Google Git
Sign in
llvm / llvm-project / llvm / d42433f828aa47533a213f63bfac32115a62fea2 / . / test / CodeGen / X86 / vector-interleaved-store-i32-stride-6.ll
blob: e3175a82992bcdd2ab083b916d3305e491da3b34 [file] [log] [blame]
; 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,AVX2-SLOW
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX2,AVX2-FAST-ALL
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX2,AVX2-FAST-PERLANE
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512bw,+avx512vl | FileCheck %s --check-prefixes=AVX512
; These patterns are produced by LoopVectorizer for interleaved stores.
define void @store_i32_stride6_vf2(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vecptr2, ptr %in.vecptr3, ptr %in.vecptr4, ptr %in.vecptr5, ptr %out.vec) nounwind {
; SSE-LABEL: store_i32_stride6_vf2:
; SSE: # %bb.0:
; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
; SSE-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
; SSE-NEXT: movsd {{.*#+}} xmm2 = mem[0],zero
; SSE-NEXT: movsd {{.*#+}} xmm3 = mem[0],zero
; SSE-NEXT: movaps %xmm0, %xmm4
; SSE-NEXT: movlhps {{.*#+}} xmm4 = xmm4[0],xmm1[0]
; SSE-NEXT: movaps %xmm2, %xmm5
; SSE-NEXT: movlhps {{.*#+}} xmm5 = xmm5[0],xmm3[0]
; SSE-NEXT: movsd {{.*#+}} xmm6 = mem[0],zero
; SSE-NEXT: movsd {{.*#+}} xmm7 = mem[0],zero
; SSE-NEXT: movlhps {{.*#+}} xmm7 = xmm7[0],xmm6[0]
; SSE-NEXT: unpcklps {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; SSE-NEXT: unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT: movlhps {{.*#+}} xmm0 = xmm0[0],xmm2[0]
; SSE-NEXT: shufps {{.*#+}} xmm5 = xmm5[1,3],xmm7[1,3]
; SSE-NEXT: shufps {{.*#+}} xmm7 = xmm7[0,2],xmm4[1,3]
; SSE-NEXT: movaps %xmm5, 32(%rax)
; SSE-NEXT: movaps %xmm7, 16(%rax)
; SSE-NEXT: movaps %xmm0, (%rax)
; SSE-NEXT: retq
;
; AVX1-LABEL: store_i32_stride6_vf2:
; AVX1: # %bb.0:
; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX1-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; AVX1-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX1-NEXT: vmovlhps {{.*#+}} xmm2 = xmm1[0],xmm0[0]
; AVX1-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
; AVX1-NEXT: vmovsd {{.*#+}} xmm4 = mem[0],zero
; AVX1-NEXT: vmovlhps {{.*#+}} xmm3 = xmm4[0],xmm3[0]
; AVX1-NEXT: vmovsd {{.*#+}} xmm4 = mem[0],zero
; AVX1-NEXT: vmovsd {{.*#+}} xmm5 = mem[0],zero
; AVX1-NEXT: vmovlhps {{.*#+}} xmm4 = xmm5[0],xmm4[0]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm3, %ymm2
; AVX1-NEXT: vpermilps {{.*#+}} ymm2 = ymm2[u,u,0,2,u,u,5,7]
; AVX1-NEXT: vunpcklps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm2[2,3,4,5,6,7]
; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm4[0,2,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm3[1,3],xmm4[1,3]
; AVX1-NEXT: vmovaps %xmm1, 32(%rax)
; AVX1-NEXT: vmovaps %ymm0, (%rax)
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-SLOW-LABEL: store_i32_stride6_vf2:
; AVX2-SLOW: # %bb.0:
; AVX2-SLOW-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX2-SLOW-NEXT: vmovlhps {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
; AVX2-SLOW-NEXT: vmovlhps {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; AVX2-SLOW-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
; AVX2-SLOW-NEXT: vmovlhps {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,1,3,4,6,5,7]
; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,1]
; AVX2-SLOW-NEXT: vpermilps {{.*#+}} xmm3 = xmm2[0,2,2,3]
