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

; These patterns are produced by LoopVectorizer for interleaved loads.

define void @load_i8_stride2_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1) nounwind {
; SSE-LABEL: load_i8_stride2_vf2:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa (%rdi), %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    psrlw $8, %xmm0
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movd %xmm1, %eax
; SSE-NEXT:    movw %ax, (%rsi)
; SSE-NEXT:    movd %xmm0, %eax
; SSE-NEXT:    movw %ax, (%rdx)
; SSE-NEXT:    retq
;
; AVX1-LABEL: load_i8_stride2_vf2:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovdqa (%rdi), %xmm0
; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[0,2,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-NEXT:    vpextrw $0, %xmm1, (%rsi)
; AVX1-NEXT:    vpextrw $0, %xmm0, (%rdx)
; AVX1-NEXT:    retq
;
; AVX512F-LABEL: load_i8_stride2_vf2:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[0,2,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpextrw $0, %xmm1, (%rsi)
; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rdx)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride2_vf2:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpmovwb %xmm0, %xmm1
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpextrw $0, %xmm1, (%rsi)
; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rdx)
; AVX512BW-NEXT:    retq
  %wide.vec = load <4 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <4 x i8> %wide.vec, <4 x i8> poison, <2 x i32> <i32 0, i32 2>
  %strided.vec1 = shufflevector <4 x i8> %wide.vec, <4 x i8> poison, <2 x i32> <i32 1, i32 3>
  store <2 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <2 x i8> %strided.vec1, ptr %out.vec1, align 64
  ret void
}

define void @load_i8_stride2_vf4(ptr %in.vec, ptr %out.vec0, ptr %out.vec1) nounwind {
; SSE-LABEL: load_i8_stride2_vf4:
; SSE:       # %bb.0:
; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    psrlw $8, %xmm0
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movd %xmm1, (%rsi)
; SSE-NEXT:    movd %xmm0, (%rdx)
; SSE-NEXT:    retq
;
; AVX1-LABEL: load_i8_stride2_vf4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[0,2,4,6,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX1-NEXT:    vmovd %xmm1, (%rsi)
; AVX1-NEXT:    vmovd %xmm0, (%rdx)
; AVX1-NEXT:    retq
;
; AVX512F-LABEL: load_i8_stride2_vf4:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[0,2,4,6,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vmovd %xmm1, (%rsi)
; AVX512F-NEXT:    vmovd %xmm0, (%rdx)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride2_vf4:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512BW-NEXT:    vpmovwb %xmm0, %xmm1
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,u,u,u,u,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vmovd %xmm1, (%rsi)
; AVX512BW-NEXT:    vmovd %xmm0, (%rdx)
; AVX512BW-NEXT:    retq
  %wide.vec = load <8 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <8 x i8> %wide.vec, <8 x i8> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %strided.vec1 = shufflevector <8 x i8> %wide.vec, <8 x i8> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  store <4 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <4 x i8> %strided.vec1, ptr %out.vec1, align 64
  ret void
}

define void @load_i8_stride2_vf8(ptr %in.vec, ptr %out.vec0, ptr %out.vec1) nounwind {
