; 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
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx  | FileCheck %s --check-prefixes=AVX1
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512bw,+avx512vl | FileCheck %s --check-prefixes=AVX512

; These patterns are produced by LoopVectorizer for interleaved loads.

define void @load_i32_stride4_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3) nounwind {
; SSE-LABEL: load_i32_stride4_vf2:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa (%rdi), %xmm0
; SSE-NEXT:    movdqa 16(%rdi), %xmm1
; SSE-NEXT:    movdqa %xmm0, %xmm2
; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[2,3,2,3]
; SSE-NEXT:    punpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,2,3]
; SSE-NEXT:    movq %xmm2, (%rsi)
; SSE-NEXT:    movq %xmm3, (%rdx)
; SSE-NEXT:    movq %xmm0, (%rcx)
; SSE-NEXT:    movq %xmm1, (%r8)
; SSE-NEXT:    retq
;
; AVX1-LABEL: load_i32_stride4_vf2:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovdqa (%rdi), %xmm0
; AVX1-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX1-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
; AVX1-NEXT:    vpblendw {{.*#+}} xmm3 = xmm3[0,1],xmm1[2,3],xmm3[4,5,6,7]
; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT:    vmovq %xmm2, (%rsi)
; AVX1-NEXT:    vmovq %xmm3, (%rdx)
; AVX1-NEXT:    vmovq %xmm0, (%rcx)
; AVX1-NEXT:    vpextrq $1, %xmm0, (%r8)
; AVX1-NEXT:    retq
;
; AVX2-LABEL: load_i32_stride4_vf2:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vmovdqa (%rdi), %xmm0
; AVX2-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; AVX2-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
; AVX2-NEXT:    vpblendd {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2,3]
; AVX2-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX2-NEXT:    vmovq %xmm2, (%rsi)
; AVX2-NEXT:    vmovq %xmm3, (%rdx)
; AVX2-NEXT:    vmovq %xmm0, (%rcx)
; AVX2-NEXT:    vpextrq $1, %xmm0, (%r8)
; AVX2-NEXT:    retq
;
; AVX512-LABEL: load_i32_stride4_vf2:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512-NEXT:    vmovdqa 16(%rdi), %xmm1
; AVX512-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; AVX512-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
; AVX512-NEXT:    vpblendd {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2,3]
; AVX512-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX512-NEXT:    vmovq %xmm2, (%rsi)
; AVX512-NEXT:    vmovq %xmm3, (%rdx)
; AVX512-NEXT:    vmovq %xmm0, (%rcx)
; AVX512-NEXT:    vpextrq $1, %xmm0, (%r8)
; AVX512-NEXT:    retq
  %wide.vec = load <8 x i32>, ptr %in.vec, align 32

  %strided.vec0 = shufflevector <8 x i32> %wide.vec, <8 x i32> poison, <2 x i32> <i32 0, i32 4>
  %strided.vec1 = shufflevector <8 x i32> %wide.vec, <8 x i32> poison, <2 x i32> <i32 1, i32 5>
  %strided.vec2 = shufflevector <8 x i32> %wide.vec, <8 x i32> poison, <2 x i32> <i32 2, i32 6>
  %strided.vec3 = shufflevector <8 x i32> %wide.vec, <8 x i32> poison, <2 x i32> <i32 3, i32 7>

  store <2 x i32> %strided.vec0, ptr %out.vec0, align 32
  store <2 x i32> %strided.vec1, ptr %out.vec1, align 32
  store <2 x i32> %strided.vec2, ptr %out.vec2, align 32
  store <2 x i32> %strided.vec3, ptr %out.vec3, align 32

  ret void
}

define void @load_i32_stride4_vf4(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3) nounwind {
; SSE-LABEL: load_i32_stride4_vf4:
; SSE:       # %bb.0:
; SSE-NEXT:    movaps (%rdi), %xmm0
; SSE-NEXT:    movaps 16(%rdi), %xmm1
; SSE-NEXT:    movaps 32(%rdi), %xmm2
; SSE-NEXT:    movaps 48(%rdi), %xmm3
; SSE-NEXT:    movaps %xmm2, %xmm4
; SSE-NEXT:    unpcklps {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; SSE-NEXT:    movaps %xmm0, %xmm5
; SSE-NEXT:    unpcklps {{.*#+}} xmm5 = xmm5[0],xmm1[0],xmm5[1],xmm1[1]
; SSE-NEXT:    movaps %xmm5, %xmm6
; SSE-NEXT:    movlhps {{.*#+}} xmm6 = xmm6[0],xmm4[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm5 = xmm5[1],xmm4[1]
; SSE-NEXT:    unpckhps {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; SSE-NEXT:    unpckhps {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    movaps %xmm0, %xmm1
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm2[1]
; SSE-NEXT:    movaps %xmm6, (%rsi)
; SSE-NEXT:    movaps %xmm5, (%rdx)
; SSE-NEXT:    movaps %xmm1, (%rcx)
; SSE-NEXT:    movaps %xmm0, (%r8)
; SSE-NEXT:    retq
;
; AVX1-LABEL: load_i32_stride4_vf4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovaps 32(%rdi), %xmm0
; AVX1-NEXT:    vmovaps 48(%rdi), %xmm1
; AVX1-NEXT:    vmovlhps {{.*#+}} xmm2 = xmm1[0],xmm0[0]
; AVX1-NEXT:    vmovaps (%rdi), %xmm3
; AVX1-NEXT:    vmovaps 16(%rdi), %xmm4
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm5 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
; AVX1-NEXT:    vshufps {{.*#+}} xmm2 = xmm5[0,1],xmm2[2,0]
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm5 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; AVX1-NEXT:    vinsertps {{.*#+}} xmm6 = xmm3[1],xmm4[1],zero,zero
; AVX1-NEXT:    vblendps {{.*#+}} xmm5 = xmm6[0,1],xmm5[2,3]
; AVX1-NEXT:    vinsertps {{.*#+}} xmm6 = zero,zero,xmm0[2],xmm1[2]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm7 = xmm3[2],xmm4[2],xmm3[3],xmm4[3]
; AVX1-NEXT:    vblendps {{.*#+}} xmm6 = xmm7[0,1],xmm6[2,3]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm4[3,0],xmm3[3,0]
; AVX1-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[2,0],xmm0[2,3]
; AVX1-NEXT:    vmovaps %xmm2, (%rsi)
; AVX1-NEXT:    vmovaps %xmm5, (%rdx)
; AVX1-NEXT:    vmovaps %xmm6, (%rcx)
; AVX1-NEXT:    vmovaps %xmm0, (%r8)
; AVX1-NEXT:    retq
;
; AVX2-LABEL: load_i32_stride4_vf4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vmovaps {{.*#+}} xmm0 = <u,u,0,4>
; AVX2-NEXT:    vmovaps 32(%rdi), %ymm1
; AVX2-NEXT:    vpermps %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vmovaps (%rdi), %xmm2
; AVX2-NEXT:    vmovaps 16(%rdi), %xmm3
; AVX2-NEXT:    vmovaps 32(%rdi), %xmm4
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm5 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; AVX2-NEXT:    vblendps {{.*#+}} xmm0 = xmm5[0,1],xmm0[2,3]
; AVX2-NEXT:    vmovaps 48(%rdi), %xmm5
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm6 = xmm4[0],xmm5[0],xmm4[1],xmm5[1]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm7 = <1,5,u,u>
; AVX2-NEXT:    vmovaps (%rdi), %ymm8
; AVX2-NEXT:    vpermps %ymm8, %ymm7, %ymm7
; AVX2-NEXT:    vblendps {{.*#+}} xmm6 = xmm7[0,1],xmm6[2,3]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm7 = <u,u,2,6>
; AVX2-NEXT:    vpermps %ymm1, %ymm7, %ymm1
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; AVX2-NEXT:    vblendps {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3]
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm2 = xmm4[2],xmm5[2],xmm4[3],xmm5[3]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm3 = <3,7,u,u>
; AVX2-NEXT:    vpermps %ymm8, %ymm3, %ymm3
; AVX2-NEXT:    vblendps {{.*#+}} xmm2 = xmm3[0,1],xmm2[2,3]
; AVX2-NEXT:    vmovaps %xmm0, (%rsi)
; AVX2-NEXT:    vmovaps %xmm6, (%rdx)
; AVX2-NEXT:    vmovaps %xmm1, (%rcx)
; AVX2-NEXT:    vmovaps %xmm2, (%r8)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512-LABEL: load_i32_stride4_vf4:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm0 = [0,4,8,12]
; AVX512-NEXT:    vmovdqa (%rdi), %ymm1
; AVX512-NEXT:    vmovdqa 32(%rdi), %ymm2
; AVX512-NEXT:    vpermi2d %ymm2, %ymm1, %ymm0
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm3 = [1,5,9,13]
; AVX512-NEXT:    vpermi2d %ymm2, %ymm1, %ymm3
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm4 = [2,6,10,14]
; AVX512-NEXT:    vpermi2d %ymm2, %ymm1, %ymm4
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm5 = [3,7,11,15]
; AVX512-NEXT:    vpermi2d %ymm2, %ymm1, %ymm5
; AVX512-NEXT:    vmovdqa %xmm0, (%rsi)
; AVX512-NEXT:    vmovdqa %xmm3, (%rdx)
; AVX512-NEXT:    vmovdqa %xmm4, (%rcx)
; AVX512-NEXT:    vmovdqa %xmm5, (%r8)
; AVX512-NEXT:    vzeroupper
; AVX512-NEXT:    retq
  %wide.vec = load <16 x i32>, ptr %in.vec, align 32

  %strided.vec0 = shufflevector <16 x i32> %wide.vec, <16 x i32> poison, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
  %strided.vec1 = shufflevector <16 x i32> %wide.vec, <16 x i32> poison, <4 x i32> <i32 1, i32 5, i32 9, i32 13>
  %strided.vec2 = shufflevector <16 x i32> %wide.vec, <16 x i32> poison, <4 x i32> <i32 2, i32 6, i32 10, i32 14>
  %strided.vec3 = shufflevector <16 x i32> %wide.vec, <16 x i32> poison, <4 x i32> <i32 3, i32 7, i32 11, i32 15>

  store <4 x i32> %strided.vec0, ptr %out.vec0, align 32
  store <4 x i32> %strided.vec1, ptr %out.vec1, align 32
  store <4 x i32> %strided.vec2, ptr %out.vec2, align 32
  store <4 x i32> %strided.vec3, ptr %out.vec3, align 32