; AVX2-SLOW-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm3[4,5],ymm0[6,7]
; AVX2-SLOW-NEXT: vshufps {{.*#+}} xmm1 = xmm1[1,3],xmm2[1,3]
; AVX2-SLOW-NEXT: vmovaps %xmm1, 32(%rax)
; AVX2-SLOW-NEXT: vmovaps %ymm0, (%rax)
; AVX2-SLOW-NEXT: vzeroupper
; AVX2-SLOW-NEXT: retq
;
; AVX2-FAST-ALL-LABEL: store_i32_stride6_vf2:
; AVX2-FAST-ALL: # %bb.0:
; AVX2-FAST-ALL-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-FAST-ALL-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; AVX2-FAST-ALL-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX2-FAST-ALL-NEXT: vmovlhps {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX2-FAST-ALL-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX2-FAST-ALL-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
; AVX2-FAST-ALL-NEXT: vmovlhps {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; AVX2-FAST-ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX2-FAST-ALL-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX2-FAST-ALL-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
; AVX2-FAST-ALL-NEXT: vmovlhps {{.*#+}} xmm3 = xmm2[0],xmm1[0]
; AVX2-FAST-ALL-NEXT: vshufps {{.*#+}} ymm3 = ymm0[1,3],ymm3[1,3],ymm0[5,7],ymm3[5,7]
; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm3 = ymm3[2,1,2,3]
; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} ymm4 = <0,2,4,6,u,u,1,3>
; AVX2-FAST-ALL-NEXT: vpermps %ymm0, %ymm4, %ymm0
; AVX2-FAST-ALL-NEXT: vunpcklps {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; AVX2-FAST-ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, (%rax)
; AVX2-FAST-ALL-NEXT: vmovaps %xmm3, 32(%rax)
; AVX2-FAST-ALL-NEXT: vzeroupper
; AVX2-FAST-ALL-NEXT: retq
;
; AVX2-FAST-PERLANE-LABEL: store_i32_stride6_vf2:
; AVX2-FAST-PERLANE: # %bb.0:
; AVX2-FAST-PERLANE-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-FAST-PERLANE-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; AVX2-FAST-PERLANE-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX2-FAST-PERLANE-NEXT: vmovlhps {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX2-FAST-PERLANE-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX2-FAST-PERLANE-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
; AVX2-FAST-PERLANE-NEXT: vmovlhps {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; AVX2-FAST-PERLANE-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX2-FAST-PERLANE-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
; AVX2-FAST-PERLANE-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
; AVX2-FAST-PERLANE-NEXT: vmovlhps {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,1,3,4,6,5,7]
; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,1]
; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} xmm3 = xmm2[0,2,2,3]
; AVX2-FAST-PERLANE-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm3
; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm3[4,5],ymm0[6,7]
; AVX2-FAST-PERLANE-NEXT: vshufps {{.*#+}} xmm1 = xmm1[1,3],xmm2[1,3]
; AVX2-FAST-PERLANE-NEXT: vmovaps %xmm1, 32(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, (%rax)
; AVX2-FAST-PERLANE-NEXT: vzeroupper
; AVX2-FAST-PERLANE-NEXT: retq
;
; AVX512-LABEL: store_i32_stride6_vf2:
; AVX512: # %bb.0:
; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX512-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; AVX512-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX512-NEXT: vmovlhps {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX512-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; AVX512-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