; SSE-LABEL: load_i8_stride2_vf8:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa (%rdi), %xmm0
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    pand %xmm0, %xmm1
; SSE-NEXT:    packuswb %xmm1, %xmm1
; SSE-NEXT:    psrlw $8, %xmm0
; SSE-NEXT:    packuswb %xmm0, %xmm0
; SSE-NEXT:    movq %xmm1, (%rsi)
; SSE-NEXT:    movq %xmm0, (%rdx)
; SSE-NEXT:    retq
;
; AVX1-LABEL: load_i8_stride2_vf8:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovdqa (%rdi), %xmm0
; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; AVX1-NEXT:    vmovq %xmm1, (%rsi)
; AVX1-NEXT:    vmovq %xmm0, (%rdx)
; AVX1-NEXT:    retq
;
; AVX512F-LABEL: load_i8_stride2_vf8:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; AVX512F-NEXT:    vmovq %xmm1, (%rsi)
; AVX512F-NEXT:    vmovq %xmm0, (%rdx)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride2_vf8:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpmovwb %xmm0, (%rsi)
; AVX512BW-NEXT:    vmovq %xmm1, (%rdx)
; AVX512BW-NEXT:    retq
  %wide.vec = load <16 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <16 x i8> %wide.vec, <16 x i8> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
  %strided.vec1 = shufflevector <16 x i8> %wide.vec, <16 x i8> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
  store <8 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <8 x i8> %strided.vec1, ptr %out.vec1, align 64
  ret void
}

define void @load_i8_stride2_vf16(ptr %in.vec, ptr %out.vec0, ptr %out.vec1) nounwind {
; SSE-LABEL: load_i8_stride2_vf16:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa (%rdi), %xmm0
; SSE-NEXT:    movdqa 16(%rdi), %xmm1
; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm1, %xmm3
; SSE-NEXT:    pand %xmm2, %xmm3
; SSE-NEXT:    pand %xmm0, %xmm2
; SSE-NEXT:    packuswb %xmm3, %xmm2
; SSE-NEXT:    psrlw $8, %xmm1
; SSE-NEXT:    psrlw $8, %xmm0
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa %xmm2, (%rsi)
; SSE-NEXT:    movdqa %xmm0, (%rdx)
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i8_stride2_vf16:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm0 = [255,255,255,255,255,255,255,255]
; AVX1-ONLY-NEXT:    vmovdqa (%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpand %xmm0, %xmm2, %xmm3
; AVX1-ONLY-NEXT:    vpand %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vpackuswb %xmm3, %xmm0, %xmm0
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm3 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX1-ONLY-NEXT:    # xmm3 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm2, %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, (%rsi)
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, (%rdx)
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i8_stride2_vf16:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    vpbroadcastw {{.*#+}} xmm0 = [255,255,255,255,255,255,255,255]
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %xmm1
; AVX2-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm2
; AVX2-ONLY-NEXT:    vpand %xmm0, %xmm2, %xmm3
; AVX2-ONLY-NEXT:    vpand %xmm0, %xmm1, %xmm0
; AVX2-ONLY-NEXT:    vpackuswb %xmm3, %xmm0, %xmm0
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX2-ONLY-NEXT:    vpshufb %xmm3, %xmm2, %xmm2
; AVX2-ONLY-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