  ret void
}

define void @load_i32_stride4_vf8(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3) nounwind {
; SSE-LABEL: load_i32_stride4_vf8:
; SSE:       # %bb.0:
; SSE-NEXT:    movaps (%rdi), %xmm5
; SSE-NEXT:    movaps 16(%rdi), %xmm8
; SSE-NEXT:    movaps 32(%rdi), %xmm6
; SSE-NEXT:    movaps 48(%rdi), %xmm9
; SSE-NEXT:    movaps 80(%rdi), %xmm10
; SSE-NEXT:    movaps 64(%rdi), %xmm4
; SSE-NEXT:    movaps 112(%rdi), %xmm11
; SSE-NEXT:    movaps 96(%rdi), %xmm3
; SSE-NEXT:    movaps %xmm3, %xmm7
; SSE-NEXT:    unpcklps {{.*#+}} xmm7 = xmm7[0],xmm11[0],xmm7[1],xmm11[1]
; SSE-NEXT:    movaps %xmm4, %xmm1
; SSE-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm10[0],xmm1[1],xmm10[1]
; SSE-NEXT:    movaps %xmm1, %xmm12
; SSE-NEXT:    movlhps {{.*#+}} xmm12 = xmm12[0],xmm7[0]
; SSE-NEXT:    movaps %xmm6, %xmm2
; SSE-NEXT:    unpcklps {{.*#+}} xmm2 = xmm2[0],xmm9[0],xmm2[1],xmm9[1]
; SSE-NEXT:    movaps %xmm5, %xmm0
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm8[0],xmm0[1],xmm8[1]
; SSE-NEXT:    movaps %xmm0, %xmm13
; SSE-NEXT:    movlhps {{.*#+}} xmm13 = xmm13[0],xmm2[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm7[1]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm2[1]
; SSE-NEXT:    unpckhps {{.*#+}} xmm3 = xmm3[2],xmm11[2],xmm3[3],xmm11[3]
; SSE-NEXT:    unpckhps {{.*#+}} xmm4 = xmm4[2],xmm10[2],xmm4[3],xmm10[3]
; SSE-NEXT:    movaps %xmm4, %xmm2
; SSE-NEXT:    movlhps {{.*#+}} xmm2 = xmm2[0],xmm3[0]
; SSE-NEXT:    unpckhps {{.*#+}} xmm6 = xmm6[2],xmm9[2],xmm6[3],xmm9[3]
; SSE-NEXT:    unpckhps {{.*#+}} xmm5 = xmm5[2],xmm8[2],xmm5[3],xmm8[3]
; SSE-NEXT:    movaps %xmm5, %xmm7
; SSE-NEXT:    movlhps {{.*#+}} xmm7 = xmm7[0],xmm6[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm4 = xmm4[1],xmm3[1]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm5 = xmm5[1],xmm6[1]
; SSE-NEXT:    movaps %xmm12, 16(%rsi)
; SSE-NEXT:    movaps %xmm13, (%rsi)
; SSE-NEXT:    movaps %xmm1, 16(%rdx)
; SSE-NEXT:    movaps %xmm0, (%rdx)
; SSE-NEXT:    movaps %xmm2, 16(%rcx)
; SSE-NEXT:    movaps %xmm7, (%rcx)
; SSE-NEXT:    movaps %xmm4, 16(%r8)
; SSE-NEXT:    movaps %xmm5, (%r8)
; SSE-NEXT:    retq
;
; AVX1-LABEL: load_i32_stride4_vf8:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovaps 64(%rdi), %ymm0
; AVX1-NEXT:    vmovaps 96(%rdi), %ymm1
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm10 = ymm1[2,3,0,1]
; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm3 = ymm1[0],ymm10[0],ymm1[2],ymm10[2]
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm4 = ymm0[2,3,0,1]
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm5 = ymm4[0],ymm0[0],ymm4[1],ymm0[1],ymm4[4],ymm0[4],ymm4[5],ymm0[5]
; AVX1-NEXT:    vshufps {{.*#+}} ymm8 = ymm5[0,1],ymm3[2,0],ymm5[4,5],ymm3[6,4]
; AVX1-NEXT:    vmovaps 32(%rdi), %xmm5
; AVX1-NEXT:    vmovaps 48(%rdi), %xmm6
; AVX1-NEXT:    vmovlhps {{.*#+}} xmm9 = xmm6[0],xmm5[0]
; AVX1-NEXT:    vmovaps (%rdi), %xmm3
; AVX1-NEXT:    vmovaps 16(%rdi), %xmm7
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm2 = xmm3[0],xmm7[0],xmm3[1],xmm7[1]
; AVX1-NEXT:    vshufps {{.*#+}} xmm2 = xmm2[0,1],xmm9[2,0]
; AVX1-NEXT:    vblendps {{.*#+}} ymm11 = ymm2[0,1,2,3],ymm8[4,5,6,7]
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm8 = ymm10[0],ymm1[0],ymm10[1],ymm1[1],ymm10[4],ymm1[4],ymm10[5],ymm1[5]
; AVX1-NEXT:    vshufps {{.*#+}} ymm9 = ymm0[1,0],ymm4[1,0],ymm0[5,4],ymm4[5,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm8 = ymm9[2,0],ymm8[2,3],ymm9[6,4],ymm8[6,7]
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm9 = xmm5[0],xmm6[0],xmm5[1],xmm6[1]
; AVX1-NEXT:    vinsertps {{.*#+}} xmm2 = xmm3[1],xmm7[1],zero,zero
; AVX1-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],xmm9[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm12 = ymm2[0,1,2,3],ymm8[4,5,6,7]
; AVX1-NEXT:    vunpckhpd {{.*#+}} ymm8 = ymm1[1],ymm10[1],ymm1[3],ymm10[3]
; AVX1-NEXT:    vunpckhps {{.*#+}} ymm9 = ymm4[2],ymm0[2],ymm4[3],ymm0[3],ymm4[6],ymm0[6],ymm4[7],ymm0[7]
; AVX1-NEXT:    vshufps {{.*#+}} ymm8 = ymm9[0,1],ymm8[2,0],ymm9[4,5],ymm8[6,4]
; AVX1-NEXT:    vinsertps {{.*#+}} xmm9 = zero,zero,xmm5[2],xmm6[2]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm2 = xmm3[2],xmm7[2],xmm3[3],xmm7[3]
; AVX1-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],xmm9[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm8[4,5,6,7]
; AVX1-NEXT:    vunpckhps {{.*#+}} ymm1 = ymm10[2],ymm1[2],ymm10[3],ymm1[3],ymm10[6],ymm1[6],ymm10[7],ymm1[7]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[3,0],ymm4[3,0],ymm0[7,4],ymm4[7,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[2,0],ymm1[2,3],ymm0[6,4],ymm1[6,7]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm1 = xmm5[2],xmm6[2],xmm5[3],xmm6[3]
; AVX1-NEXT:    vshufps {{.*#+}} xmm3 = xmm7[3,0],xmm3[3,0]
; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm3[2,0],xmm1[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovaps %ymm11, (%rsi)
; AVX1-NEXT:    vmovaps %ymm12, (%rdx)
; AVX1-NEXT:    vmovaps %ymm2, (%rcx)
; AVX1-NEXT:    vmovaps %ymm0, (%r8)
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: load_i32_stride4_vf8:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vmovaps (%rdi), %ymm8
; AVX2-NEXT:    vmovaps 32(%rdi), %ymm9
; AVX2-NEXT:    vmovaps 64(%rdi), %ymm1
; AVX2-NEXT:    vmovaps 96(%rdi), %ymm2
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm3 = [17179869184,17179869184,17179869184,17179869184]
; AVX2-NEXT:    vpermps %ymm2, %ymm3, %ymm5
; AVX2-NEXT:    vpermps %ymm1, %ymm3, %ymm3
; AVX2-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm5[6,7]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm5 = <u,u,0,4>
; AVX2-NEXT:    vpermps %ymm9, %ymm5, %ymm6
; AVX2-NEXT:    vmovaps (%rdi), %xmm7
; AVX2-NEXT:    vmovaps 16(%rdi), %xmm0
; AVX2-NEXT:    vmovaps 32(%rdi), %xmm5
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm4 = xmm7[0],xmm0[0],xmm7[1],xmm0[1]
; AVX2-NEXT:    vblendps {{.*#+}} xmm4 = xmm4[0,1],xmm6[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm10 = ymm4[0,1,2,3],ymm3[4,5,6,7]
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm4 = [21474836481,21474836481,21474836481,21474836481]
; AVX2-NEXT:    vpermps %ymm2, %ymm4, %ymm6
; AVX2-NEXT:    vpermps %ymm1, %ymm4, %ymm4
; AVX2-NEXT:    vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3,4,5],ymm6[6,7]
; AVX2-NEXT:    vmovaps 48(%rdi), %xmm6
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm3 = xmm5[0],xmm6[0],xmm5[1],xmm6[1]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm11 = <1,5,u,u>
; AVX2-NEXT:    vpermps %ymm8, %ymm11, %ymm11
; AVX2-NEXT:    vblendps {{.*#+}} xmm3 = xmm11[0,1],xmm3[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm4[4,5,6,7]
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm4 = [25769803778,25769803778,25769803778,25769803778]
; AVX2-NEXT:    vpermps %ymm2, %ymm4, %ymm11
; AVX2-NEXT:    vpermps %ymm1, %ymm4, %ymm4
; AVX2-NEXT:    vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3,4,5],ymm11[6,7]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm11 = <u,u,2,6>
; AVX2-NEXT:    vpermps %ymm9, %ymm11, %ymm9
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm0 = xmm7[2],xmm0[2],xmm7[3],xmm0[3]
; AVX2-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0,1],xmm9[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm4[4,5,6,7]
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm4 = [30064771075,30064771075,30064771075,30064771075]
; AVX2-NEXT:    vpermps %ymm2, %ymm4, %ymm2
; AVX2-NEXT:    vpermps %ymm1, %ymm4, %ymm1
; AVX2-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4,5],ymm2[6,7]
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm2 = xmm5[2],xmm6[2],xmm5[3],xmm6[3]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm4 = <3,7,u,u>
; AVX2-NEXT:    vpermps %ymm8, %ymm4, %ymm4
; AVX2-NEXT:    vblendps {{.*#+}} xmm2 = xmm4[0,1],xmm2[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm1 = ymm2[0,1,2,3],ymm1[4,5,6,7]
; AVX2-NEXT:    vmovaps %ymm10, (%rsi)
; AVX2-NEXT:    vmovaps %ymm3, (%rdx)
; AVX2-NEXT:    vmovaps %ymm0, (%rcx)
; AVX2-NEXT:    vmovaps %ymm1, (%r8)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512-LABEL: load_i32_stride4_vf8:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm0 = [0,4,8,12,16,20,24,28]
; AVX512-NEXT:    vmovdqu64 (%rdi), %zmm1
; AVX512-NEXT:    vmovdqu64 64(%rdi), %zmm2
; AVX512-NEXT:    vpermi2d %zmm2, %zmm1, %zmm0
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm3 = [1,5,9,13,17,21,25,29]
; AVX512-NEXT:    vpermi2d %zmm2, %zmm1, %zmm3
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm4 = [2,6,10,14,18,22,26,30]
; AVX512-NEXT:    vpermi2d %zmm2, %zmm1, %zmm4
; AVX512-NEXT:    vmovdqa {{.*#+}} ymm5 = [3,7,11,15,19,23,27,31]
; AVX512-NEXT:    vpermi2d %zmm2, %zmm1, %zmm5
; AVX512-NEXT:    vmovdqa %ymm0, (%rsi)
; AVX512-NEXT:    vmovdqa %ymm3, (%rdx)
; AVX512-NEXT:    vmovdqa %ymm4, (%rcx)
; AVX512-NEXT:    vmovdqa %ymm5, (%r8)
; AVX512-NEXT:    vzeroupper
; AVX512-NEXT:    retq
  %wide.vec = load <32 x i32>, ptr %in.vec, align 32

