLY-NEXT:    vinserti128 $1, 96(%rdi), %ymm0, %ymm3
; AVX2-ONLY-NEXT:    vpblendd $127, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = mem[0,1,2,3,4,5,6],ymm3[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm3 = ymm3[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm1[0,1,2],ymm3[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 128(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm15
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm11, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = mem[0,1,2,3],ymm11[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm2, %ymm3
; AVX2-ONLY-NEXT:    vinserti128 $1, 416(%rdi), %ymm0, %ymm15
; AVX2-ONLY-NEXT:    vpblendd $127, (%rsp), %ymm15, %ymm15 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm15 = mem[0,1,2,3,4,5,6],ymm15[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm15 = ymm15[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm15 = ymm3[0,1,2],ymm15[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 448(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm0, %ymm13
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm15[0,1,2,3,4,5],ymm13[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm9 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm11 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm11[0,1,2,3],ymm9[4,5],ymm11[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm13, %ymm2, %ymm2
; AVX2-ONLY-NEXT:    vinserti128 $1, 736(%rdi), %ymm0, %ymm13
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm14 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm14[0,1,2,3,4,5,6],ymm13[7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm13 = ymm13[2,3,0,1,6,7,4,5]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm2[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqa 768(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vpermd %ymm2, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, %ymm3
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm13[0,1,2,3,4,5],ymm0[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm0 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm0 = ymm4[0,1,2,3],mem[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm13 = ymm12[12,13,14,15],ymm2[0,1,2,3,4,5,6,7,8,9,10,11],ymm12[28,29,30,31],ymm2[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm13 = ymm13[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm13[0,1,2],ymm0[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm0 = [1,6,1,6,1,6,1,6]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm15 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm13[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm13 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm13 = ymm2[0,1,2,3],mem[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm13 = ymm13[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm15 = ymm10[12,13,14,15],ymm2[0,1,2,3,4,5,6,7,8,9,10,11],ymm10[28,29,30,31],ymm2[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm15 = ymm15[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm15[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm15 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm13[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm13 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm13 = ymm5[0,1,2,3],mem[4,5],ymm5[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm13 = ymm13[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpalignr $12, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm15 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm15 = mem[12,13,14,15],ymm2[0,1,2,3,4,5,6,7,8,9,10,11],mem[28,29,30,31],ymm2[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm15 = ymm15[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm15[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm15 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm13[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm13 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm13 = ymm2[0,1,2,3],mem[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm13 = ymm13[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm15 = ymm7[12,13,14,15],ymm6[0,1,2,3,4,5,6,7,8,9,10,11],ymm7[28,29,30,31],ymm6[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm15 = ymm15[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm15[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm15 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm13[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm13 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm13 = mem[0,1,2,3],ymm2[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm13 = ymm13[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpalignr $12, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm15 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm15 = mem[12,13,14,15],ymm2[0,1,2,3,4,5,6,7,8,9,10,11],mem[28,29,30,31],ymm2[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm15 = ymm15[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm15[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm8, %ymm0, %ymm15