; AVX2-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; AVX2-ONLY-NEXT:    vmovdqa %xmm0, (%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %xmm1, (%rdx)
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-LABEL: load_i8_stride2_vf16:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vpbroadcastd {{.*#+}} xmm0 = [255,255,255,255,255,255,255,255]
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm1
; AVX512F-NEXT:    vmovdqa 16(%rdi), %xmm2
; AVX512F-NEXT:    vpand %xmm0, %xmm2, %xmm3
; AVX512F-NEXT:    vpand %xmm0, %xmm1, %xmm0
; AVX512F-NEXT:    vpackuswb %xmm3, %xmm0, %xmm0
; AVX512F-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX512F-NEXT:    vpshufb %xmm3, %xmm2, %xmm2
; AVX512F-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
; AVX512F-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; AVX512F-NEXT:    vmovdqa %xmm0, (%rsi)
; AVX512F-NEXT:    vmovdqa %xmm1, (%rdx)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride2_vf16:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512BW-NEXT:    vpsrlw $8, %ymm0, %ymm1
; AVX512BW-NEXT:    vpmovwb %ymm0, (%rsi)
; AVX512BW-NEXT:    vpmovwb %ymm1, (%rdx)
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <32 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <32 x i8> %wide.vec, <32 x i8> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
  %strided.vec1 = shufflevector <32 x i8> %wide.vec, <32 x i8> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
  store <16 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <16 x i8> %strided.vec1, ptr %out.vec1, align 64
  ret void
}

define void @load_i8_stride2_vf32(ptr %in.vec, ptr %out.vec0, ptr %out.vec1) nounwind {
; SSE-LABEL: load_i8_stride2_vf32:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa (%rdi), %xmm0
; SSE-NEXT:    movdqa 16(%rdi), %xmm1
; SSE-NEXT:    movdqa 32(%rdi), %xmm2
; SSE-NEXT:    movdqa 48(%rdi), %xmm3
; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm3, %xmm5
; SSE-NEXT:    pand %xmm4, %xmm5
; SSE-NEXT:    movdqa %xmm2, %xmm6
; SSE-NEXT:    pand %xmm4, %xmm6
; SSE-NEXT:    packuswb %xmm5, %xmm6
; SSE-NEXT:    movdqa %xmm1, %xmm5
; SSE-NEXT:    pand %xmm4, %xmm5
; SSE-NEXT:    pand %xmm0, %xmm4
; SSE-NEXT:    packuswb %xmm5, %xmm4
; SSE-NEXT:    psrlw $8, %xmm3
; SSE-NEXT:    psrlw $8, %xmm2
; SSE-NEXT:    packuswb %xmm3, %xmm2
; SSE-NEXT:    psrlw $8, %xmm1
; SSE-NEXT:    psrlw $8, %xmm0
; SSE-NEXT:    packuswb %xmm1, %xmm0
; SSE-NEXT:    movdqa %xmm4, (%rsi)
; SSE-NEXT:    movdqa %xmm6, 16(%rsi)
; SSE-NEXT:    movdqa %xmm0, (%rdx)
; SSE-NEXT:    movdqa %xmm2, 16(%rdx)
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i8_stride2_vf32:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm0 = [255,255,255,255,255,255,255,255]
; AVX1-ONLY-NEXT:    vmovdqa (%rdi), %xmm1
; AVX1-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vmovdqa 32(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vmovdqa 48(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vpand %xmm0, %xmm4, %xmm5
; AVX1-ONLY-NEXT:    vpand %xmm0, %xmm3, %xmm6
; AVX1-ONLY-NEXT:    vpackuswb %xmm5, %xmm6, %xmm5
; AVX1-ONLY-NEXT:    vpand %xmm0, %xmm2, %xmm6
; AVX1-ONLY-NEXT:    vpand %xmm0, %xmm1, %xmm0
; AVX1-ONLY-NEXT:    vpackuswb %xmm6, %xmm0, %xmm0
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm6 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX1-ONLY-NEXT:    # xmm6 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm4, %xmm4