  %strided.vec0 = shufflevector <32 x i32> %wide.vec, <32 x i32> poison, <8 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28>
  %strided.vec1 = shufflevector <32 x i32> %wide.vec, <32 x i32> poison, <8 x i32> <i32 1, i32 5, i32 9, i32 13, i32 17, i32 21, i32 25, i32 29>
  %strided.vec2 = shufflevector <32 x i32> %wide.vec, <32 x i32> poison, <8 x i32> <i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30>
  %strided.vec3 = shufflevector <32 x i32> %wide.vec, <32 x i32> poison, <8 x i32> <i32 3, i32 7, i32 11, i32 15, i32 19, i32 23, i32 27, i32 31>

  store <8 x i32> %strided.vec0, ptr %out.vec0, align 32
  store <8 x i32> %strided.vec1, ptr %out.vec1, align 32
  store <8 x i32> %strided.vec2, ptr %out.vec2, align 32
  store <8 x i32> %strided.vec3, ptr %out.vec3, align 32

  ret void
}

define void @load_i32_stride4_vf16(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3) nounwind {
; SSE-LABEL: load_i32_stride4_vf16:
; SSE:       # %bb.0:
; SSE-NEXT:    subq $24, %rsp
; SSE-NEXT:    movaps 208(%rdi), %xmm10
; SSE-NEXT:    movaps %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 240(%rdi), %xmm5
; SSE-NEXT:    movaps %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 224(%rdi), %xmm2
; SSE-NEXT:    movaps 80(%rdi), %xmm3
; SSE-NEXT:    movaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 64(%rdi), %xmm9
; SSE-NEXT:    movaps 112(%rdi), %xmm6
; SSE-NEXT:    movaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 96(%rdi), %xmm7
; SSE-NEXT:    movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 144(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 128(%rdi), %xmm12
; SSE-NEXT:    movaps 176(%rdi), %xmm4
; SSE-NEXT:    movaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 160(%rdi), %xmm8
; SSE-NEXT:    movaps %xmm8, %xmm0
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
; SSE-NEXT:    movaps %xmm12, %xmm11
; SSE-NEXT:    unpcklps {{.*#+}} xmm11 = xmm11[0],xmm1[0],xmm11[1],xmm1[1]
; SSE-NEXT:    movaps %xmm11, %xmm1
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm0[0]
; SSE-NEXT:    movaps %xmm1, (%rsp) # 16-byte Spill
; SSE-NEXT:    movaps %xmm7, %xmm1
; SSE-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1]
; SSE-NEXT:    movaps %xmm9, %xmm7
; SSE-NEXT:    unpcklps {{.*#+}} xmm7 = xmm7[0],xmm3[0],xmm7[1],xmm3[1]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm11 = xmm11[1],xmm0[1]
; SSE-NEXT:    movaps %xmm7, %xmm0
; SSE-NEXT:    movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm7 = xmm7[1],xmm1[1]
; SSE-NEXT:    movaps %xmm2, %xmm0
; SSE-NEXT:    movaps %xmm2, %xmm6
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1]
; SSE-NEXT:    movaps 192(%rdi), %xmm5
; SSE-NEXT:    movaps %xmm5, %xmm1
; SSE-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm10[0],xmm1[1],xmm10[1]
; SSE-NEXT:    movaps %xmm1, %xmm15
; SSE-NEXT:    movlhps {{.*#+}} xmm15 = xmm15[0],xmm0[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm0[1]
; SSE-NEXT:    movaps 32(%rdi), %xmm4
; SSE-NEXT:    movaps 48(%rdi), %xmm13
; SSE-NEXT:    movaps %xmm4, %xmm3
; SSE-NEXT:    unpcklps {{.*#+}} xmm3 = xmm3[0],xmm13[0],xmm3[1],xmm13[1]
; SSE-NEXT:    movaps (%rdi), %xmm0
; SSE-NEXT:    movaps 16(%rdi), %xmm10
; SSE-NEXT:    movaps %xmm0, %xmm2
; SSE-NEXT:    unpcklps {{.*#+}} xmm2 = xmm2[0],xmm10[0],xmm2[1],xmm10[1]
; SSE-NEXT:    movaps %xmm2, %xmm14
; SSE-NEXT:    movlhps {{.*#+}} xmm14 = xmm14[0],xmm3[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm2 = xmm2[1],xmm3[1]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
; SSE-NEXT:    # xmm8 = xmm8[2],mem[2],xmm8[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
; SSE-NEXT:    # xmm12 = xmm12[2],mem[2],xmm12[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
; SSE-NEXT:    # xmm6 = xmm6[2],mem[2],xmm6[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Folded Reload
; SSE-NEXT:    # xmm5 = xmm5[2],mem[2],xmm5[3],mem[3]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
; SSE-NEXT:    # xmm3 = xmm3[2],mem[2],xmm3[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Folded Reload
; SSE-NEXT:    # xmm9 = xmm9[2],mem[2],xmm9[3],mem[3]
; SSE-NEXT:    unpckhps {{.*#+}} xmm4 = xmm4[2],xmm13[2],xmm4[3],xmm13[3]
; SSE-NEXT:    unpckhps {{.*#+}} xmm0 = xmm0[2],xmm10[2],xmm0[3],xmm10[3]
; SSE-NEXT:    movaps %xmm12, %xmm10
; SSE-NEXT:    movlhps {{.*#+}} xmm10 = xmm10[0],xmm8[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm12 = xmm12[1],xmm8[1]
; SSE-NEXT:    movaps %xmm9, %xmm8
; SSE-NEXT:    movlhps {{.*#+}} xmm8 = xmm8[0],xmm3[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm9 = xmm9[1],xmm3[1]
; SSE-NEXT:    movaps %xmm5, %xmm3
; SSE-NEXT:    movlhps {{.*#+}} xmm3 = xmm3[0],xmm6[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm5 = xmm5[1],xmm6[1]
; SSE-NEXT:    movaps %xmm0, %xmm13
; SSE-NEXT:    movlhps {{.*#+}} xmm13 = xmm13[0],xmm4[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm4[1]
; SSE-NEXT:    movaps %xmm15, 48(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; SSE-NEXT:    movaps %xmm4, 16(%rsi)
; SSE-NEXT:    movaps (%rsp), %xmm4 # 16-byte Reload
; SSE-NEXT:    movaps %xmm4, 32(%rsi)
; SSE-NEXT:    movaps %xmm14, (%rsi)
; SSE-NEXT:    movaps %xmm1, 48(%rdx)
; SSE-NEXT:    movaps %xmm7, 16(%rdx)
; SSE-NEXT:    movaps %xmm2, (%rdx)
; SSE-NEXT:    movaps %xmm11, 32(%rdx)
; SSE-NEXT:    movaps %xmm8, 16(%rcx)
; SSE-NEXT:    movaps %xmm3, 48(%rcx)
; SSE-NEXT:    movaps %xmm10, 32(%rcx)
; SSE-NEXT:    movaps %xmm13, (%rcx)
; SSE-NEXT:    movaps %xmm5, 48(%r8)
; SSE-NEXT:    movaps %xmm9, 16(%r8)
; SSE-NEXT:    movaps %xmm12, 32(%r8)
; SSE-NEXT:    movaps %xmm0, (%r8)
; SSE-NEXT:    addq $24, %rsp
; SSE-NEXT:    retq
;
; AVX1-LABEL: load_i32_stride4_vf16:
; AVX1:       # %bb.0:
; AVX1-NEXT:    subq $312, %rsp # imm = 0x138
; AVX1-NEXT:    vmovaps 64(%rdi), %ymm5
; AVX1-NEXT:    vmovaps 96(%rdi), %ymm8
; AVX1-NEXT:    vmovaps 192(%rdi), %ymm2
; AVX1-NEXT:    vmovaps 224(%rdi), %ymm15
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm11 = ymm15[2,3,0,1]
; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm0 = ymm15[0],ymm11[0],ymm15[2],ymm11[2]
; AVX1-NEXT:    vmovups %ymm11, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm7 = ymm2[2,3,0,1]
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm1 = ymm7[0],ymm2[0],ymm7[1],ymm2[1],ymm7[4],ymm2[4],ymm7[5],ymm2[5]
; AVX1-NEXT:    vmovaps %ymm2, %ymm4
; AVX1-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vshufps {{.*#+}} ymm10 = ymm1[0,1],ymm0[2,0],ymm1[4,5],ymm0[6,4]
; AVX1-NEXT:    vmovaps 160(%rdi), %xmm1
; AVX1-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps 176(%rdi), %xmm0
; AVX1-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovlhps {{.*#+}} xmm1 = xmm0[0],xmm1[0]
; AVX1-NEXT:    vmovaps 144(%rdi), %xmm0
; AVX1-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps 128(%rdi), %xmm2
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm3 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
; AVX1-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm3[0,1],xmm1[2,0]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm10[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm8[2,3,0,1]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm0 = ymm8[0],ymm0[0],ymm8[2],ymm0[2]
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm5[2,3,0,1]
; AVX1-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm1 = ymm1[0],ymm5[0],ymm1[1],ymm5[1],ymm1[4],ymm5[4],ymm1[5],ymm5[5]
; AVX1-NEXT:    vshufps {{.*#+}} ymm9 = ymm1[0,1],ymm0[2,0],ymm1[4,5],ymm0[6,4]
; AVX1-NEXT:    vmovaps 32(%rdi), %xmm1
; AVX1-NEXT:    vmovaps 48(%rdi), %xmm0
; AVX1-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; AVX1-NEXT:    vmovaps %xmm1, %xmm14
; AVX1-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps (%rdi), %xmm12
; AVX1-NEXT:    vmovaps 16(%rdi), %xmm6
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm10 = xmm12[0],xmm6[0],xmm12[1],xmm6[1]
; AVX1-NEXT:    vmovaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vshufps {{.*#+}} xmm0 = xmm10[0,1],xmm0[2,0]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm9[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm0 = ymm11[0],ymm15[0],ymm11[1],ymm15[1],ymm11[4],ymm15[4],ymm11[5],ymm15[5]
; AVX1-NEXT:    vshufps {{.*#+}} ymm10 = ymm4[1,0],ymm7[1,0],ymm4[5,4],ymm7[5,4]
; AVX1-NEXT:    vmovaps %ymm7, %ymm11
; AVX1-NEXT:    vmovups %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vshufps {{.*#+}} ymm10 = ymm10[2,0],ymm0[2,3],ymm10[6,4],ymm0[6,7]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX1-NEXT:    vinsertps {{.*#+}} xmm0 = xmm2[1],xmm4[1],zero,zero
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm13 = xmm7[0],xmm9[0],xmm7[1],xmm9[1]
; AVX1-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0,1],xmm13[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm10[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm0 = ymm3[0],ymm8[0],ymm3[1],ymm8[1],ymm3[4],ymm8[4],ymm3[5],ymm8[5]
; AVX1-NEXT:    vmovaps %ymm5, %ymm2
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-NEXT:    vshufps {{.*#+}} ymm10 = ymm5[1,0],ymm1[1,0],ymm5[5,4],ymm1[5,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm10 = ymm10[2,0],ymm0[2,3],ymm10[6,4],ymm0[6,7]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm0 = xmm14[0],xmm5[0],xmm14[1],xmm5[1]
; AVX1-NEXT:    vinsertps {{.*#+}} xmm6 = xmm12[1],xmm6[1],zero,zero
; AVX1-NEXT:    vblendps {{.*#+}} xmm0 = xmm6[0,1],xmm0[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm10[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm13 # 32-byte Reload
; AVX1-NEXT:    vunpckhpd {{.*#+}} ymm0 = ymm15[1],ymm13[1],ymm15[3],ymm13[3]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm14 # 32-byte Reload
; AVX1-NEXT:    vunpckhps {{.*#+}} ymm6 = ymm11[2],ymm14[2],ymm11[3],ymm14[3],ymm11[6],ymm14[6],ymm11[7],ymm14[7]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm6[0,1],ymm0[2,0],ymm6[4,5],ymm0[6,4]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm6 = xmm11[2],xmm4[2],xmm11[3],xmm4[3]
; AVX1-NEXT:    vmovaps %xmm9, %xmm10
; AVX1-NEXT:    vmovaps %xmm7, %xmm4
; AVX1-NEXT:    vinsertps {{.*#+}} xmm7 = zero,zero,xmm7[2],xmm9[2]
; AVX1-NEXT:    vblendps {{.*#+}} xmm6 = xmm6[0,1],xmm7[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm6[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpckhpd {{.*#+}} ymm6 = ymm8[1],ymm3[1],ymm8[3],ymm3[3]
; AVX1-NEXT:    vunpckhps {{.*#+}} ymm7 = ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[6],ymm2[6],ymm1[7],ymm2[7]
; AVX1-NEXT:    vmovaps %ymm2, %ymm9
; AVX1-NEXT:    vshufps {{.*#+}} ymm6 = ymm7[0,1],ymm6[2,0],ymm7[4,5],ymm6[6,4]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; AVX1-NEXT:    vinsertps {{.*#+}} xmm7 = zero,zero,xmm1[2],xmm5[2]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm0 = xmm12[2],xmm3[2],xmm12[3],xmm3[3]
; AVX1-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0,1],xmm7[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm6[4,5,6,7]
; AVX1-NEXT:    vunpckhps {{.*#+}} ymm6 = ymm13[2],ymm15[2],ymm13[3],ymm15[3],ymm13[6],ymm15[6],ymm13[7],ymm15[7]
; AVX1-NEXT:    vshufps $51, {{[-0-9]+}}(%r{{[sb]}}p), %ymm14, %ymm2 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm2 = ymm14[3,0],mem[3,0],ymm14[7,4],mem[7,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm2 = ymm2[2,0],ymm6[2,3],ymm2[6,4],ymm6[6,7]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm4 = xmm4[2],xmm10[2],xmm4[3],xmm10[3]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; AVX1-NEXT:    vshufps {{.*#+}} xmm6 = xmm6[3,0],xmm11[3,0]
; AVX1-NEXT:    vshufps {{.*#+}} xmm4 = xmm6[2,0],xmm4[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm2 = ymm4[0,1,2,3],ymm2[4,5,6,7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-NEXT:    vunpckhps {{.*#+}} ymm4 = ymm4[2],ymm8[2],ymm4[3],ymm8[3],ymm4[6],ymm8[6],ymm4[7],ymm8[7]
; AVX1-NEXT:    vshufps $51, {{[-0-9]+}}(%r{{[sb]}}p), %ymm9, %ymm6 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm6 = ymm9[3,0],mem[3,0],ymm9[7,4],mem[7,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm4 = ymm6[2,0],ymm4[2,3],ymm6[6,4],ymm4[6,7]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm5 = xmm1[2],xmm5[2],xmm1[3],xmm5[3]
; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm3[3,0],xmm12[3,0]
; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[2,0],xmm5[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm4[4,5,6,7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 32(%rsi)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, (%rsi)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 32(%rdx)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, (%rdx)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 32(%rcx)
; AVX1-NEXT:    vmovaps %ymm0, (%rcx)
; AVX1-NEXT:    vmovaps %ymm2, 32(%r8)
; AVX1-NEXT:    vmovaps %ymm1, (%r8)
; AVX1-NEXT:    addq $312, %rsp # imm = 0x138
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: load_i32_stride4_vf16:
; AVX2:       # %bb.0:
; AVX2-NEXT:    subq $104, %rsp
; AVX2-NEXT:    vmovaps (%rdi), %ymm13
; AVX2-NEXT:    vmovups %ymm13, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 32(%rdi), %ymm9
; AVX2-NEXT:    vmovups %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 64(%rdi), %ymm4
; AVX2-NEXT:    vmovaps 96(%rdi), %ymm5
; AVX2-NEXT:    vmovaps 160(%rdi), %ymm8
; AVX2-NEXT:    vmovups %ymm8, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 192(%rdi), %ymm3
; AVX2-NEXT:    vmovaps 224(%rdi), %ymm2
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm0 = [17179869184,17179869184,17179869184,17179869184]
; AVX2-NEXT:    vpermps %ymm2, %ymm0, %ymm1
; AVX2-NEXT:    vpermps %ymm3, %ymm0, %ymm6
; AVX2-NEXT:    vblendps {{.*#+}} ymm1 = ymm6[0,1,2,3,4,5],ymm1[6,7]
; AVX2-NEXT:    vmovaps 144(%rdi), %xmm15
; AVX2-NEXT:    vmovaps 128(%rdi), %xmm10
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm6 = xmm10[0],xmm15[0],xmm10[1],xmm15[1]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm7 = <u,u,0,4>
; AVX2-NEXT:    vpermps %ymm8, %ymm7, %ymm8
; AVX2-NEXT:    vblendps {{.*#+}} xmm6 = xmm6[0,1],xmm8[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm1 = ymm6[0,1,2,3],ymm1[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm5, %ymm0, %ymm1
; AVX2-NEXT:    vpermps %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
; AVX2-NEXT:    vpermps %ymm9, %ymm7, %ymm1
; AVX2-NEXT:    vmovaps (%rdi), %xmm7
; AVX2-NEXT:    vmovaps 16(%rdi), %xmm8
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm6 = xmm7[0],xmm8[0],xmm7[1],xmm8[1]
; AVX2-NEXT:    vblendps {{.*#+}} xmm1 = xmm6[0,1],xmm1[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm0 = [21474836481,21474836481,21474836481,21474836481]
; AVX2-NEXT:    vpermps %ymm5, %ymm0, %ymm1
; AVX2-NEXT:    vpermps %ymm4, %ymm0, %ymm6
; AVX2-NEXT:    vblendps {{.*#+}} ymm12 = ymm6[0,1,2,3,4,5],ymm1[6,7]
; AVX2-NEXT:    vmovaps 32(%rdi), %xmm9
; AVX2-NEXT:    vmovaps 48(%rdi), %xmm6
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm1 = xmm9[0],xmm6[0],xmm9[1],xmm6[1]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm11 = <1,5,u,u>
; AVX2-NEXT:    vpermps %ymm13, %ymm11, %ymm14
; AVX2-NEXT:    vblendps {{.*#+}} xmm1 = xmm14[0,1],xmm1[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm12[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm1, (%rsp) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm2, %ymm0, %ymm1
; AVX2-NEXT:    vpermps %ymm3, %ymm0, %ymm0
; AVX2-NEXT:    vblendps {{.*#+}} ymm12 = ymm0[0,1,2,3,4,5],ymm1[6,7]
; AVX2-NEXT:    vmovaps 128(%rdi), %ymm14
; AVX2-NEXT:    vpermps %ymm14, %ymm11, %ymm11
; AVX2-NEXT:    vmovaps 176(%rdi), %xmm1
; AVX2-NEXT:    vmovaps 160(%rdi), %xmm0
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm13 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; AVX2-NEXT:    vblendps {{.*#+}} xmm11 = xmm11[0,1],xmm13[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm11 = ymm11[0,1,2,3],ymm12[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm11, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm11 = [25769803778,25769803778,25769803778,25769803778]
; AVX2-NEXT:    vpermps %ymm2, %ymm11, %ymm13
; AVX2-NEXT:    vpermps %ymm3, %ymm11, %ymm12
; AVX2-NEXT:    vblendps {{.*#+}} ymm12 = ymm12[0,1,2,3,4,5],ymm13[6,7]
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm10 = xmm10[2],xmm15[2],xmm10[3],xmm15[3]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm13 = <u,u,2,6>
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm13, %ymm15 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} xmm10 = xmm10[0,1],xmm15[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm10 = ymm10[0,1,2,3],ymm12[4,5,6,7]
; AVX2-NEXT:    vpermps %ymm5, %ymm11, %ymm12
; AVX2-NEXT:    vpermps %ymm4, %ymm11, %ymm11
; AVX2-NEXT:    vblendps {{.*#+}} ymm11 = ymm11[0,1,2,3,4,5],ymm12[6,7]
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm13, %ymm12 # 32-byte Folded Reload
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm7 = xmm7[2],xmm8[2],xmm7[3],xmm8[3]
; AVX2-NEXT:    vblendps {{.*#+}} xmm7 = xmm7[0,1],xmm12[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3],ymm11[4,5,6,7]
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm8 = [30064771075,30064771075,30064771075,30064771075]
; AVX2-NEXT:    vpermps %ymm5, %ymm8, %ymm5
; AVX2-NEXT:    vpermps %ymm4, %ymm8, %ymm4
; AVX2-NEXT:    vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3,4,5],ymm5[6,7]
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm5 = xmm9[2],xmm6[2],xmm9[3],xmm6[3]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm6 = <3,7,u,u>
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm9 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} xmm5 = xmm9[0,1],xmm5[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3],ymm4[4,5,6,7]
; AVX2-NEXT:    vpermps %ymm2, %ymm8, %ymm2
; AVX2-NEXT:    vpermps %ymm3, %ymm8, %ymm3
; AVX2-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2,3,4,5],ymm2[6,7]
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX2-NEXT:    vpermps %ymm14, %ymm6, %ymm1
; AVX2-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm2[4,5,6,7]
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm1, 32(%rsi)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm1, (%rsi)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm1, 32(%rdx)
; AVX2-NEXT:    vmovups (%rsp), %ymm1 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm1, (%rdx)
; AVX2-NEXT:    vmovaps %ymm10, 32(%rcx)
; AVX2-NEXT:    vmovaps %ymm7, (%rcx)
; AVX2-NEXT:    vmovaps %ymm0, 32(%r8)
; AVX2-NEXT:    vmovaps %ymm4, (%r8)
; AVX2-NEXT:    addq $104, %rsp
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512-LABEL: load_i32_stride4_vf16:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmovdqu64 (%rdi), %zmm0
; AVX512-NEXT:    vmovdqu64 64(%rdi), %zmm1
; AVX512-NEXT:    vmovdqu64 128(%rdi), %zmm2
; AVX512-NEXT:    vmovdqu64 192(%rdi), %zmm3
; AVX512-NEXT:    vbroadcasti64x4 {{.*#+}} zmm4 = [0,4,8,12,16,20,24,28,0,4,8,12,16,20,24,28]
; AVX512-NEXT:    # zmm4 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT:    vmovdqa64 %zmm2, %zmm5
; AVX512-NEXT:    vpermt2d %zmm3, %zmm4, %zmm5
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm4
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm4 = zmm4[0,1,2,3],zmm5[4,5,6,7]
; AVX512-NEXT:    vbroadcasti64x4 {{.*#+}} zmm5 = [1,5,9,13,17,21,25,29,1,5,9,13,17,21,25,29]
; AVX512-NEXT:    # zmm5 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT:    vmovdqa64 %zmm2, %zmm6
; AVX512-NEXT:    vpermt2d %zmm3, %zmm5, %zmm6
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm5
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm5 = zmm5[0,1,2,3],zmm6[4,5,6,7]
; AVX512-NEXT:    vbroadcasti64x4 {{.*#+}} zmm6 = [2,6,10,14,18,22,26,30,2,6,10,14,18,22,26,30]
; AVX512-NEXT:    # zmm6 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT:    vmovdqa64 %zmm2, %zmm7
; AVX512-NEXT:    vpermt2d %zmm3, %zmm6, %zmm7
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm6
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm6 = zmm6[0,1,2,3],zmm7[4,5,6,7]
; AVX512-NEXT:    vbroadcasti64x4 {{.*#+}} zmm7 = [3,7,11,15,19,23,27,31,3,7,11,15,19,23,27,31]
; AVX512-NEXT:    # zmm7 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT:    vpermt2d %zmm3, %zmm7, %zmm2
; AVX512-NEXT:    vpermt2d %zmm1, %zmm7, %zmm0
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm0 = zmm0[0,1,2,3],zmm2[4,5,6,7]
; AVX512-NEXT:    vmovdqu64 %zmm4, (%rsi)
; AVX512-NEXT:    vmovdqu64 %zmm5, (%rdx)
; AVX512-NEXT:    vmovdqu64 %zmm6, (%rcx)
; AVX512-NEXT:    vmovdqu64 %zmm0, (%r8)
; AVX512-NEXT:    vzeroupper
; AVX512-NEXT:    retq
  %wide.vec = load <64 x i32>, ptr %in.vec, align 32