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm13[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm12 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm14[0,1,2,3],ymm12[4,5],ymm14[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm13 = ymm13[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm15 = ymm11[12,13,14,15],ymm9[0,1,2,3,4,5,6,7,8,9,10,11],ymm11[28,29,30,31],ymm9[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vmovdqa %ymm11, %ymm14
; AVX2-ONLY-NEXT:    vmovdqa %ymm9, %ymm11
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm15 = ymm15[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm15[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm0, %ymm15
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm13[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu (%rsp), %ymm9 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm9[0,1,2,3],ymm8[4,5],ymm9[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm13 = ymm13[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm15 = ymm3[12,13,14,15],ymm2[0,1,2,3,4,5,6,7,8,9,10,11],ymm3[28,29,30,31],ymm2[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm15 = ymm15[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm15[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpermd %ymm5, %ymm0, %ymm15
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm13[0,1,2,3,4,5],ymm15[6,7]
; AVX2-ONLY-NEXT:    vmovdqu %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm10[0,1,2,3],ymm7[4,5],ymm10[6,7]
; AVX2-ONLY-NEXT:    vpshufd {{.*#+}} ymm13 = ymm13[3,0,2,2,7,4,6,6]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpalignr {{.*#+}} ymm15 = ymm4[12,13,14,15],ymm6[0,1,2,3,4,5,6,7,8,9,10,11],ymm4[28,29,30,31],ymm6[16,17,18,19,20,21,22,23,24,25,26,27]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm15 = ymm15[0,3,2,3]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm15[0,1,2],ymm13[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm15 = ymm13[0,1,2,3,4,5],ymm0[6,7]
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm10[0,1],ymm7[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm7[5],ymm0[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm6[0,1,2,3],ymm4[4,5],ymm6[6,7]
; AVX2-ONLY-NEXT:    vmovdqa {{.*#+}} xmm7 = <4,1,6,u>
; AVX2-ONLY-NEXT:    vpermd %ymm4, %ymm7, %ymm4
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm4[0,1,2],ymm0[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm6 = [2,7,2,7,2,7,2,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm6, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm1 = mem[0,1,2,3],ymm1[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-ONLY-NEXT:    vperm2i128 $2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm4 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm4 = mem[0,1],ymm10[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm4[0,1,2,3,4],ymm10[5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm7, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm4[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm4 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm13 = ymm1[0,1,2,3,4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2,3],ymm3[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm9[0,1],ymm8[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm4[0,1,2,3,4],ymm8[5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm1, %ymm7, %ymm1
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm4[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm5, %ymm6, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = mem[0,1,2,3],ymm2[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX2-ONLY-NEXT:    vperm2i128 $2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm8, %ymm4 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm4 = mem[0,1],ymm8[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm4[0,1,2,3,4],ymm8[5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm7, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm4[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm4 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm3[0,1,2,3,4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm11[0,1,2,3],ymm14[4,5],ymm11[6,7]
; AVX2-ONLY-NEXT:    vperm2i128 $2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm12, %ymm8 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm8 = mem[0,1],ymm12[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm8[0,1,2,3,4],ymm12[5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm7, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm8[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm2 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm3[0,1,2,3,4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm3 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm3 = mem[0,1,2,3],ymm3[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-ONLY-NEXT:    vperm2i128 $2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm8 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm8 = mem[0,1],ymm10[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm8[0,1,2,3,4],ymm10[5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm3, %ymm7, %ymm3