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm3, %xmm3
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm4[0]
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm2, %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm6, %xmm1, %xmm1
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm1, %ymm1
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, (%rsi)
; AVX1-ONLY-NEXT:    vmovdqa %xmm5, 16(%rsi)
; AVX1-ONLY-NEXT:    vmovaps %ymm1, (%rdx)
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i8_stride2_vf32:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm2 = ymm1[u,u,u,u,u,u,u,u,0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm3 = ymm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5],ymm2[6,7]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,1,3]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u,17,19,21,23,25,27,29,31]
; AVX2-ONLY-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u,17,19,21,23,25,27,29,31,u,u,u,u,u,u,u,u]
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, (%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, (%rdx)
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-SLOW-LABEL: load_i8_stride2_vf32:
; AVX512F-SLOW:       # %bb.0:
; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512F-SLOW-NEXT:    vmovdqa 32(%rdi), %ymm1
; AVX512F-SLOW-NEXT:    vpshufb {{.*#+}} ymm2 = ymm1[u,u,u,u,u,u,u,u,0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30]
; AVX512F-SLOW-NEXT:    vpshufb {{.*#+}} ymm3 = ymm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u]
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5],ymm2[6,7]
; AVX512F-SLOW-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,1,3]
; AVX512F-SLOW-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u,17,19,21,23,25,27,29,31]
; AVX512F-SLOW-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u,17,19,21,23,25,27,29,31,u,u,u,u,u,u,u,u]
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX512F-SLOW-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX512F-SLOW-NEXT:    vmovdqa %ymm2, (%rsi)
; AVX512F-SLOW-NEXT:    vmovdqa %ymm0, (%rdx)
; AVX512F-SLOW-NEXT:    vzeroupper
; AVX512F-SLOW-NEXT:    retq
;
; AVX512F-FAST-LABEL: load_i8_stride2_vf32:
; AVX512F-FAST:       # %bb.0:
; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512F-FAST-NEXT:    vmovdqa 32(%rdi), %ymm1
; AVX512F-FAST-NEXT:    vpshufb {{.*#+}} ymm2 = ymm1[u,u,u,u,u,u,u,u,0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30]
; AVX512F-FAST-NEXT:    vpshufb {{.*#+}} ymm3 = ymm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u]
; AVX512F-FAST-NEXT:    vmovdqa {{.*#+}} ymm4 = [0,2,5,7]
; AVX512F-FAST-NEXT:    vpermt2q %ymm2, %ymm4, %ymm3
; AVX512F-FAST-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u,17,19,21,23,25,27,29,31]
; AVX512F-FAST-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u,17,19,21,23,25,27,29,31,u,u,u,u,u,u,u,u]
; AVX512F-FAST-NEXT:    vpermt2q %ymm1, %ymm4, %ymm0
; AVX512F-FAST-NEXT:    vmovdqa %ymm3, (%rsi)
; AVX512F-FAST-NEXT:    vmovdqa %ymm0, (%rdx)
; AVX512F-FAST-NEXT:    vzeroupper
; AVX512F-FAST-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride2_vf32:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovdqa64 (%rdi), %zmm0
; AVX512BW-NEXT:    vpsrlw $8, %zmm0, %zmm1
; AVX512BW-NEXT:    vpmovwb %zmm0, (%rsi)
; AVX512BW-NEXT:    vpmovwb %zmm1, (%rdx)