  %strided.vec0 = shufflevector <64 x i32> %wide.vec, <64 x i32> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 32, i32 36, i32 40, i32 44, i32 48, i32 52, i32 56, i32 60>
  %strided.vec1 = shufflevector <64 x i32> %wide.vec, <64 x i32> poison, <16 x i32> <i32 1, i32 5, i32 9, i32 13, i32 17, i32 21, i32 25, i32 29, i32 33, i32 37, i32 41, i32 45, i32 49, i32 53, i32 57, i32 61>
  %strided.vec2 = shufflevector <64 x i32> %wide.vec, <64 x i32> poison, <16 x i32> <i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30, i32 34, i32 38, i32 42, i32 46, i32 50, i32 54, i32 58, i32 62>
  %strided.vec3 = shufflevector <64 x i32> %wide.vec, <64 x i32> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 19, i32 23, i32 27, i32 31, i32 35, i32 39, i32 43, i32 47, i32 51, i32 55, i32 59, i32 63>

  store <16 x i32> %strided.vec0, ptr %out.vec0, align 32
  store <16 x i32> %strided.vec1, ptr %out.vec1, align 32
  store <16 x i32> %strided.vec2, ptr %out.vec2, align 32
  store <16 x i32> %strided.vec3, ptr %out.vec3, align 32