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm8[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm8 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm8[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm8 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm8 = ymm5[0,1,2,3],mem[4,5],ymm5[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vperm2i128 $2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm10 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm10 = mem[0,1],ymm5[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm10 = ymm10[0,1,2,3,4],ymm5[5],ymm10[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm8, %ymm7, %ymm8
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm8[0,1,2],ymm10[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm9 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm8 = ymm8[0,1,2,3,4,5],ymm9[6,7]
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vpblendd $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm9 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm9 = mem[0,1,2,3],ymm5[4,5],mem[6,7]
; AVX2-ONLY-NEXT:    vpermd %ymm9, %ymm7, %ymm7
; AVX2-ONLY-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-ONLY-NEXT:    vperm2i128 $2, {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm9 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    # ymm9 = mem[0,1],ymm10[0,1]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm9 = ymm9[0,1,2,3,4],ymm10[5],ymm9[6,7]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm7 = ymm7[0,1,2],ymm9[3,4,5,6,7]
; AVX2-ONLY-NEXT:    vpermd {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm6 # 32-byte Folded Reload
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm6 = ymm7[0,1,2,3,4,5],ymm6[6,7]
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 192(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 128(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm7, 64(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, (%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm7, 224(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm7, 160(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm7, 96(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm7 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm7, 32(%rsi)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 192(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 128(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 64(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, (%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 224(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 160(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 96(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 32(%rdx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 128(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 64(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, (%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 192(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 224(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 160(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 96(%rcx)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 32(%rcx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm15, (%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 64(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 128(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 192(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 224(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 160(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 96(%r8)
; AVX2-ONLY-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-ONLY-NEXT:    vmovaps %ymm5, 32(%r8)
; AVX2-ONLY-NEXT:    vmovdqa %ymm6, 224(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm8, 192(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm3, 160(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, 128(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm4, 96(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm1, 64(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm13, 32(%r9)
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, (%r9)
; AVX2-ONLY-NEXT:    addq $2120, %rsp # imm = 0x848
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-LABEL: load_i32_stride5_vf64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    subq $584, %rsp # imm = 0x248
; AVX512F-NEXT:    vmovdqa64 1152(%rdi), %zmm11
; AVX512F-NEXT:    vmovdqa64 1088(%rdi), %zmm0
; AVX512F-NEXT:    vmovdqa64 832(%rdi), %zmm10
; AVX512F-NEXT:    vmovdqa64 768(%rdi), %zmm1