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <64 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <64 x i8> %wide.vec, <64 x i8> poison, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
  %strided.vec1 = shufflevector <64 x i8> %wide.vec, <64 x i8> poison, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
  store <32 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <32 x i8> %strided.vec1, ptr %out.vec1, align 64
  ret void
}

define void @load_i8_stride2_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1) nounwind {
; SSE-LABEL: load_i8_stride2_vf64:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa 64(%rdi), %xmm0
; SSE-NEXT:    movdqa 80(%rdi), %xmm4
; SSE-NEXT:    movdqa 96(%rdi), %xmm1
; SSE-NEXT:    movdqa 112(%rdi), %xmm7
; SSE-NEXT:    movdqa (%rdi), %xmm2
; SSE-NEXT:    movdqa 16(%rdi), %xmm9
; SSE-NEXT:    movdqa 32(%rdi), %xmm3
; SSE-NEXT:    movdqa 48(%rdi), %xmm11
; SSE-NEXT:    movdqa {{.*#+}} xmm6 = [255,255,255,255,255,255,255,255]
; SSE-NEXT:    movdqa %xmm11, %xmm8
; SSE-NEXT:    pand %xmm6, %xmm8
; SSE-NEXT:    movdqa %xmm3, %xmm5
; SSE-NEXT:    pand %xmm6, %xmm5
; SSE-NEXT:    packuswb %xmm8, %xmm5
; SSE-NEXT:    movdqa %xmm9, %xmm10
; SSE-NEXT:    pand %xmm6, %xmm10
; SSE-NEXT:    movdqa %xmm2, %xmm8
; SSE-NEXT:    pand %xmm6, %xmm8
; SSE-NEXT:    packuswb %xmm10, %xmm8
; SSE-NEXT:    movdqa %xmm7, %xmm12
; SSE-NEXT:    pand %xmm6, %xmm12
; SSE-NEXT:    movdqa %xmm1, %xmm10
; SSE-NEXT:    pand %xmm6, %xmm10
; SSE-NEXT:    packuswb %xmm12, %xmm10
; SSE-NEXT:    movdqa %xmm4, %xmm12
; SSE-NEXT:    pand %xmm6, %xmm12
; SSE-NEXT:    pand %xmm0, %xmm6
; SSE-NEXT:    packuswb %xmm12, %xmm6
; SSE-NEXT:    psrlw $8, %xmm11
; SSE-NEXT:    psrlw $8, %xmm3
; SSE-NEXT:    packuswb %xmm11, %xmm3
; SSE-NEXT:    psrlw $8, %xmm9
; SSE-NEXT:    psrlw $8, %xmm2
; SSE-NEXT:    packuswb %xmm9, %xmm2
; SSE-NEXT:    psrlw $8, %xmm7
; SSE-NEXT:    psrlw $8, %xmm1
; SSE-NEXT:    packuswb %xmm7, %xmm1
; SSE-NEXT:    psrlw $8, %xmm4
; SSE-NEXT:    psrlw $8, %xmm0
; SSE-NEXT:    packuswb %xmm4, %xmm0
; SSE-NEXT:    movdqa %xmm6, 32(%rsi)
; SSE-NEXT:    movdqa %xmm10, 48(%rsi)
; SSE-NEXT:    movdqa %xmm8, (%rsi)
; SSE-NEXT:    movdqa %xmm5, 16(%rsi)
; SSE-NEXT:    movdqa %xmm0, 32(%rdx)
; SSE-NEXT:    movdqa %xmm1, 48(%rdx)
; SSE-NEXT:    movdqa %xmm2, (%rdx)
; SSE-NEXT:    movdqa %xmm3, 16(%rdx)
; SSE-NEXT:    retq
;
; AVX1-ONLY-LABEL: load_i8_stride2_vf64:
; AVX1-ONLY:       # %bb.0:
; AVX1-ONLY-NEXT:    vbroadcastss {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
; AVX1-ONLY-NEXT:    vmovdqa 112(%rdi), %xmm2
; AVX1-ONLY-NEXT:    vpand %xmm1, %xmm2, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa 96(%rdi), %xmm3
; AVX1-ONLY-NEXT:    vpand %xmm1, %xmm3, %xmm4
; AVX1-ONLY-NEXT:    vpackuswb %xmm0, %xmm4, %xmm0
; AVX1-ONLY-NEXT:    vmovdqa 80(%rdi), %xmm4
; AVX1-ONLY-NEXT:    vpand %xmm1, %xmm4, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa 64(%rdi), %xmm6
; AVX1-ONLY-NEXT:    vpand %xmm1, %xmm6, %xmm7
; AVX1-ONLY-NEXT:    vpackuswb %xmm5, %xmm7, %xmm5
; AVX1-ONLY-NEXT:    vmovdqa (%rdi), %xmm7
; AVX1-ONLY-NEXT:    vmovdqa 16(%rdi), %xmm8
; AVX1-ONLY-NEXT:    vmovdqa 32(%rdi), %xmm9
; AVX1-ONLY-NEXT:    vmovdqa 48(%rdi), %xmm10
; AVX1-ONLY-NEXT:    vpand %xmm1, %xmm10, %xmm11
; AVX1-ONLY-NEXT:    vpand %xmm1, %xmm9, %xmm12
; AVX1-ONLY-NEXT:    vpackuswb %xmm11, %xmm12, %xmm11