; AVX512-NEXT: vmovlhps {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; AVX512-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
; AVX512-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
; AVX512-NEXT: vmovlhps {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; AVX512-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX512-NEXT: vinsertf64x4 $1, %ymm2, %zmm0, %zmm0
; AVX512-NEXT: vmovaps {{.*#+}} zmm1 = <0,2,4,6,8,10,1,3,5,7,9,11,u,u,u,u>
; AVX512-NEXT: vpermps %zmm0, %zmm1, %zmm0
; AVX512-NEXT: vextractf32x4 $2, %zmm0, 32(%rax)
; AVX512-NEXT: vmovaps %ymm0, (%rax)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%in.vec0 = load <2 x i32>, ptr %in.vecptr0, align 32
%in.vec1 = load <2 x i32>, ptr %in.vecptr1, align 32
%in.vec2 = load <2 x i32>, ptr %in.vecptr2, align 32
%in.vec3 = load <2 x i32>, ptr %in.vecptr3, align 32
%in.vec4 = load <2 x i32>, ptr %in.vecptr4, align 32
%in.vec5 = load <2 x i32>, ptr %in.vecptr5, align 32
%concat01 = shufflevector <2 x i32> %in.vec0, <2 x i32> %in.vec1, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
%concat23 = shufflevector <2 x i32> %in.vec2, <2 x i32> %in.vec3, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
%concat45 = shufflevector <2 x i32> %in.vec4, <2 x i32> %in.vec5, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
%concat0123 = shufflevector <4 x i32> %concat01, <4 x i32> %concat23, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%concat45uu = shufflevector <4 x i32> %concat45, <4 x i32> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
%concat012345 = shufflevector <8 x i32> %concat0123, <8 x i32> %concat45uu, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
%interleaved.vec = shufflevector <12 x i32> %concat012345, <12 x i32> poison, <12 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 1, i32 3, i32 5, i32 7, i32 9, i32 11>
store <12 x i32> %interleaved.vec, ptr %out.vec, align 32
ret void
}
define void @store_i32_stride6_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vecptr2, ptr %in.vecptr3, ptr %in.vecptr4, ptr %in.vecptr5, ptr %out.vec) nounwind {
; SSE-LABEL: store_i32_stride6_vf4:
; SSE: # %bb.0:
; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT: movaps (%rdi), %xmm0
; SSE-NEXT: movaps (%rsi), %xmm8
; SSE-NEXT: movaps (%rdx), %xmm1
; SSE-NEXT: movaps (%rcx), %xmm9
; SSE-NEXT: movaps (%r8), %xmm2
; SSE-NEXT: movaps (%r9), %xmm6
; SSE-NEXT: movaps %xmm1, %xmm3
; SSE-NEXT: unpcklps {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1]
; SSE-NEXT: movaps %xmm2, %xmm4
; SSE-NEXT: shufps {{.*#+}} xmm4 = xmm4[1,1],xmm6[1,1]
; SSE-NEXT: movaps %xmm2, %xmm5
; SSE-NEXT: unpckhpd {{.*#+}} xmm5 = xmm5[1],xmm6[1]
; SSE-NEXT: movaps %xmm2, %xmm7
; SSE-NEXT: shufps {{.*#+}} xmm7 = xmm7[3,3],xmm6[3,3]
; SSE-NEXT: movlhps {{.*#+}} xmm6 = xmm6[0],xmm2[0]
; SSE-NEXT: movaps %xmm0, %xmm2
; SSE-NEXT: unpcklps {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1]
; SSE-NEXT: shufps {{.*#+}} xmm6 = xmm6[2,0],xmm2[2,3]
; SSE-NEXT: movlhps {{.*#+}} xmm2 = xmm2[0],xmm3[0]
; SSE-NEXT: shufps {{.*#+}} xmm3 = xmm3[2,3],xmm4[0,2]
; SSE-NEXT: unpckhps {{.*#+}} xmm0 = xmm0[2],xmm8[2],xmm0[3],xmm8[3]
; SSE-NEXT: shufps {{.*#+}} xmm5 = xmm5[0,2],xmm0[2,3]