; AVX512F-NEXT:    vmovdqa64 384(%rdi), %zmm15
; AVX512F-NEXT:    vmovdqa64 512(%rdi), %zmm17
; AVX512F-NEXT:    vmovdqa64 448(%rdi), %zmm2
; AVX512F-NEXT:    vmovdqa64 (%rdi), %zmm20
; AVX512F-NEXT:    vmovdqa64 64(%rdi), %zmm13
; AVX512F-NEXT:    vmovdqa64 128(%rdi), %zmm3
; AVX512F-NEXT:    vmovdqa64 192(%rdi), %zmm21
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm6 = [8,13,18,23,28,0,0,3,8,13,18,23,28,0,0,3]
; AVX512F-NEXT:    # zmm6 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vmovdqa64 %zmm3, %zmm24
; AVX512F-NEXT:    vmovdqa64 %zmm3, %zmm4
; AVX512F-NEXT:    vpermt2d %zmm21, %zmm6, %zmm24
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm19 = <0,5,10,15,20,25,30,u>
; AVX512F-NEXT:    vmovdqa64 %zmm20, %zmm3
; AVX512F-NEXT:    vpermt2d %zmm13, %zmm19, %zmm3
; AVX512F-NEXT:    vmovdqa64 %zmm2, %zmm5
; AVX512F-NEXT:    vpermt2d %zmm17, %zmm6, %zmm5
; AVX512F-NEXT:    vmovdqu64 %zmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa64 %zmm1, %zmm5
; AVX512F-NEXT:    vpermt2d %zmm10, %zmm6, %zmm5
; AVX512F-NEXT:    vmovdqu64 %zmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vpermi2d %zmm11, %zmm0, %zmm6
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm5 = [9,14,19,24,29,0,0,4,9,14,19,24,29,0,0,4]
; AVX512F-NEXT:    # zmm5 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vmovdqa64 %zmm4, %zmm7
; AVX512F-NEXT:    vpermt2d %zmm21, %zmm5, %zmm7
; AVX512F-NEXT:    vmovdqu64 %zmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm18 = <17,22,27,0,5,10,15,u>
; AVX512F-NEXT:    vmovdqa64 %zmm13, %zmm27
; AVX512F-NEXT:    vpermt2d %zmm20, %zmm18, %zmm27
; AVX512F-NEXT:    vmovdqa64 %zmm1, %zmm7
; AVX512F-NEXT:    vpermt2d %zmm10, %zmm5, %zmm7
; AVX512F-NEXT:    vmovdqu64 %zmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa64 %zmm2, %zmm7
; AVX512F-NEXT:    vpermt2d %zmm17, %zmm5, %zmm7
; AVX512F-NEXT:    vmovdqu64 %zmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vpermi2d %zmm11, %zmm0, %zmm5
; AVX512F-NEXT:    vmovdqu64 %zmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm26 = <2,7,12,17,22,27,u,u>
; AVX512F-NEXT:    vmovdqa64 %zmm20, %zmm5
; AVX512F-NEXT:    vpermt2d %zmm13, %zmm26, %zmm5
; AVX512F-NEXT:    vmovdqu64 %zmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm22 = [10,15,20,25,30,0,0,5,10,15,20,25,30,0,0,5]
; AVX512F-NEXT:    # zmm22 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vmovdqa64 %zmm4, %zmm23
; AVX512F-NEXT:    vpermt2d %zmm21, %zmm22, %zmm23
; AVX512F-NEXT:    vmovdqa64 %zmm1, %zmm25
; AVX512F-NEXT:    vpermt2d %zmm10, %zmm22, %zmm25
; AVX512F-NEXT:    vmovdqa64 %zmm2, %zmm28
; AVX512F-NEXT:    vpermt2d %zmm17, %zmm22, %zmm28
; AVX512F-NEXT:    vpermi2d %zmm11, %zmm0, %zmm22
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm29 = [27,0,5,10,15,0,17,22,27,0,5,10,15,0,17,22]
; AVX512F-NEXT:    # zmm29 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vmovdqa64 %zmm21, %zmm30
; AVX512F-NEXT:    vpermt2d %zmm4, %zmm29, %zmm30
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm31 = [12,17,22,27,0,0,2,7,12,17,22,27,0,0,2,7]
; AVX512F-NEXT:    # zmm31 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpermt2d %zmm21, %zmm31, %zmm4
; AVX512F-NEXT:    vmovdqu64 %zmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa64 %zmm17, %zmm21
; AVX512F-NEXT:    vpermt2d %zmm2, %zmm29, %zmm21
; AVX512F-NEXT:    vpermt2d %zmm17, %zmm31, %zmm2
; AVX512F-NEXT:    vmovdqu64 %zmm2, (%rsp) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa64 %zmm10, %zmm17
; AVX512F-NEXT:    vpermt2d %zmm1, %zmm29, %zmm17
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm11, %zmm29
; AVX512F-NEXT:    vpermt2d %zmm11, %zmm31, %zmm0
; AVX512F-NEXT:    vmovdqu64 %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm11 = <3,8,13,18,23,28,u,u>
; AVX512F-NEXT:    vpermt2d %zmm10, %zmm31, %zmm1
; AVX512F-NEXT:    vmovdqu64 %zmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512F-NEXT:    vmovdqa64 %zmm20, %zmm16
; AVX512F-NEXT:    vpermt2d %zmm13, %zmm11, %zmm16
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} ymm31 = <4,9,14,19,24,29,u,u>
; AVX512F-NEXT:    vpermt2d %zmm13, %zmm31, %zmm20
; AVX512F-NEXT:    vmovdqa64 320(%rdi), %zmm13
; AVX512F-NEXT:    vmovdqa64 %zmm13, %zmm7
; AVX512F-NEXT:    vpermt2d %zmm15, %zmm19, %zmm7
; AVX512F-NEXT:    vmovdqa64 %zmm15, %zmm9
; AVX512F-NEXT:    vpermt2d %zmm13, %zmm18, %zmm9
; AVX512F-NEXT:    vmovdqa64 %zmm13, %zmm12
; AVX512F-NEXT:    vpermt2d %zmm15, %zmm26, %zmm12
; AVX512F-NEXT:    vmovdqa64 %zmm13, %zmm14
; AVX512F-NEXT:    vpermt2d %zmm15, %zmm11, %zmm14
; AVX512F-NEXT:    vpermt2d %zmm15, %zmm31, %zmm13
; AVX512F-NEXT:    vmovdqa64 704(%rdi), %zmm15
; AVX512F-NEXT:    vmovdqa64 640(%rdi), %zmm2
; AVX512F-NEXT:    vmovdqa64 %zmm2, %zmm4
; AVX512F-NEXT:    vpermt2d %zmm15, %zmm19, %zmm4
; AVX512F-NEXT:    vmovdqa64 1024(%rdi), %zmm0
; AVX512F-NEXT:    vmovdqa64 960(%rdi), %zmm1
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm1, %zmm19