; AVX1-ONLY-NEXT:    vpand %xmm1, %xmm8, %xmm12
; AVX1-ONLY-NEXT:    vpand %xmm1, %xmm7, %xmm1
; AVX1-ONLY-NEXT:    vpackuswb %xmm12, %xmm1, %xmm1
; AVX1-ONLY-NEXT:    vmovddup {{.*#+}} xmm12 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX1-ONLY-NEXT:    # xmm12 = mem[0,0]
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm2, %xmm2
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm3, %xmm3
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm4, %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm6, %xmm4
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm4[0],xmm3[0]
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm2, %ymm3, %ymm2
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm10, %xmm3
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm9, %xmm4
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm3 = xmm4[0],xmm3[0]
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm8, %xmm4
; AVX1-ONLY-NEXT:    vpshufb %xmm12, %xmm7, %xmm6
; AVX1-ONLY-NEXT:    vpunpcklqdq {{.*#+}} xmm4 = xmm6[0],xmm4[0]
; AVX1-ONLY-NEXT:    vinsertf128 $1, %xmm3, %ymm4, %ymm3
; AVX1-ONLY-NEXT:    vmovdqa %xmm1, (%rsi)
; AVX1-ONLY-NEXT:    vmovdqa %xmm11, 16(%rsi)
; AVX1-ONLY-NEXT:    vmovdqa %xmm5, 32(%rsi)
; AVX1-ONLY-NEXT:    vmovdqa %xmm0, 48(%rsi)
; AVX1-ONLY-NEXT:    vmovaps %ymm3, (%rdx)
; AVX1-ONLY-NEXT:    vmovaps %ymm2, 32(%rdx)
; AVX1-ONLY-NEXT:    vzeroupper
; AVX1-ONLY-NEXT:    retq
;
; AVX2-ONLY-LABEL: load_i8_stride2_vf64:
; AVX2-ONLY:       # %bb.0:
; AVX2-ONLY-NEXT:    vmovdqa (%rdi), %ymm0
; AVX2-ONLY-NEXT:    vmovdqa 32(%rdi), %ymm1
; AVX2-ONLY-NEXT:    vmovdqa 64(%rdi), %ymm2
; AVX2-ONLY-NEXT:    vmovdqa 96(%rdi), %ymm3
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm4 = [0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14]
; AVX2-ONLY-NEXT:    vpshufb %ymm4, %ymm3, %ymm5
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm6 = [0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14]
; AVX2-ONLY-NEXT:    vpshufb %ymm6, %ymm2, %ymm7
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm5 = ymm7[0,1],ymm5[2,3],ymm7[4,5],ymm5[6,7]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm5 = ymm5[0,2,1,3]
; AVX2-ONLY-NEXT:    vpshufb %ymm4, %ymm1, %ymm4
; AVX2-ONLY-NEXT:    vpshufb %ymm6, %ymm0, %ymm6
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm4 = ymm6[0,1],ymm4[2,3],ymm6[4,5],ymm4[6,7]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm4 = ymm4[0,2,1,3]
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm6 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX2-ONLY-NEXT:    vpshufb %ymm6, %ymm3, %ymm3
; AVX2-ONLY-NEXT:    vpbroadcastq {{.*#+}} ymm7 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX2-ONLY-NEXT:    vpshufb %ymm7, %ymm2, %ymm2
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1],ymm3[2,3],ymm2[4,5],ymm3[6,7]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,1,3]
; AVX2-ONLY-NEXT:    vpshufb %ymm6, %ymm1, %ymm1
; AVX2-ONLY-NEXT:    vpshufb %ymm7, %ymm0, %ymm0
; AVX2-ONLY-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX2-ONLY-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-ONLY-NEXT:    vmovdqa %ymm4, (%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %ymm5, 32(%rsi)
; AVX2-ONLY-NEXT:    vmovdqa %ymm0, (%rdx)
; AVX2-ONLY-NEXT:    vmovdqa %ymm2, 32(%rdx)
; AVX2-ONLY-NEXT:    vzeroupper
; AVX2-ONLY-NEXT:    retq