  ret void
}

define void @load_i32_stride4_vf32(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr %out.vec2, ptr %out.vec3) nounwind {
; SSE-LABEL: load_i32_stride4_vf32:
; SSE:       # %bb.0:
; SSE-NEXT:    subq $456, %rsp # imm = 0x1C8
; SSE-NEXT:    movaps 336(%rdi), %xmm8
; SSE-NEXT:    movaps %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 368(%rdi), %xmm9
; SSE-NEXT:    movaps 352(%rdi), %xmm10
; SSE-NEXT:    movaps %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 208(%rdi), %xmm7
; SSE-NEXT:    movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 192(%rdi), %xmm5
; SSE-NEXT:    movaps %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 240(%rdi), %xmm6
; SSE-NEXT:    movaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 224(%rdi), %xmm14
; SSE-NEXT:    movaps 80(%rdi), %xmm4
; SSE-NEXT:    movaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 64(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 112(%rdi), %xmm0
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 96(%rdi), %xmm11
; SSE-NEXT:    movaps %xmm11, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpcklps {{.*#+}} xmm11 = xmm11[0],xmm0[0],xmm11[1],xmm0[1]
; SSE-NEXT:    movaps %xmm1, %xmm2
; SSE-NEXT:    unpcklps {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
; SSE-NEXT:    movaps %xmm2, %xmm1
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm11[0]
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps %xmm14, %xmm1
; SSE-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1]
; SSE-NEXT:    movaps %xmm5, %xmm3
; SSE-NEXT:    unpcklps {{.*#+}} xmm3 = xmm3[0],xmm7[0],xmm3[1],xmm7[1]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm2 = xmm2[1],xmm11[1]
; SSE-NEXT:    movaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps %xmm3, %xmm0
; SSE-NEXT:    movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm3 = xmm3[1],xmm1[1]
; SSE-NEXT:    movaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps %xmm10, %xmm0
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1]
; SSE-NEXT:    movaps %xmm9, %xmm7
; SSE-NEXT:    movaps 320(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1]
; SSE-NEXT:    movaps %xmm1, %xmm2
; SSE-NEXT:    movlhps {{.*#+}} xmm2 = xmm2[0],xmm0[0]
; SSE-NEXT:    movaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm0[1]
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 496(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 480(%rdi), %xmm0
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movaps 464(%rdi), %xmm3
; SSE-NEXT:    movaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 448(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps %xmm1, %xmm2
; SSE-NEXT:    unpcklps {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movaps %xmm2, %xmm1
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm0[0]
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm2 = xmm2[1],xmm0[1]
; SSE-NEXT:    movaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 176(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 160(%rdi), %xmm9
; SSE-NEXT:    movaps %xmm9, %xmm0
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movaps 144(%rdi), %xmm3
; SSE-NEXT:    movaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 128(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps %xmm1, %xmm2
; SSE-NEXT:    unpcklps {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT:    movaps %xmm2, %xmm1
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm0[0]
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm2 = xmm2[1],xmm0[1]
; SSE-NEXT:    movaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 304(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 288(%rdi), %xmm13
; SSE-NEXT:    movaps %xmm13, %xmm0
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movaps 272(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 256(%rdi), %xmm15
; SSE-NEXT:    movaps %xmm15, %xmm12
; SSE-NEXT:    unpcklps {{.*#+}} xmm12 = xmm12[0],xmm1[0],xmm12[1],xmm1[1]
; SSE-NEXT:    movaps %xmm12, %xmm1
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm0[0]
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm12 = xmm12[1],xmm0[1]
; SSE-NEXT:    movaps 432(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 416(%rdi), %xmm10
; SSE-NEXT:    movaps %xmm10, %xmm0
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movaps 400(%rdi), %xmm1
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps 384(%rdi), %xmm8
; SSE-NEXT:    movaps %xmm8, %xmm3
; SSE-NEXT:    unpcklps {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; SSE-NEXT:    movaps %xmm3, %xmm1
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm0[0]
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm3 = xmm3[1],xmm0[1]
; SSE-NEXT:    movaps 32(%rdi), %xmm2
; SSE-NEXT:    movaps 48(%rdi), %xmm0
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps %xmm2, %xmm1
; SSE-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT:    movaps (%rdi), %xmm5
; SSE-NEXT:    movaps 16(%rdi), %xmm0
; SSE-NEXT:    movaps %xmm0, (%rsp) # 16-byte Spill
; SSE-NEXT:    movaps %xmm5, %xmm4
; SSE-NEXT:    unpcklps {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
; SSE-NEXT:    movaps %xmm4, %xmm0
; SSE-NEXT:    movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm4 = xmm4[1],xmm1[1]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
; SSE-NEXT:    # xmm6 = xmm6[2],mem[2],xmm6[3],mem[3]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE-NEXT:    # xmm0 = xmm0[2],mem[2],xmm0[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
; SSE-NEXT:    # xmm14 = xmm14[2],mem[2],xmm14[3],mem[3]
; SSE-NEXT:    movaps %xmm14, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
; SSE-NEXT:    # xmm14 = xmm14[2],mem[2],xmm14[3],mem[3]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    unpckhps {{.*#+}} xmm11 = xmm11[2],xmm7[2],xmm11[3],xmm7[3]
; SSE-NEXT:    movaps %xmm11, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Folded Reload
; SSE-NEXT:    # xmm7 = xmm7[2],mem[2],xmm7[3],mem[3]
; SSE-NEXT:    movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
; SSE-NEXT:    # xmm11 = xmm11[2],mem[2],xmm11[3],mem[3]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Folded Reload
; SSE-NEXT:    # xmm7 = xmm7[2],mem[2],xmm7[3],mem[3]
; SSE-NEXT:    movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Folded Reload
; SSE-NEXT:    # xmm9 = xmm9[2],mem[2],xmm9[3],mem[3]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Folded Reload
; SSE-NEXT:    # xmm7 = xmm7[2],mem[2],xmm7[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Folded Reload
; SSE-NEXT:    # xmm13 = xmm13[2],mem[2],xmm13[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Folded Reload
; SSE-NEXT:    # xmm15 = xmm15[2],mem[2],xmm15[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
; SSE-NEXT:    # xmm10 = xmm10[2],mem[2],xmm10[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
; SSE-NEXT:    # xmm8 = xmm8[2],mem[2],xmm8[3],mem[3]
; SSE-NEXT:    unpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
; SSE-NEXT:    # xmm2 = xmm2[2],mem[2],xmm2[3],mem[3]
; SSE-NEXT:    unpckhps (%rsp), %xmm5 # 16-byte Folded Reload
; SSE-NEXT:    # xmm5 = xmm5[2],mem[2],xmm5[3],mem[3]
; SSE-NEXT:    movaps %xmm0, %xmm1
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm6[0]
; SSE-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm6[1]
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps %xmm14, %xmm0
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE-NEXT:    movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    unpckhpd {{.*#+}} xmm14 = xmm14[1],xmm1[1]
; SSE-NEXT:    movaps %xmm14, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, %xmm1
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
; SSE-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0],xmm14[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm14[1]
; SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; SSE-NEXT:    movaps %xmm6, %xmm0
; SSE-NEXT:    movlhps {{.*#+}} xmm0 = xmm0[0],xmm11[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm6 = xmm6[1],xmm11[1]
; SSE-NEXT:    movaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT:    movaps %xmm7, %xmm14
; SSE-NEXT:    movaps %xmm7, %xmm11
; SSE-NEXT:    movlhps {{.*#+}} xmm11 = xmm11[0],xmm9[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm14 = xmm14[1],xmm9[1]
; SSE-NEXT:    movaps %xmm15, %xmm9
; SSE-NEXT:    movlhps {{.*#+}} xmm9 = xmm9[0],xmm13[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm15 = xmm15[1],xmm13[1]
; SSE-NEXT:    movaps %xmm8, %xmm7
; SSE-NEXT:    movlhps {{.*#+}} xmm7 = xmm7[0],xmm10[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm8 = xmm8[1],xmm10[1]
; SSE-NEXT:    movaps %xmm5, %xmm6
; SSE-NEXT:    movlhps {{.*#+}} xmm6 = xmm6[0],xmm2[0]
; SSE-NEXT:    unpckhpd {{.*#+}} xmm5 = xmm5[1],xmm2[1]
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 96(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 64(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 32(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, (%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 112(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 80(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 48(%rsi)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 16(%rsi)
; SSE-NEXT:    movaps %xmm3, 96(%rdx)
; SSE-NEXT:    movaps %xmm12, 64(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 32(%rdx)
; SSE-NEXT:    movaps %xmm4, (%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 112(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 80(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 48(%rdx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; SSE-NEXT:    movaps %xmm2, 16(%rdx)
; SSE-NEXT:    movaps %xmm7, 96(%rcx)
; SSE-NEXT:    movaps %xmm9, 64(%rcx)
; SSE-NEXT:    movaps %xmm11, 32(%rcx)
; SSE-NEXT:    movaps %xmm6, (%rcx)
; SSE-NEXT:    movaps %xmm0, 112(%rcx)
; SSE-NEXT:    movaps %xmm1, 80(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 48(%rcx)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 16(%rcx)
; SSE-NEXT:    movaps %xmm8, 96(%r8)
; SSE-NEXT:    movaps %xmm15, 64(%r8)
; SSE-NEXT:    movaps %xmm14, 32(%r8)
; SSE-NEXT:    movaps %xmm5, (%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 112(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 80(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 48(%r8)
; SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT:    movaps %xmm0, 16(%r8)
; SSE-NEXT:    addq $456, %rsp # imm = 0x1C8
; SSE-NEXT:    retq
;
; AVX1-LABEL: load_i32_stride4_vf32:
; AVX1:       # %bb.0:
; AVX1-NEXT:    subq $1000, %rsp # imm = 0x3E8
; AVX1-NEXT:    vmovaps 320(%rdi), %ymm3
; AVX1-NEXT:    vmovaps 352(%rdi), %ymm4
; AVX1-NEXT:    vmovaps 448(%rdi), %ymm2
; AVX1-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovaps 480(%rdi), %ymm1
; AVX1-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm1[2,3,0,1]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[2],ymm0[2]
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm2[2,3,0,1]
; AVX1-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm1 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[4],ymm2[4],ymm1[5],ymm2[5]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,0],ymm1[4,5],ymm0[6,4]
; AVX1-NEXT:    vmovaps 416(%rdi), %xmm2
; AVX1-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps 432(%rdi), %xmm1
; AVX1-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovlhps {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; AVX1-NEXT:    vmovaps 400(%rdi), %xmm5
; AVX1-NEXT:    vmovaps %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps 384(%rdi), %xmm2
; AVX1-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,0]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovups %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm4[2,3,0,1]
; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm0 = ymm4[0],ymm1[0],ymm4[2],ymm1[2]
; AVX1-NEXT:    vmovaps %ymm1, %ymm6
; AVX1-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovaps %ymm3, %ymm5
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm3[2,3,0,1]
; AVX1-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm1 = ymm1[0],ymm3[0],ymm1[1],ymm3[1],ymm1[4],ymm3[4],ymm1[5],ymm3[5]
; AVX1-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,0],ymm1[4,5],ymm0[6,4]
; AVX1-NEXT:    vmovaps 288(%rdi), %xmm2
; AVX1-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps 304(%rdi), %xmm1
; AVX1-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovlhps {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; AVX1-NEXT:    vmovaps 272(%rdi), %xmm3
; AVX1-NEXT:    vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps 256(%rdi), %xmm2
; AVX1-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,0]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovaps 192(%rdi), %ymm0
; AVX1-NEXT:    vmovaps 224(%rdi), %ymm1
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm3 = ymm1[2,3,0,1]
; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm2 = ymm1[0],ymm3[0],ymm1[2],ymm3[2]
; AVX1-NEXT:    vmovaps %ymm3, %ymm13
; AVX1-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovaps %ymm1, %ymm12
; AVX1-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm9 = ymm0[2,3,0,1]
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm3 = ymm9[0],ymm0[0],ymm9[1],ymm0[1],ymm9[4],ymm0[4],ymm9[5],ymm0[5]
; AVX1-NEXT:    vmovups %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovaps %ymm0, %ymm15
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vshufps {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,0],ymm3[4,5],ymm2[6,4]
; AVX1-NEXT:    vmovaps 160(%rdi), %xmm0