; AVX512F-NEXT:    vmovdqa64 %zmm15, %zmm5
; AVX512F-NEXT:    vpermt2d %zmm2, %zmm18, %zmm5
; AVX512F-NEXT:    vpermi2d %zmm1, %zmm0, %zmm18
; AVX512F-NEXT:    vmovdqa64 %zmm2, %zmm8
; AVX512F-NEXT:    vpermt2d %zmm15, %zmm26, %zmm8
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm1, %zmm26
; AVX512F-NEXT:    vmovdqa64 %zmm2, %zmm10
; AVX512F-NEXT:    vpermt2d %zmm15, %zmm11, %zmm10
; AVX512F-NEXT:    vpermi2d %zmm0, %zmm1, %zmm11
; AVX512F-NEXT:    vpermt2d %zmm0, %zmm31, %zmm1
; AVX512F-NEXT:    vpermt2d %zmm15, %zmm31, %zmm2
; AVX512F-NEXT:    movw $8064, %ax # imm = 0x1F80
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqa64 %zmm3, %zmm0
; AVX512F-NEXT:    vmovdqa32 %zmm24, %zmm0 {%k1}
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm3 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa32 %zmm3, %zmm7 {%k1}
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm3 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa32 %zmm3, %zmm4 {%k1}
; AVX512F-NEXT:    vmovdqa32 %zmm6, %zmm19 {%k1}
; AVX512F-NEXT:    vmovdqa64 256(%rdi), %zmm3
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm15 = [0,1,2,3,4,5,6,7,8,9,10,11,12,17,22,27]
; AVX512F-NEXT:    vpermt2d %zmm3, %zmm15, %zmm0
; AVX512F-NEXT:    vmovdqa64 %zmm0, %zmm24
; AVX512F-NEXT:    vmovdqa64 576(%rdi), %zmm31
; AVX512F-NEXT:    vpermt2d %zmm31, %zmm15, %zmm7
; AVX512F-NEXT:    vmovdqa64 896(%rdi), %zmm6
; AVX512F-NEXT:    vpermt2d %zmm6, %zmm15, %zmm4
; AVX512F-NEXT:    vmovdqa64 1216(%rdi), %zmm0
; AVX512F-NEXT:    vpermt2d %zmm0, %zmm15, %zmm19
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa32 %zmm15, %zmm27 {%k1}
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa32 %zmm15, %zmm5 {%k1}
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa32 %zmm15, %zmm9 {%k1}
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa32 %zmm15, %zmm18 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm15 = [0,1,2,3,4,5,6,7,8,9,10,11,12,18,23,28]
; AVX512F-NEXT:    vpermt2d %zmm3, %zmm15, %zmm27
; AVX512F-NEXT:    vpermt2d %zmm6, %zmm15, %zmm5
; AVX512F-NEXT:    vpermt2d %zmm31, %zmm15, %zmm9
; AVX512F-NEXT:    vpermt2d %zmm0, %zmm15, %zmm18
; AVX512F-NEXT:    movb $7, %al
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa64 %zmm15, %zmm23 {%k1}
; AVX512F-NEXT:    vmovdqa64 %zmm8, %zmm25 {%k1}
; AVX512F-NEXT:    vmovdqa64 %zmm12, %zmm28 {%k1}
; AVX512F-NEXT:    vmovdqa64 %zmm26, %zmm22 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm8 = [0,1,2,3,4,5,6,7,8,9,10,11,12,19,24,29]
; AVX512F-NEXT:    vpermt2d %zmm3, %zmm8, %zmm23
; AVX512F-NEXT:    vpermt2d %zmm6, %zmm8, %zmm25
; AVX512F-NEXT:    vpermt2d %zmm31, %zmm8, %zmm28
; AVX512F-NEXT:    vpermt2d %zmm0, %zmm8, %zmm22
; AVX512F-NEXT:    vmovdqa64 %zmm16, %zmm30 {%k1}
; AVX512F-NEXT:    vmovdqa64 %zmm10, %zmm17 {%k1}
; AVX512F-NEXT:    vmovdqa64 %zmm14, %zmm21 {%k1}
; AVX512F-NEXT:    vmovdqa64 %zmm11, %zmm29 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm8 = [0,1,2,3,4,5,6,7,8,9,10,11,12,20,25,30]
; AVX512F-NEXT:    vpermt2d %zmm3, %zmm8, %zmm30
; AVX512F-NEXT:    vpermt2d %zmm6, %zmm8, %zmm17
; AVX512F-NEXT:    vpermt2d %zmm31, %zmm8, %zmm21
; AVX512F-NEXT:    vpermt2d %zmm0, %zmm8, %zmm29
; AVX512F-NEXT:    movb $56, %al
; AVX512F-NEXT:    kmovw %eax, %k1
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm8 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa64 %zmm8, %zmm20 {%k1}
; AVX512F-NEXT:    vmovdqa64 {{.*#+}} zmm8 = [0,1,2,3,4,5,6,7,8,9,10,11,16,21,26,31]
; AVX512F-NEXT:    vpermt2d %zmm3, %zmm8, %zmm20
; AVX512F-NEXT:    vmovdqu64 (%rsp), %zmm3 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa64 %zmm3, %zmm13 {%k1}
; AVX512F-NEXT:    vpermt2d %zmm31, %zmm8, %zmm13
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm3 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa64 %zmm3, %zmm1 {%k1}
; AVX512F-NEXT:    vpermt2d %zmm0, %zmm8, %zmm1
; AVX512F-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
; AVX512F-NEXT:    vmovdqa64 %zmm0, %zmm2 {%k1}
; AVX512F-NEXT:    vpermt2d %zmm6, %zmm8, %zmm2
; AVX512F-NEXT:    vmovdqa64 %zmm19, 192(%rsi)
; AVX512F-NEXT:    vmovdqa64 %zmm4, 128(%rsi)
; AVX512F-NEXT:    vmovdqa64 %zmm7, 64(%rsi)
; AVX512F-NEXT:    vmovdqa64 %zmm24, (%rsi)
; AVX512F-NEXT:    vmovdqa64 %zmm18, 192(%rdx)
; AVX512F-NEXT:    vmovdqa64 %zmm27, (%rdx)
; AVX512F-NEXT:    vmovdqa64 %zmm9, 64(%rdx)
; AVX512F-NEXT:    vmovdqa64 %zmm5, 128(%rdx)
; AVX512F-NEXT:    vmovdqa64 %zmm22, 192(%rcx)
; AVX512F-NEXT:    vmovdqa64 %zmm23, (%rcx)
; AVX512F-NEXT:    vmovdqa64 %zmm28, 64(%rcx)
; AVX512F-NEXT:    vmovdqa64 %zmm25, 128(%rcx)
; AVX512F-NEXT:    vmovdqa64 %zmm29, 192(%r8)
; AVX512F-NEXT:    vmovdqa64 %zmm30, (%r8)
; AVX512F-NEXT:    vmovdqa64 %zmm21, 64(%r8)
; AVX512F-NEXT:    vmovdqa64 %zmm17, 128(%r8)
; AVX512F-NEXT:    vmovdqa64 %zmm2, 128(%r9)
; AVX512F-NEXT:    vmovdqa64 %zmm1, 192(%r9)
; AVX512F-NEXT:    vmovdqa64 %zmm20, (%r9)
; AVX512F-NEXT:    vmovdqa64 %zmm13, 64(%r9)