;
; AVX512F-SLOW-LABEL: load_i8_stride2_vf64:
; AVX512F-SLOW:       # %bb.0:
; AVX512F-SLOW-NEXT:    vpbroadcastq {{.*#+}} ymm0 = [0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14]
; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %ymm1
; AVX512F-SLOW-NEXT:    vmovdqa 32(%rdi), %ymm2
; AVX512F-SLOW-NEXT:    vmovdqa 64(%rdi), %ymm3
; AVX512F-SLOW-NEXT:    vmovdqa 96(%rdi), %ymm4
; AVX512F-SLOW-NEXT:    vpshufb %ymm0, %ymm4, %ymm5
; AVX512F-SLOW-NEXT:    vpbroadcastq {{.*#+}} ymm6 = [0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14]
; AVX512F-SLOW-NEXT:    vpshufb %ymm6, %ymm3, %ymm7
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} ymm5 = ymm7[0,1],ymm5[2,3],ymm7[4,5],ymm5[6,7]
; AVX512F-SLOW-NEXT:    vpshufb %ymm0, %ymm2, %ymm0
; AVX512F-SLOW-NEXT:    vpshufb %ymm6, %ymm1, %ymm6
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} ymm0 = ymm6[0,1],ymm0[2,3],ymm6[4,5],ymm0[6,7]
; AVX512F-SLOW-NEXT:    vinserti64x4 $1, %ymm5, %zmm0, %zmm0
; AVX512F-SLOW-NEXT:    vpermq {{.*#+}} zmm0 = zmm0[0,2,1,3,4,6,5,7]
; AVX512F-SLOW-NEXT:    vpbroadcastq {{.*#+}} ymm5 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX512F-SLOW-NEXT:    vpshufb %ymm5, %ymm4, %ymm4
; AVX512F-SLOW-NEXT:    vpbroadcastq {{.*#+}} ymm6 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX512F-SLOW-NEXT:    vpshufb %ymm6, %ymm3, %ymm3
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3],ymm3[4,5],ymm4[6,7]
; AVX512F-SLOW-NEXT:    vpshufb %ymm5, %ymm2, %ymm2
; AVX512F-SLOW-NEXT:    vpshufb %ymm6, %ymm1, %ymm1
; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm2[2,3],ymm1[4,5],ymm2[6,7]
; AVX512F-SLOW-NEXT:    vinserti64x4 $1, %ymm3, %zmm1, %zmm1
; AVX512F-SLOW-NEXT:    vpermq {{.*#+}} zmm1 = zmm1[0,2,1,3,4,6,5,7]
; AVX512F-SLOW-NEXT:    vmovdqa64 %zmm0, (%rsi)
; AVX512F-SLOW-NEXT:    vmovdqa64 %zmm1, (%rdx)
; AVX512F-SLOW-NEXT:    vzeroupper
; AVX512F-SLOW-NEXT:    retq
;
; AVX512F-FAST-LABEL: load_i8_stride2_vf64:
; AVX512F-FAST:       # %bb.0:
; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} ymm0 = [0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14]
; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %ymm1
; AVX512F-FAST-NEXT:    vmovdqa 32(%rdi), %ymm2
; AVX512F-FAST-NEXT:    vmovdqa 64(%rdi), %ymm3
; AVX512F-FAST-NEXT:    vmovdqa 96(%rdi), %ymm4
; AVX512F-FAST-NEXT:    vpshufb %ymm0, %ymm4, %ymm5
; AVX512F-FAST-NEXT:    vpshufb %ymm0, %ymm2, %ymm0
; AVX512F-FAST-NEXT:    vinserti64x4 $1, %ymm5, %zmm0, %zmm0
; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} ymm5 = [0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14,0,2,4,6,8,10,12,14]
; AVX512F-FAST-NEXT:    vpshufb %ymm5, %ymm3, %ymm6
; AVX512F-FAST-NEXT:    vpshufb %ymm5, %ymm1, %ymm5
; AVX512F-FAST-NEXT:    vinserti64x4 $1, %ymm6, %zmm5, %zmm5
; AVX512F-FAST-NEXT:    vmovdqa64 {{.*#+}} zmm6 = [0,2,9,11,4,6,13,15]
; AVX512F-FAST-NEXT:    vpermt2q %zmm0, %zmm6, %zmm5
; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} ymm0 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX512F-FAST-NEXT:    vpshufb %ymm0, %ymm4, %ymm4
; AVX512F-FAST-NEXT:    vpshufb %ymm0, %ymm2, %ymm0
; AVX512F-FAST-NEXT:    vinserti64x4 $1, %ymm4, %zmm0, %zmm0
; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15,1,3,5,7,9,11,13,15]
; AVX512F-FAST-NEXT:    vpshufb %ymm2, %ymm3, %ymm3
; AVX512F-FAST-NEXT:    vpshufb %ymm2, %ymm1, %ymm1