; AVX1-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
; AVX1-NEXT:    vmovaps 176(%rdi), %xmm1
; AVX1-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovlhps {{.*#+}} xmm3 = xmm1[0],xmm0[0]
; AVX1-NEXT:    vmovaps 144(%rdi), %xmm4
; AVX1-NEXT:    vmovaps 128(%rdi), %xmm8
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm7 = xmm8[0],xmm4[0],xmm8[1],xmm4[1]
; AVX1-NEXT:    vmovaps %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vshufps {{.*#+}} xmm3 = xmm7[0,1],xmm3[2,0]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3],ymm2[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovaps 64(%rdi), %ymm0
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovaps 96(%rdi), %ymm1
; AVX1-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
; AVX1-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklpd {{.*#+}} ymm10 = ymm1[0],ymm2[0],ymm1[2],ymm2[2]
; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
; AVX1-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm11 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[4],ymm0[4],ymm1[5],ymm0[5]
; AVX1-NEXT:    vshufps {{.*#+}} ymm1 = ymm11[0,1],ymm10[2,0],ymm11[4,5],ymm10[6,4]
; AVX1-NEXT:    vmovaps 32(%rdi), %xmm0
; AVX1-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps 48(%rdi), %xmm2
; AVX1-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovlhps {{.*#+}} xmm0 = xmm2[0],xmm0[0]
; AVX1-NEXT:    vmovaps (%rdi), %xmm2
; AVX1-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vmovaps 16(%rdi), %xmm3
; AVX1-NEXT:    vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm14 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; AVX1-NEXT:    vshufps {{.*#+}} xmm0 = xmm14[0,1],xmm0[2,0]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm0 = ymm6[0],ymm10[0],ymm6[1],ymm10[1],ymm6[4],ymm10[4],ymm6[5],ymm10[5]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
; AVX1-NEXT:    vshufps {{.*#+}} ymm1 = ymm5[1,0],ymm6[1,0],ymm5[5,4],ymm6[5,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[2,0],ymm0[2,3],ymm1[6,4],ymm0[6,7]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; AVX1-NEXT:    vinsertps {{.*#+}} xmm1 = xmm11[1],xmm5[1],zero,zero
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX1-NEXT:    vunpcklps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm7 # 16-byte Folded Reload
; AVX1-NEXT:    # xmm7 = xmm2[0],mem[0],xmm2[1],mem[1]
; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm7[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm0 = ymm13[0],ymm12[0],ymm13[1],ymm12[1],ymm13[4],ymm12[4],ymm13[5],ymm12[5]
; AVX1-NEXT:    vshufps {{.*#+}} ymm1 = ymm15[1,0],ymm9[1,0],ymm15[5,4],ymm9[5,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[2,0],ymm0[2,3],ymm1[6,4],ymm0[6,7]
; AVX1-NEXT:    vinsertps {{.*#+}} xmm1 = xmm8[1],xmm4[1],zero,zero
; AVX1-NEXT:    vmovaps (%rsp), %xmm2 # 16-byte Reload
; AVX1-NEXT:    vunpcklps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm7 # 16-byte Folded Reload
; AVX1-NEXT:    # xmm7 = xmm2[0],mem[0],xmm2[1],mem[1]
; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm7[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm14 # 32-byte Reload
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
; AVX1-NEXT:    vunpcklps {{.*#+}} ymm0 = ymm14[0],ymm2[0],ymm14[1],ymm2[1],ymm14[4],ymm2[4],ymm14[5],ymm2[5]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm12 # 32-byte Reload
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vshufps {{.*#+}} ymm1 = ymm3[1,0],ymm12[1,0],ymm3[5,4],ymm12[5,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[2,0],ymm0[2,3],ymm1[6,4],ymm0[6,7]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; AVX1-NEXT:    vinsertps {{.*#+}} xmm1 = xmm8[1],xmm1[1],zero,zero
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm7 = xmm4[0],xmm9[0],xmm4[1],xmm9[1]
; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm7[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-NEXT:    vunpcklps {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm0 = ymm0[0],mem[0],ymm0[1],mem[1],ymm0[4],mem[4],ymm0[5],mem[5]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm13 # 32-byte Reload
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[1,0],ymm13[1,0],ymm1[5,4],ymm13[5,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[2,0],ymm0[2,3],ymm1[6,4],ymm0[6,7]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; AVX1-NEXT:    vunpcklps {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
; AVX1-NEXT:    vinsertps $12, {{[-0-9]+}}(%r{{[sb]}}p), %xmm7, %xmm7 # 16-byte Folded Reload
; AVX1-NEXT:    # xmm7 = mem[0],xmm7[1],zero,zero
; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm7[0,1],xmm1[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpckhpd {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm0 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm0 = ymm10[1],mem[1],ymm10[3],mem[3]
; AVX1-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm1 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm1 = ymm6[2],mem[2],ymm6[3],mem[3],ymm6[6],mem[6],ymm6[7],mem[7]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,0],ymm1[4,5],ymm0[6,4]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm1 = xmm11[2],xmm5[2],xmm11[3],xmm5[3]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; AVX1-NEXT:    vinsertps $51, {{[-0-9]+}}(%r{{[sb]}}p), %xmm5, %xmm7 # 16-byte Folded Reload
; AVX1-NEXT:    # xmm7 = zero,zero,xmm5[2],mem[0]
; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm7[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-NEXT:    vunpckhpd {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm0 = ymm0[1],mem[1],ymm0[3],mem[3]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX1-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %ymm10, %ymm1 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm1 = ymm10[2],mem[2],ymm10[3],mem[3],ymm10[6],mem[6],ymm10[7],mem[7]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,0],ymm1[4,5],ymm0[6,4]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Reload
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm1 = xmm1[2],xmm11[2],xmm1[3],xmm11[3]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
; AVX1-NEXT:    vmovaps (%rsp), %xmm5 # 16-byte Reload
; AVX1-NEXT:    vinsertps {{.*#+}} xmm7 = zero,zero,xmm5[2],xmm6[2]
; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm7[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vunpckhpd {{.*#+}} ymm0 = ymm2[1],ymm14[1],ymm2[3],ymm14[3]
; AVX1-NEXT:    vunpckhps {{.*#+}} ymm1 = ymm12[2],ymm3[2],ymm12[3],ymm3[3],ymm12[6],ymm3[6],ymm12[7],ymm3[7]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,0],ymm1[4,5],ymm0[6,4]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm1 = xmm8[2],xmm2[2],xmm8[3],xmm2[3]
; AVX1-NEXT:    vinsertps {{.*#+}} xmm7 = zero,zero,xmm4[2],xmm9[2]
; AVX1-NEXT:    vmovaps %xmm9, %xmm3
; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm7[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm14 # 32-byte Reload
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm12 # 32-byte Reload
; AVX1-NEXT:    vunpckhpd {{.*#+}} ymm0 = ymm14[1],ymm12[1],ymm14[3],ymm12[3]
; AVX1-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %ymm13, %ymm1 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm1 = ymm13[2],mem[2],ymm13[3],mem[3],ymm13[6],mem[6],ymm13[7],mem[7]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,0],ymm1[4,5],ymm0[6,4]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
; AVX1-NEXT:    vinsertps {{.*#+}} xmm1 = zero,zero,xmm13[2],xmm15[2]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm7 = xmm8[2],xmm9[2],xmm8[3],xmm9[3]
; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm7[0,1],xmm1[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm7 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm0 = ymm0[2],mem[2],ymm0[3],mem[3],ymm0[6],mem[6],ymm0[7],mem[7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[3,0],ymm10[3,0],ymm1[7,4],ymm10[7,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm1[2,0],ymm0[2,3],ymm1[6,4],ymm0[6,7]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm1 = xmm5[2],xmm6[2],xmm5[3],xmm6[3]
; AVX1-NEXT:    vshufps $51, {{[-0-9]+}}(%r{{[sb]}}p), %xmm11, %xmm4 # 16-byte Folded Reload
; AVX1-NEXT:    # xmm4 = xmm11[3,0],mem[3,0]
; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm4[2,0],xmm1[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
; AVX1-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm1 = ymm1[2],mem[2],ymm1[3],mem[3],ymm1[6],mem[6],ymm1[7],mem[7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX1-NEXT:    vshufps $51, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm4 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm4 = ymm4[3,0],mem[3,0],ymm4[7,4],mem[7,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm1 = ymm4[2,0],ymm1[2,3],ymm4[6,4],ymm1[6,7]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm4 = xmm4[2],xmm3[2],xmm4[3],xmm3[3]
; AVX1-NEXT:    vshufps $51, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm5 # 16-byte Folded Reload
; AVX1-NEXT:    # xmm5 = xmm2[3,0],mem[3,0]
; AVX1-NEXT:    vshufps {{.*#+}} xmm4 = xmm5[2,0],xmm4[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm1 = ymm4[0,1,2,3],ymm1[4,5,6,7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm4 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm4 = ymm3[2],mem[2],ymm3[3],mem[3],ymm3[6],mem[6],ymm3[7],mem[7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vshufps $51, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm5 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm5 = ymm3[3,0],mem[3,0],ymm3[7,4],mem[7,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm4 = ymm5[2,0],ymm4[2,3],ymm5[6,4],ymm4[6,7]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; AVX1-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm5 # 16-byte Folded Reload
; AVX1-NEXT:    # xmm5 = xmm3[2],mem[2],xmm3[3],mem[3]
; AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX1-NEXT:    vshufps $51, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm6 # 16-byte Folded Reload
; AVX1-NEXT:    # xmm6 = xmm2[3,0],mem[3,0]
; AVX1-NEXT:    vshufps {{.*#+}} xmm5 = xmm6[2,0],xmm5[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3],ymm4[4,5,6,7]
; AVX1-NEXT:    vunpckhps {{.*#+}} ymm2 = ymm12[2],ymm14[2],ymm12[3],ymm14[3],ymm12[6],ymm14[6],ymm12[7],ymm14[7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vshufps $51, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm3 # 32-byte Folded Reload
; AVX1-NEXT:    # ymm3 = ymm3[3,0],mem[3,0],ymm3[7,4],mem[7,4]
; AVX1-NEXT:    vshufps {{.*#+}} ymm2 = ymm3[2,0],ymm2[2,3],ymm3[6,4],ymm2[6,7]
; AVX1-NEXT:    vunpckhps {{.*#+}} xmm3 = xmm13[2],xmm15[2],xmm13[3],xmm15[3]
; AVX1-NEXT:    vshufps {{.*#+}} xmm5 = xmm9[3,0],xmm8[3,0]
; AVX1-NEXT:    vshufps {{.*#+}} xmm3 = xmm5[2,0],xmm3[2,3]
; AVX1-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2,3],ymm2[4,5,6,7]
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, (%rsi)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 32(%rsi)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 64(%rsi)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 96(%rsi)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 96(%rdx)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, (%rdx)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 32(%rdx)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 64(%rdx)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 96(%rcx)
; AVX1-NEXT:    vmovaps %ymm7, (%rcx)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 32(%rcx)
; AVX1-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
; AVX1-NEXT:    vmovaps %ymm3, 64(%rcx)
; AVX1-NEXT:    vmovaps %ymm4, 64(%r8)
; AVX1-NEXT:    vmovaps %ymm1, 96(%r8)
; AVX1-NEXT:    vmovaps %ymm2, (%r8)
; AVX1-NEXT:    vmovaps %ymm0, 32(%r8)
; AVX1-NEXT:    addq $1000, %rsp # imm = 0x3E8
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: load_i32_stride4_vf32:
; AVX2:       # %bb.0:
; AVX2-NEXT:    subq $696, %rsp # imm = 0x2B8
; AVX2-NEXT:    vmovaps 288(%rdi), %ymm12
; AVX2-NEXT:    vmovaps 320(%rdi), %ymm5
; AVX2-NEXT:    vmovaps 352(%rdi), %ymm6
; AVX2-NEXT:    vmovaps 416(%rdi), %ymm7
; AVX2-NEXT:    vmovups %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 448(%rdi), %ymm2
; AVX2-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 480(%rdi), %ymm1
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm3 = [17179869184,17179869184,17179869184,17179869184]
; AVX2-NEXT:    vpermps %ymm1, %ymm3, %ymm0
; AVX2-NEXT:    vmovaps %ymm1, %ymm14
; AVX2-NEXT:    vpermps %ymm2, %ymm3, %ymm1
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3,4,5],ymm0[6,7]
; AVX2-NEXT:    vmovaps 400(%rdi), %xmm2
; AVX2-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vmovaps 384(%rdi), %xmm1
; AVX2-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm4 = <u,u,0,4>
; AVX2-NEXT:    vpermps %ymm7, %ymm4, %ymm2
; AVX2-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm6, %ymm3, %ymm0
; AVX2-NEXT:    vpermps %ymm5, %ymm3, %ymm1
; AVX2-NEXT:    vmovaps %ymm5, %ymm9
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3,4,5],ymm0[6,7]
; AVX2-NEXT:    vmovaps 272(%rdi), %xmm2
; AVX2-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vmovaps 256(%rdi), %xmm1
; AVX2-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
; AVX2-NEXT:    vpermps %ymm12, %ymm4, %ymm2
; AVX2-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3]
; AVX2-NEXT:    vmovaps 192(%rdi), %ymm5
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 224(%rdi), %ymm10
; AVX2-NEXT:    vpermps %ymm10, %ymm3, %ymm0
; AVX2-NEXT:    vpermps %ymm5, %ymm3, %ymm2