; AVX512F-NEXT:    addq $584, %rsp # imm = 0x248
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i32_stride5_vf64:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    subq $584, %rsp # imm = 0x248
; AVX512BW-NEXT:    vmovdqa64 1152(%rdi), %zmm11
; AVX512BW-NEXT:    vmovdqa64 1088(%rdi), %zmm0
; AVX512BW-NEXT:    vmovdqa64 832(%rdi), %zmm10
; AVX512BW-NEXT:    vmovdqa64 768(%rdi), %zmm1
; AVX512BW-NEXT:    vmovdqa64 384(%rdi), %zmm15
; AVX512BW-NEXT:    vmovdqa64 512(%rdi), %zmm17
; AVX512BW-NEXT:    vmovdqa64 448(%rdi), %zmm2
; AVX512BW-NEXT:    vmovdqa64 (%rdi), %zmm20
; AVX512BW-NEXT:    vmovdqa64 64(%rdi), %zmm13
; AVX512BW-NEXT:    vmovdqa64 128(%rdi), %zmm3
; AVX512BW-NEXT:    vmovdqa64 192(%rdi), %zmm21
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm6 = [8,13,18,23,28,0,0,3,8,13,18,23,28,0,0,3]
; AVX512BW-NEXT:    # zmm6 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vmovdqa64 %zmm3, %zmm24
; AVX512BW-NEXT:    vmovdqa64 %zmm3, %zmm4
; AVX512BW-NEXT:    vpermt2d %zmm21, %zmm6, %zmm24
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} ymm19 = <0,5,10,15,20,25,30,u>
; AVX512BW-NEXT:    vmovdqa64 %zmm20, %zmm3
; AVX512BW-NEXT:    vpermt2d %zmm13, %zmm19, %zmm3
; AVX512BW-NEXT:    vmovdqa64 %zmm2, %zmm5
; AVX512BW-NEXT:    vpermt2d %zmm17, %zmm6, %zmm5
; AVX512BW-NEXT:    vmovdqu64 %zmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vmovdqa64 %zmm1, %zmm5
; AVX512BW-NEXT:    vpermt2d %zmm10, %zmm6, %zmm5
; AVX512BW-NEXT:    vmovdqu64 %zmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vpermi2d %zmm11, %zmm0, %zmm6
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm5 = [9,14,19,24,29,0,0,4,9,14,19,24,29,0,0,4]
; AVX512BW-NEXT:    # zmm5 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vmovdqa64 %zmm4, %zmm7
; AVX512BW-NEXT:    vpermt2d %zmm21, %zmm5, %zmm7
; AVX512BW-NEXT:    vmovdqu64 %zmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} ymm18 = <17,22,27,0,5,10,15,u>
; AVX512BW-NEXT:    vmovdqa64 %zmm13, %zmm27
; AVX512BW-NEXT:    vpermt2d %zmm20, %zmm18, %zmm27
; AVX512BW-NEXT:    vmovdqa64 %zmm1, %zmm7
; AVX512BW-NEXT:    vpermt2d %zmm10, %zmm5, %zmm7
; AVX512BW-NEXT:    vmovdqu64 %zmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vmovdqa64 %zmm2, %zmm7
; AVX512BW-NEXT:    vpermt2d %zmm17, %zmm5, %zmm7
; AVX512BW-NEXT:    vmovdqu64 %zmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vpermi2d %zmm11, %zmm0, %zmm5
; AVX512BW-NEXT:    vmovdqu64 %zmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} ymm26 = <2,7,12,17,22,27,u,u>
; AVX512BW-NEXT:    vmovdqa64 %zmm20, %zmm5
; AVX512BW-NEXT:    vpermt2d %zmm13, %zmm26, %zmm5
; AVX512BW-NEXT:    vmovdqu64 %zmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm22 = [10,15,20,25,30,0,0,5,10,15,20,25,30,0,0,5]
; AVX512BW-NEXT:    # zmm22 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vmovdqa64 %zmm4, %zmm23
; AVX512BW-NEXT:    vpermt2d %zmm21, %zmm22, %zmm23
; AVX512BW-NEXT:    vmovdqa64 %zmm1, %zmm25
; AVX512BW-NEXT:    vpermt2d %zmm10, %zmm22, %zmm25
; AVX512BW-NEXT:    vmovdqa64 %zmm2, %zmm28
; AVX512BW-NEXT:    vpermt2d %zmm17, %zmm22, %zmm28
; AVX512BW-NEXT:    vpermi2d %zmm11, %zmm0, %zmm22
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm29 = [27,0,5,10,15,0,17,22,27,0,5,10,15,0,17,22]
; AVX512BW-NEXT:    # zmm29 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vmovdqa64 %zmm21, %zmm30
; AVX512BW-NEXT:    vpermt2d %zmm4, %zmm29, %zmm30
; AVX512BW-NEXT:    vbroadcasti64x4 {{.*#+}} zmm31 = [12,17,22,27,0,0,2,7,12,17,22,27,0,0,2,7]
; AVX512BW-NEXT:    # zmm31 = mem[0,1,2,3,0,1,2,3]
; AVX512BW-NEXT:    vpermt2d %zmm21, %zmm31, %zmm4
; AVX512BW-NEXT:    vmovdqu64 %zmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vmovdqa64 %zmm17, %zmm21
; AVX512BW-NEXT:    vpermt2d %zmm2, %zmm29, %zmm21
; AVX512BW-NEXT:    vpermt2d %zmm17, %zmm31, %zmm2
; AVX512BW-NEXT:    vmovdqu64 %zmm2, (%rsp) # 64-byte Spill
; AVX512BW-NEXT:    vmovdqa64 %zmm10, %zmm17
; AVX512BW-NEXT:    vpermt2d %zmm1, %zmm29, %zmm17
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm11, %zmm29
; AVX512BW-NEXT:    vpermt2d %zmm11, %zmm31, %zmm0
; AVX512BW-NEXT:    vmovdqu64 %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm11 = <3,8,13,18,23,28,u,u>
; AVX512BW-NEXT:    vpermt2d %zmm10, %zmm31, %zmm1
; AVX512BW-NEXT:    vmovdqu64 %zmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
; AVX512BW-NEXT:    vmovdqa64 %zmm20, %zmm16
; AVX512BW-NEXT:    vpermt2d %zmm13, %zmm11, %zmm16
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} ymm31 = <4,9,14,19,24,29,u,u>
; AVX512BW-NEXT:    vpermt2d %zmm13, %zmm31, %zmm20
; AVX512BW-NEXT:    vmovdqa64 320(%rdi), %zmm13
; AVX512BW-NEXT:    vmovdqa64 %zmm13, %zmm7
; AVX512BW-NEXT:    vpermt2d %zmm15, %zmm19, %zmm7
; AVX512BW-NEXT:    vmovdqa64 %zmm15, %zmm9
; AVX512BW-NEXT:    vpermt2d %zmm13, %zmm18, %zmm9
; AVX512BW-NEXT:    vmovdqa64 %zmm13, %zmm12
; AVX512BW-NEXT:    vpermt2d %zmm15, %zmm26, %zmm12
; AVX512BW-NEXT:    vmovdqa64 %zmm13, %zmm14