; AVX512F-FAST-NEXT:    vinserti64x4 $1, %ymm3, %zmm1, %zmm1
; AVX512F-FAST-NEXT:    vpermt2q %zmm0, %zmm6, %zmm1
; AVX512F-FAST-NEXT:    vmovdqa64 %zmm5, (%rsi)
; AVX512F-FAST-NEXT:    vmovdqa64 %zmm1, (%rdx)
; AVX512F-FAST-NEXT:    vzeroupper
; AVX512F-FAST-NEXT:    retq
;
; AVX512BW-LABEL: load_i8_stride2_vf64:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovdqa64 (%rdi), %zmm0
; AVX512BW-NEXT:    vmovdqa64 64(%rdi), %zmm1
; AVX512BW-NEXT:    vpshufb {{.*#+}} zmm2 = zmm1[u,u,u,u,u,u,u,u,0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u,32,34,36,38,40,42,44,46,u,u,u,u,u,u,u,u,48,50,52,54,56,58,60,62]
; AVX512BW-NEXT:    vpshufb {{.*#+}} zmm3 = zmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u,32,34,36,38,40,42,44,46,u,u,u,u,u,u,u,u,48,50,52,54,56,58,60,62,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm4 = [0,2,4,6,9,11,13,15]
; AVX512BW-NEXT:    vpermt2q %zmm2, %zmm4, %zmm3
; AVX512BW-NEXT:    vpshufb {{.*#+}} zmm1 = zmm1[u,u,u,u,u,u,u,u,1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u,17,19,21,23,25,27,29,31,u,u,u,u,u,u,u,u,33,35,37,39,41,43,45,47,u,u,u,u,u,u,u,u,49,51,53,55,57,59,61,63]
; AVX512BW-NEXT:    vpshufb {{.*#+}} zmm0 = zmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u,17,19,21,23,25,27,29,31,u,u,u,u,u,u,u,u,33,35,37,39,41,43,45,47,u,u,u,u,u,u,u,u,49,51,53,55,57,59,61,63,u,u,u,u,u,u,u,u]
; AVX512BW-NEXT:    vpermt2q %zmm1, %zmm4, %zmm0
; AVX512BW-NEXT:    vmovdqa64 %zmm3, (%rsi)
; AVX512BW-NEXT:    vmovdqa64 %zmm0, (%rdx)
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
  %wide.vec = load <128 x i8>, ptr %in.vec, align 64
  %strided.vec0 = shufflevector <128 x i8> %wide.vec, <128 x i8> poison, <64 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62, i32 64, i32 66, i32 68, i32 70, i32 72, i32 74, i32 76, i32 78, i32 80, i32 82, i32 84, i32 86, i32 88, i32 90, i32 92, i32 94, i32 96, i32 98, i32 100, i32 102, i32 104, i32 106, i32 108, i32 110, i32 112, i32 114, i32 116, i32 118, i32 120, i32 122, i32 124, i32 126>
  %strided.vec1 = shufflevector <128 x i8> %wide.vec, <128 x i8> poison, <64 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63, i32 65, i32 67, i32 69, i32 71, i32 73, i32 75, i32 77, i32 79, i32 81, i32 83, i32 85, i32 87, i32 89, i32 91, i32 93, i32 95, i32 97, i32 99, i32 101, i32 103, i32 105, i32 107, i32 109, i32 111, i32 113, i32 115, i32 117, i32 119, i32 121, i32 123, i32 125, i32 127>
  store <64 x i8> %strided.vec0, ptr %out.vec0, align 64
  store <64 x i8> %strided.vec1, ptr %out.vec1, align 64
  ret void
}
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; AVX: {{.*}}
; AVX2: {{.*}}
; AVX2-FAST: {{.*}}
; AVX2-FAST-PERLANE: {{.*}}
; AVX2-SLOW: {{.*}}
; AVX512: {{.*}}
; AVX512-FAST: {{.*}}
; AVX512-SLOW: {{.*}}
; AVX512BW-FAST: {{.*}}
; AVX512BW-ONLY: {{.*}}
; AVX512BW-ONLY-FAST: {{.*}}
; AVX512BW-ONLY-SLOW: {{.*}}
; AVX512BW-SLOW: {{.*}}
; AVX512DQ-FAST: {{.*}}
; AVX512DQ-ONLY: {{.*}}
; AVX512DQ-SLOW: {{.*}}
; AVX512DQBW-FAST: {{.*}}
; AVX512DQBW-ONLY: {{.*}}
; AVX512DQBW-SLOW: {{.*}}
; AVX512F-ONLY: {{.*}}
; AVX512F-ONLY-FAST: {{.*}}
; AVX512F-ONLY-SLOW: {{.*}}
; FALLBACK0: {{.*}}
; FALLBACK1: {{.*}}
; FALLBACK10: {{.*}}
; FALLBACK11: {{.*}}
; FALLBACK12: {{.*}}
; FALLBACK2: {{.*}}
; FALLBACK3: {{.*}}
; FALLBACK4: {{.*}}
; FALLBACK5: {{.*}}
; FALLBACK6: {{.*}}
; FALLBACK7: {{.*}}
; FALLBACK8: {{.*}}
; FALLBACK9: {{.*}}