; AVX2-NEXT:    vmovaps %ymm5, %ymm8
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3,4,5],ymm0[6,7]
; AVX2-NEXT:    vmovaps 160(%rdi), %ymm1
; AVX2-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 144(%rdi), %xmm5
; AVX2-NEXT:    vmovaps %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vmovaps 128(%rdi), %xmm2
; AVX2-NEXT:    vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
; AVX2-NEXT:    vpermps %ymm1, %ymm4, %ymm5
; AVX2-NEXT:    vblendps {{.*#+}} xmm2 = xmm2[0,1],xmm5[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 64(%rdi), %ymm15
; AVX2-NEXT:    vmovaps 96(%rdi), %ymm11
; AVX2-NEXT:    vpermps %ymm11, %ymm3, %ymm5
; AVX2-NEXT:    vpermps %ymm15, %ymm3, %ymm3
; AVX2-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3,4,5],ymm5[6,7]
; AVX2-NEXT:    vmovaps 32(%rdi), %ymm0
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm0, %ymm4, %ymm4
; AVX2-NEXT:    vmovaps (%rdi), %xmm0
; AVX2-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vmovaps 16(%rdi), %xmm1
; AVX2-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm5 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; AVX2-NEXT:    vblendps {{.*#+}} xmm4 = xmm5[0,1],xmm4[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm4[0,1,2,3],ymm3[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm3 = [21474836481,21474836481,21474836481,21474836481]
; AVX2-NEXT:    vpermps %ymm6, %ymm3, %ymm4
; AVX2-NEXT:    vmovaps %ymm6, %ymm2
; AVX2-NEXT:    vmovups %ymm6, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps %ymm9, %ymm1
; AVX2-NEXT:    vmovups %ymm9, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm9, %ymm3, %ymm5
; AVX2-NEXT:    vblendps {{.*#+}} ymm5 = ymm5[0,1,2,3,4,5],ymm4[6,7]
; AVX2-NEXT:    vmovaps 256(%rdi), %ymm0
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 304(%rdi), %xmm4
; AVX2-NEXT:    vmovaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vmovaps 288(%rdi), %xmm6
; AVX2-NEXT:    vmovaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm6 = xmm6[0],xmm4[0],xmm6[1],xmm4[1]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm4 = <1,5,u,u>
; AVX2-NEXT:    vpermps %ymm0, %ymm4, %ymm7
; AVX2-NEXT:    vblendps {{.*#+}} xmm6 = xmm7[0,1],xmm6[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm6[0,1,2,3],ymm5[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm10, %ymm3, %ymm5
; AVX2-NEXT:    vpermps %ymm8, %ymm3, %ymm6
; AVX2-NEXT:    vmovaps %ymm8, %ymm7
; AVX2-NEXT:    vblendps {{.*#+}} ymm8 = ymm6[0,1,2,3,4,5],ymm5[6,7]
; AVX2-NEXT:    vmovaps 128(%rdi), %ymm0
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 176(%rdi), %xmm6
; AVX2-NEXT:    vmovaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vmovaps 160(%rdi), %xmm5
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm13 = xmm5[0],xmm6[0],xmm5[1],xmm6[1]
; AVX2-NEXT:    vpermps %ymm0, %ymm4, %ymm9
; AVX2-NEXT:    vblendps {{.*#+}} xmm6 = xmm9[0,1],xmm13[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm6[0,1,2,3],ymm8[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm11, %ymm3, %ymm6
; AVX2-NEXT:    vpermps %ymm15, %ymm3, %ymm8
; AVX2-NEXT:    vblendps {{.*#+}} ymm9 = ymm8[0,1,2,3,4,5],ymm6[6,7]
; AVX2-NEXT:    vmovaps (%rdi), %ymm13
; AVX2-NEXT:    vmovups %ymm13, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps 32(%rdi), %xmm6
; AVX2-NEXT:    vmovaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vmovaps 48(%rdi), %xmm0
; AVX2-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm8 = xmm6[0],xmm0[0],xmm6[1],xmm0[1]
; AVX2-NEXT:    vpermps %ymm13, %ymm4, %ymm13
; AVX2-NEXT:    vblendps {{.*#+}} xmm0 = xmm13[0,1],xmm8[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm9[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vmovaps %ymm14, %ymm9
; AVX2-NEXT:    vpermps %ymm14, %ymm3, %ymm0
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm8 # 32-byte Reload
; AVX2-NEXT:    vpermps %ymm8, %ymm3, %ymm3
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4,5],ymm0[6,7]
; AVX2-NEXT:    vmovaps 384(%rdi), %ymm3
; AVX2-NEXT:    vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm3, %ymm4, %ymm13
; AVX2-NEXT:    vmovaps 432(%rdi), %xmm4
; AVX2-NEXT:    vmovaps 416(%rdi), %xmm3
; AVX2-NEXT:    vunpcklps {{.*#+}} xmm14 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
; AVX2-NEXT:    vblendps {{.*#+}} xmm13 = xmm13[0,1],xmm14[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm13[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm0 = [25769803778,25769803778,25769803778,25769803778]
; AVX2-NEXT:    vpermps %ymm2, %ymm0, %ymm13
; AVX2-NEXT:    vpermps %ymm1, %ymm0, %ymm14
; AVX2-NEXT:    vblendps {{.*#+}} ymm13 = ymm14[0,1,2,3,4,5],ymm13[6,7]
; AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; AVX2-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm14 # 16-byte Folded Reload
; AVX2-NEXT:    # xmm14 = xmm1[2],mem[2],xmm1[3],mem[3]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm1 = <u,u,2,6>
; AVX2-NEXT:    vpermps %ymm12, %ymm1, %ymm12
; AVX2-NEXT:    vblendps {{.*#+}} xmm12 = xmm14[0,1],xmm12[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm2 = ymm12[0,1,2,3],ymm13[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm10, %ymm0, %ymm12
; AVX2-NEXT:    vpermps %ymm7, %ymm0, %ymm13
; AVX2-NEXT:    vblendps {{.*#+}} ymm12 = ymm13[0,1,2,3,4,5],ymm12[6,7]
; AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX2-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm13 # 16-byte Folded Reload
; AVX2-NEXT:    # xmm13 = xmm2[2],mem[2],xmm2[3],mem[3]
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm14 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} xmm13 = xmm13[0,1],xmm14[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm2 = ymm13[0,1,2,3],ymm12[4,5,6,7]
; AVX2-NEXT:    vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-NEXT:    vpermps %ymm9, %ymm0, %ymm12
; AVX2-NEXT:    vpermps %ymm8, %ymm0, %ymm13
; AVX2-NEXT:    vblendps {{.*#+}} ymm12 = ymm13[0,1,2,3,4,5],ymm12[6,7]
; AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX2-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm13 # 16-byte Folded Reload
; AVX2-NEXT:    # xmm13 = xmm2[2],mem[2],xmm2[3],mem[3]
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm14 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} xmm13 = xmm13[0,1],xmm14[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm14 = ymm13[0,1,2,3],ymm12[4,5,6,7]
; AVX2-NEXT:    vpermps %ymm11, %ymm0, %ymm12
; AVX2-NEXT:    vpermps %ymm15, %ymm0, %ymm0
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm12[6,7]
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm1 # 32-byte Folded Reload
; AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX2-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm6 # 16-byte Folded Reload
; AVX2-NEXT:    # xmm6 = xmm2[2],mem[2],xmm2[3],mem[3]
; AVX2-NEXT:    vblendps {{.*#+}} xmm1 = xmm6[0,1],xmm1[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm13 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT:    vbroadcastsd {{.*#+}} ymm6 = [30064771075,30064771075,30064771075,30064771075]
; AVX2-NEXT:    vpermps %ymm10, %ymm6, %ymm0
; AVX2-NEXT:    vpermps %ymm7, %ymm6, %ymm1
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3,4,5],ymm0[6,7]
; AVX2-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5, %xmm1 # 16-byte Folded Reload
; AVX2-NEXT:    # xmm1 = xmm5[2],mem[2],xmm5[3],mem[3]
; AVX2-NEXT:    vmovaps {{.*#+}} xmm5 = <3,7,u,u>
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm7 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} xmm1 = xmm7[0,1],xmm1[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT:    vpermps %ymm11, %ymm6, %ymm0
; AVX2-NEXT:    vpermps %ymm15, %ymm6, %ymm2
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3,4,5],ymm0[6,7]
; AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX2-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 16-byte Folded Reload
; AVX2-NEXT:    # xmm2 = xmm2[2],mem[2],xmm2[3],mem[3]
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm7 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} xmm2 = xmm7[0,1],xmm2[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5,6,7]
; AVX2-NEXT:    vpermps %ymm9, %ymm6, %ymm2
; AVX2-NEXT:    vpermps %ymm8, %ymm6, %ymm7
; AVX2-NEXT:    vblendps {{.*#+}} ymm2 = ymm7[0,1,2,3,4,5],ymm2[6,7]
; AVX2-NEXT:    vunpckhps {{.*#+}} xmm3 = xmm3[2],xmm4[2],xmm3[3],xmm4[3]
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm4 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} xmm3 = xmm4[0,1],xmm3[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm2 = ymm3[0,1,2,3],ymm2[4,5,6,7]
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm3 # 32-byte Folded Reload
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm6, %ymm4 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2,3,4,5],ymm3[6,7]
; AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
; AVX2-NEXT:    vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm4, %xmm4 # 16-byte Folded Reload
; AVX2-NEXT:    # xmm4 = xmm4[2],mem[2],xmm4[3],mem[3]
; AVX2-NEXT:    vpermps {{[-0-9]+}}(%r{{[sb]}}p), %ymm5, %ymm5 # 32-byte Folded Reload
; AVX2-NEXT:    vblendps {{.*#+}} xmm4 = xmm5[0,1],xmm4[2,3]
; AVX2-NEXT:    vblendps {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm3[4,5,6,7]
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, (%rsi)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, 32(%rsi)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, 64(%rsi)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, 96(%rsi)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, 96(%rdx)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, (%rdx)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, 32(%rdx)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, 64(%rdx)
; AVX2-NEXT:    vmovaps %ymm14, 96(%rcx)
; AVX2-NEXT:    vmovaps %ymm13, (%rcx)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, 32(%rcx)
; AVX2-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
; AVX2-NEXT:    vmovaps %ymm4, 64(%rcx)
; AVX2-NEXT:    vmovaps %ymm3, 64(%r8)
; AVX2-NEXT:    vmovaps %ymm2, 96(%r8)
; AVX2-NEXT:    vmovaps %ymm0, (%r8)
; AVX2-NEXT:    vmovaps %ymm1, 32(%r8)
; AVX2-NEXT:    addq $696, %rsp # imm = 0x2B8
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512-LABEL: load_i32_stride4_vf32:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmovdqu64 (%rdi), %zmm0
; AVX512-NEXT:    vmovdqu64 64(%rdi), %zmm1
; AVX512-NEXT:    vmovdqu64 128(%rdi), %zmm2
; AVX512-NEXT:    vmovdqu64 192(%rdi), %zmm3
; AVX512-NEXT:    vmovdqu64 320(%rdi), %zmm4
; AVX512-NEXT:    vmovdqu64 256(%rdi), %zmm5
; AVX512-NEXT:    vmovdqu64 448(%rdi), %zmm6
; AVX512-NEXT:    vmovdqu64 384(%rdi), %zmm7
; AVX512-NEXT:    vbroadcasti64x4 {{.*#+}} zmm8 = [0,4,8,12,16,20,24,28,0,4,8,12,16,20,24,28]
; AVX512-NEXT:    # zmm8 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT:    vmovdqa64 %zmm7, %zmm9
; AVX512-NEXT:    vpermt2d %zmm6, %zmm8, %zmm9
; AVX512-NEXT:    vmovdqa64 %zmm5, %zmm10
; AVX512-NEXT:    vpermt2d %zmm4, %zmm8, %zmm10
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm9 = zmm10[0,1,2,3],zmm9[4,5,6,7]
; AVX512-NEXT:    vmovdqa64 %zmm2, %zmm10
; AVX512-NEXT:    vpermt2d %zmm3, %zmm8, %zmm10
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm8
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm8 = zmm8[0,1,2,3],zmm10[4,5,6,7]
; AVX512-NEXT:    vbroadcasti64x4 {{.*#+}} zmm10 = [1,5,9,13,17,21,25,29,1,5,9,13,17,21,25,29]
; AVX512-NEXT:    # zmm10 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT:    vmovdqa64 %zmm7, %zmm11
; AVX512-NEXT:    vpermt2d %zmm6, %zmm10, %zmm11
; AVX512-NEXT:    vmovdqa64 %zmm5, %zmm12
; AVX512-NEXT:    vpermt2d %zmm4, %zmm10, %zmm12
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm11 = zmm12[0,1,2,3],zmm11[4,5,6,7]
; AVX512-NEXT:    vmovdqa64 %zmm2, %zmm12
; AVX512-NEXT:    vpermt2d %zmm3, %zmm10, %zmm12
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm10
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm10 = zmm10[0,1,2,3],zmm12[4,5,6,7]
; AVX512-NEXT:    vbroadcasti64x4 {{.*#+}} zmm12 = [2,6,10,14,18,22,26,30,2,6,10,14,18,22,26,30]
; AVX512-NEXT:    # zmm12 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT:    vmovdqa64 %zmm7, %zmm13
; AVX512-NEXT:    vpermt2d %zmm6, %zmm12, %zmm13
; AVX512-NEXT:    vmovdqa64 %zmm5, %zmm14
; AVX512-NEXT:    vpermt2d %zmm4, %zmm12, %zmm14
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm13 = zmm14[0,1,2,3],zmm13[4,5,6,7]
; AVX512-NEXT:    vmovdqa64 %zmm2, %zmm14
; AVX512-NEXT:    vpermt2d %zmm3, %zmm12, %zmm14
; AVX512-NEXT:    vpermi2d %zmm1, %zmm0, %zmm12
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm12 = zmm12[0,1,2,3],zmm14[4,5,6,7]
; AVX512-NEXT:    vbroadcasti64x4 {{.*#+}} zmm14 = [3,7,11,15,19,23,27,31,3,7,11,15,19,23,27,31]
; AVX512-NEXT:    # zmm14 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT:    vpermt2d %zmm6, %zmm14, %zmm7
; AVX512-NEXT:    vpermt2d %zmm4, %zmm14, %zmm5
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm4 = zmm5[0,1,2,3],zmm7[4,5,6,7]
; AVX512-NEXT:    vpermt2d %zmm3, %zmm14, %zmm2
; AVX512-NEXT:    vpermt2d %zmm1, %zmm14, %zmm0
; AVX512-NEXT:    vshufi64x2 {{.*#+}} zmm0 = zmm0[0,1,2,3],zmm2[4,5,6,7]
; AVX512-NEXT:    vmovdqu64 %zmm9, 64(%rsi)
; AVX512-NEXT:    vmovdqu64 %zmm8, (%rsi)
; AVX512-NEXT:    vmovdqu64 %zmm11, 64(%rdx)
; AVX512-NEXT:    vmovdqu64 %zmm10, (%rdx)
; AVX512-NEXT:    vmovdqu64 %zmm13, 64(%rcx)
; AVX512-NEXT:    vmovdqu64 %zmm12, (%rcx)
; AVX512-NEXT:    vmovdqu64 %zmm4, 64(%r8)
; AVX512-NEXT:    vmovdqu64 %zmm0, (%r8)
; AVX512-NEXT:    vzeroupper
; AVX512-NEXT:    retq
  %wide.vec = load <128 x i32>, ptr %in.vec, align 32