; AVX512BW-NEXT:    vpermt2d %zmm15, %zmm11, %zmm14
; AVX512BW-NEXT:    vpermt2d %zmm15, %zmm31, %zmm13
; AVX512BW-NEXT:    vmovdqa64 704(%rdi), %zmm15
; AVX512BW-NEXT:    vmovdqa64 640(%rdi), %zmm2
; AVX512BW-NEXT:    vmovdqa64 %zmm2, %zmm4
; AVX512BW-NEXT:    vpermt2d %zmm15, %zmm19, %zmm4
; AVX512BW-NEXT:    vmovdqa64 1024(%rdi), %zmm0
; AVX512BW-NEXT:    vmovdqa64 960(%rdi), %zmm1
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm1, %zmm19
; AVX512BW-NEXT:    vmovdqa64 %zmm15, %zmm5
; AVX512BW-NEXT:    vpermt2d %zmm2, %zmm18, %zmm5
; AVX512BW-NEXT:    vpermi2d %zmm1, %zmm0, %zmm18
; AVX512BW-NEXT:    vmovdqa64 %zmm2, %zmm8
; AVX512BW-NEXT:    vpermt2d %zmm15, %zmm26, %zmm8
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm1, %zmm26
; AVX512BW-NEXT:    vmovdqa64 %zmm2, %zmm10
; AVX512BW-NEXT:    vpermt2d %zmm15, %zmm11, %zmm10
; AVX512BW-NEXT:    vpermi2d %zmm0, %zmm1, %zmm11
; AVX512BW-NEXT:    vpermt2d %zmm0, %zmm31, %zmm1
; AVX512BW-NEXT:    vpermt2d %zmm15, %zmm31, %zmm2
; AVX512BW-NEXT:    movw $8064, %ax # imm = 0x1F80
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqa64 %zmm3, %zmm0
; AVX512BW-NEXT:    vmovdqa32 %zmm24, %zmm0 {%k1}
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm3 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa32 %zmm3, %zmm7 {%k1}
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm3 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa32 %zmm3, %zmm4 {%k1}
; AVX512BW-NEXT:    vmovdqa32 %zmm6, %zmm19 {%k1}
; AVX512BW-NEXT:    vmovdqa64 256(%rdi), %zmm3
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm15 = [0,1,2,3,4,5,6,7,8,9,10,11,12,17,22,27]
; AVX512BW-NEXT:    vpermt2d %zmm3, %zmm15, %zmm0
; AVX512BW-NEXT:    vmovdqa64 %zmm0, %zmm24
; AVX512BW-NEXT:    vmovdqa64 576(%rdi), %zmm31
; AVX512BW-NEXT:    vpermt2d %zmm31, %zmm15, %zmm7
; AVX512BW-NEXT:    vmovdqa64 896(%rdi), %zmm6
; AVX512BW-NEXT:    vpermt2d %zmm6, %zmm15, %zmm4
; AVX512BW-NEXT:    vmovdqa64 1216(%rdi), %zmm0
; AVX512BW-NEXT:    vpermt2d %zmm0, %zmm15, %zmm19
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa32 %zmm15, %zmm27 {%k1}
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa32 %zmm15, %zmm5 {%k1}
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa32 %zmm15, %zmm9 {%k1}
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa32 %zmm15, %zmm18 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm15 = [0,1,2,3,4,5,6,7,8,9,10,11,12,18,23,28]
; AVX512BW-NEXT:    vpermt2d %zmm3, %zmm15, %zmm27
; AVX512BW-NEXT:    vpermt2d %zmm6, %zmm15, %zmm5
; AVX512BW-NEXT:    vpermt2d %zmm31, %zmm15, %zmm9
; AVX512BW-NEXT:    vpermt2d %zmm0, %zmm15, %zmm18
; AVX512BW-NEXT:    movb $7, %al
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm15 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa64 %zmm15, %zmm23 {%k1}
; AVX512BW-NEXT:    vmovdqa64 %zmm8, %zmm25 {%k1}
; AVX512BW-NEXT:    vmovdqa64 %zmm12, %zmm28 {%k1}
; AVX512BW-NEXT:    vmovdqa64 %zmm26, %zmm22 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm8 = [0,1,2,3,4,5,6,7,8,9,10,11,12,19,24,29]
; AVX512BW-NEXT:    vpermt2d %zmm3, %zmm8, %zmm23
; AVX512BW-NEXT:    vpermt2d %zmm6, %zmm8, %zmm25
; AVX512BW-NEXT:    vpermt2d %zmm31, %zmm8, %zmm28
; AVX512BW-NEXT:    vpermt2d %zmm0, %zmm8, %zmm22
; AVX512BW-NEXT:    vmovdqa64 %zmm16, %zmm30 {%k1}
; AVX512BW-NEXT:    vmovdqa64 %zmm10, %zmm17 {%k1}
; AVX512BW-NEXT:    vmovdqa64 %zmm14, %zmm21 {%k1}
; AVX512BW-NEXT:    vmovdqa64 %zmm11, %zmm29 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm8 = [0,1,2,3,4,5,6,7,8,9,10,11,12,20,25,30]
; AVX512BW-NEXT:    vpermt2d %zmm3, %zmm8, %zmm30
; AVX512BW-NEXT:    vpermt2d %zmm6, %zmm8, %zmm17
; AVX512BW-NEXT:    vpermt2d %zmm31, %zmm8, %zmm21
; AVX512BW-NEXT:    vpermt2d %zmm0, %zmm8, %zmm29
; AVX512BW-NEXT:    movb $56, %al
; AVX512BW-NEXT:    kmovd %eax, %k1
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm8 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa64 %zmm8, %zmm20 {%k1}
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm8 = [0,1,2,3,4,5,6,7,8,9,10,11,16,21,26,31]
; AVX512BW-NEXT:    vpermt2d %zmm3, %zmm8, %zmm20
; AVX512BW-NEXT:    vmovdqu64 (%rsp), %zmm3 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa64 %zmm3, %zmm13 {%k1}
; AVX512BW-NEXT:    vpermt2d %zmm31, %zmm8, %zmm13
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm3 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa64 %zmm3, %zmm1 {%k1}
; AVX512BW-NEXT:    vpermt2d %zmm0, %zmm8, %zmm1
; AVX512BW-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
; AVX512BW-NEXT:    vmovdqa64 %zmm0, %zmm2 {%k1}
; AVX512BW-NEXT:    vpermt2d %zmm6, %zmm8, %zmm2
; AVX512BW-NEXT:    vmovdqa64 %zmm19, 192(%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm4, 128(%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm7, 64(%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm24, (%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm18, 192(%rdx)
; AVX512BW-NEXT:    vmovdqa64 %zmm27, (%rdx)
; AVX512BW-NEXT:    vmovdqa64 %zmm9, 64(%rdx)
; AVX512BW-NEXT:    vmovdqa64 %zmm5, 128(%rdx)
; AVX512BW-NEXT:    vmovdqa64 %zmm22, 192(%rcx)
; AVX512BW-NEXT:    vmovdqa64 %zmm23, (%rcx)
; AVX512BW-NEXT:    vmovdqa64 %zmm28, 64(%rcx)
; AVX512BW-NEXT:    vmovdqa64 %zmm25, 128(%rcx)
; AVX512BW-NEXT:    vmovdqa64 %zmm29, 192(%r8)
; AVX512BW-NEXT:    vmovdqa64 %zmm30, (%r8)
; AVX512BW-NEXT:    vmovdqa64 %zmm21, 64(%r8)
; AVX512BW-NEXT:    vmovdqa64 %zmm17, 128(%r8)
; AVX512BW-NEXT:    vmovdqa64 %zmm2, 128(%r9)
; AVX512BW-NEXT:    vmovdqa64 %zmm1, 192(%r9)
; AVX512BW-NEXT:    vmovdqa64 %zmm20, (%r9)
; AVX512BW-NEXT:    vmovdqa64 %zmm13, 64(%r9)
; AVX512BW-NEXT:    addq $584, %rsp # imm = 0x248
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <320 x i32>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <320 x i32> %wide.vec, <320 x i32> poison, <64 x i32> <i32 0, i32 5, i32 10, i32 15, i32 20, i32 25, i32 30, i32 35, i32 40, i32 45, i32 50, i32 55, i32 60, i32 65, i32 70, i32 75, i32 80, i32 85, i32 90, i32 95, i32 100, i32 105, i32 110, i32 115, i32 120, i32 125, i32 130, i32 135, i32 140, i32 145, i32 150, i32 155, i32 160, i32 165, i32 170, i32 175, i32 180, i32 185, i32 190, i32 195, i32 200, i32 205, i32 210, i32 215, i32 220, i32 225, i32 230, i32 235, i32 240, i32 245, i32 250, i32 255, i32 260, i32 265, i32 270, i32 275, i32 280, i32 285, i32 290, i32 295, i32 300, i32 305, i32 310, i32 315>
  %strided.vec1 = shufflevector <320 x i32> %wide.vec, <320 x i32> poison, <64 x i32> <i32 1, i32 6, i32 11, i32 16, i32 21, i32 26, i32 31, i32 36, i32 41, i32 46, i32 51, i32 56, i32 61, i32 66, i32 71, i32 76, i32 81, i32 86, i32 91, i32 96, i32 101, i32 106, i32 111, i32 116, i32 121, i32 126, i32 131, i32 136, i32 141, i32 146, i32 151, i32 156, i32 161, i32 166, i32 171, i32 176, i32 181, i32 186, i32 191, i32 196, i32 201, i32 206, i32 211, i32 216, i32 221, i32 226, i32 231, i32 236, i32 241, i32 246, i32 251, i32 256, i32 261, i32 266, i32 271, i32 276, i32 281, i32 286, i32 291, i32 296, i32 301, i32 306, i32 311, i32 316>
  %strided.vec2 = shufflevector <320 x i32> %wide.vec, <320 x i32> poison, <64 x i32> <i32 2, i32 7, i32 12, i32 17, i32 22, i32 27, i32 32, i32 37, i32 42, i32 47, i32 52, i32 57, i32 62, i32 67, i32 72, i32 77, i32 82, i32 87, i32 92, i32 97, i32 102, i32 107, i32 112, i32 117, i32 122, i32 127, i32 132, i32 137, i32 142, i32 147, i32 152, i32 157, i32 162, i32 167, i32 172, i32 177, i32 182, i32 187, i32 192, i32 197, i32 202, i32 207, i32 212, i32 217, i32 222, i32 227, i32 232, i32 237, i32 242, i32 247, i32 252, i32 257, i32 262, i32 267, i32 272, i32 277, i32 282, i32 287, i32 292, i32 297, i32 302, i32 307, i32 312, i32 317>
  %strided.vec3 = shufflevector <320 x i32> %wide.vec, <320 x i32> poison, <64 x i32> <i32 3, i32 8, i32 13, i32 18, i32 23, i32 28, i32 33, i32 38, i32 43, i32 48, i32 53, i32 58, i32 63, i32 68, i32 73, i32 78, i32 83, i32 88, i32 93, i32 98, i32 103, i32 108, i32 113, i32 118, i32 123, i32 128, i32 133, i32 138, i32 143, i32 148, i32 153, i32 158, i32 163, i32 168, i32 173, i32 178, i32 183, i32 188, i32 193, i32 198, i32 203, i32 208, i32 213, i32 218, i32 223, i32 228, i32 233, i32 238, i32 243, i32 248, i32 253, i32 258, i32 263, i32 268, i32 273, i32 278, i32 283, i32 288, i32 293, i32 298, i32 303, i32 308, i32 313, i32 318>
  %strided.vec4 = shufflevector <320 x i32> %wide.vec, <320 x i32> poison, <64 x i32> <i32 4, i32 9, i32 14, i32 19, i32 24, i32 29, i32 34, i32 39, i32 44, i32 49, i32 54, i32 59, i32 64, i32 69, i32 74, i32 79, i32 84, i32 89, i32 94, i32 99, i32 104, i32 109, i32 114, i32 119, i32 124, i32 129, i32 134, i32 139, i32 144, i32 149, i32 154, i32 159, i32 164, i32 169, i32 174, i32 179, i32 184, i32 189, i32 194, i32 199, i32 204, i32 209, i32 214, i32 219, i32 224, i32 229, i32 234, i32 239, i32 244, i32 249, i32 254, i32 259, i32 264, i32 269, i32 274, i32 279, i32 284, i32 289, i32 294, i32 299, i32 304, i32 309, i32 314, i32 319>
  store <64 x i32> %strided.vec0, ptr %out.vec0, align 64
  store <64 x i32> %strided.vec1, ptr %out.vec1, align 64
  store <64 x i32> %strided.vec2, ptr %out.vec2, align 64
  store <64 x i32> %strided.vec3, ptr %out.vec3, align 64
  store <64 x i32> %strided.vec4, ptr %out.vec4, align 64
  ret void
}
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; AVX: {{.*}}
; AVX1: {{.*}}
; AVX2: {{.*}}
; AVX2-FAST: {{.*}}
; AVX2-FAST-PERLANE: {{.*}}
; AVX2-SLOW: {{.*}}
; AVX512BW-ONLY-FAST: {{.*}}
; AVX512BW-ONLY-SLOW: {{.*}}
; AVX512DQ-FAST: {{.*}}
; AVX512DQ-SLOW: {{.*}}
; AVX512DQBW-FAST: {{.*}}
; AVX512DQBW-SLOW: {{.*}}
; AVX512F-ONLY-FAST: {{.*}}
; AVX512F-ONLY-SLOW: {{.*}}
; FALLBACK0: {{.*}}
; FALLBACK1: {{.*}}
; FALLBACK10: {{.*}}
; FALLBACK11: {{.*}}
; FALLBACK12: {{.*}}
; FALLBACK2: {{.*}}
; FALLBACK3: {{.*}}
; FALLBACK4: {{.*}}
; FALLBACK5: {{.*}}
; FALLBACK6: {{.*}}
; FALLBACK7: {{.*}}
; FALLBACK8: {{.*}}
; FALLBACK9: {{.*}}