  %strided.vec0 = shufflevector <128 x i32> %wide.vec, <128 x i32> poison, <32 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 32, i32 36, i32 40, i32 44, i32 48, i32 52, i32 56, i32 60, i32 64, i32 68, i32 72, i32 76, i32 80, i32 84, i32 88, i32 92, i32 96, i32 100, i32 104, i32 108, i32 112, i32 116, i32 120, i32 124>
  %strided.vec1 = shufflevector <128 x i32> %wide.vec, <128 x i32> poison, <32 x i32> <i32 1, i32 5, i32 9, i32 13, i32 17, i32 21, i32 25, i32 29, i32 33, i32 37, i32 41, i32 45, i32 49, i32 53, i32 57, i32 61, i32 65, i32 69, i32 73, i32 77, i32 81, i32 85, i32 89, i32 93, i32 97, i32 101, i32 105, i32 109, i32 113, i32 117, i32 121, i32 125>
  %strided.vec2 = shufflevector <128 x i32> %wide.vec, <128 x i32> poison, <32 x i32> <i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30, i32 34, i32 38, i32 42, i32 46, i32 50, i32 54, i32 58, i32 62, i32 66, i32 70, i32 74, i32 78, i32 82, i32 86, i32 90, i32 94, i32 98, i32 102, i32 106, i32 110, i32 114, i32 118, i32 122, i32 126>
  %strided.vec3 = shufflevector <128 x i32> %wide.vec, <128 x i32> poison, <32 x i32> <i32 3, i32 7, i32 11, i32 15, i32 19, i32 23, i32 27, i32 31, i32 35, i32 39, i32 43, i32 47, i32 51, i32 55, i32 59, i32 63, i32 67, i32 71, i32 75, i32 79, i32 83, i32 87, i32 91, i32 95, i32 99, i32 103, i32 107, i32 111, i32 115, i32 119, i32 123, i32 127>

  store <32 x i32> %strided.vec0, ptr %out.vec0, align 32
  store <32 x i32> %strided.vec1, ptr %out.vec1, align 32
  store <32 x i32> %strided.vec2, ptr %out.vec2, align 32
  store <32 x i32> %strided.vec3, ptr %out.vec3, align 32

  ret void
}