; SSE-NEXT: unpckhps {{.*#+}} xmm1 = xmm1[2],xmm9[2],xmm1[3],xmm9[3]
; SSE-NEXT: movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[2,3],xmm7[0,2]
; SSE-NEXT: movaps %xmm6, 16(%rax)
; SSE-NEXT: movaps %xmm0, 48(%rax)
; SSE-NEXT: movaps %xmm1, 80(%rax)
; SSE-NEXT: movaps %xmm5, 64(%rax)
; SSE-NEXT: movaps %xmm3, 32(%rax)
; SSE-NEXT: movaps %xmm2, (%rax)
; SSE-NEXT: retq
;
; AVX1-LABEL: store_i32_stride6_vf4:
; AVX1: # %bb.0:
; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX1-NEXT: vmovaps (%rdi), %xmm0
; AVX1-NEXT: vmovaps (%rsi), %xmm2
; AVX1-NEXT: vmovaps (%rdx), %xmm1
; AVX1-NEXT: vmovaps (%rcx), %xmm3
; AVX1-NEXT: vmovaps (%r8), %xmm4
; AVX1-NEXT: vmovaps (%r9), %xmm5
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm6
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm1, %ymm13
; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm4, %ymm8
; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm5, %ymm9
; AVX1-NEXT: vunpcklps {{.*#+}} ymm10 = ymm9[0],ymm8[0],ymm9[1],ymm8[1],ymm9[4],ymm8[4],ymm9[5],ymm8[5]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm2, %ymm11
; AVX1-NEXT: vunpcklpd {{.*#+}} ymm12 = ymm6[0],ymm11[0],ymm6[2],ymm11[2]
; AVX1-NEXT: vpermilps {{.*#+}} ymm12 = ymm12[0,2,3,1,4,6,7,5]
; AVX1-NEXT: vshufps {{.*#+}} xmm7 = xmm3[0,0],xmm1[0,0]
; AVX1-NEXT: vpermilps {{.*#+}} xmm7 = xmm7[0,1,2,0]
; AVX1-NEXT: vblendps {{.*#+}} ymm7 = ymm12[0,1],ymm7[2,3],ymm12[4,5,6,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3],ymm10[4,5],ymm7[6,7]
; AVX1-NEXT: vunpckhps {{.*#+}} ymm6 = ymm11[2],ymm6[2],ymm11[3],ymm6[3],ymm11[6],ymm6[6],ymm11[7],ymm6[7]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm3, %ymm10
; AVX1-NEXT: vshufps {{.*#+}} ymm10 = ymm13[1,2],ymm10[1,2],ymm13[5,6],ymm10[5,6]
; AVX1-NEXT: vpermilps {{.*#+}} ymm10 = ymm10[0,2,3,1,4,6,7,5]
; AVX1-NEXT: vblendps {{.*#+}} ymm6 = ymm10[0,1,2,3],ymm6[4,5],ymm10[6,7]
; AVX1-NEXT: vunpcklps {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1]
; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm6[0,1],ymm4[2,3],ymm6[4,5,6,7]
; AVX1-NEXT: vunpckhpd {{.*#+}} ymm5 = ymm8[1],ymm9[1],ymm8[3],ymm9[3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm5 = ymm5[0,2,3,1,4,6,7,5]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm2, %ymm2
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: vshufps {{.*#+}} ymm0 = ymm0[3,3],ymm2[3,3],ymm0[7,7],ymm2[7,7]
; AVX1-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,1,3,4,6,5,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm5[0,1],ymm0[2,3,4,5],ymm5[6,7]
; AVX1-NEXT: vmovaps %ymm0, 64(%rax)
; AVX1-NEXT: vmovaps %ymm4, 32(%rax)
; AVX1-NEXT: vmovaps %ymm7, (%rax)
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: store_i32_stride6_vf4:
; AVX2: # %bb.0:
; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-NEXT: vmovaps (%rdi), %xmm0
; AVX2-NEXT: vmovaps (%rsi), %xmm1
; AVX2-NEXT: vmovaps (%rdx), %xmm2
; AVX2-NEXT: vmovaps (%rcx), %xmm3
; AVX2-NEXT: vmovaps (%r8), %xmm4
; AVX2-NEXT: vmovaps (%r9), %xmm5
; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm6
; AVX2-NEXT: vinsertf128 $1, %xmm3, %ymm2, %ymm7
; AVX2-NEXT: vinsertf128 $1, %xmm5, %ymm4, %ymm8
; AVX2-NEXT: vmovaps {{.*#+}} xmm9 = <u,u,0,4>
; AVX2-NEXT: vpermps %ymm7, %ymm9, %ymm9
; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm10 = [0,4,1,5,0,4,1,5]
; AVX2-NEXT: # ymm10 = mem[0,1,0,1]
; AVX2-NEXT: vpermps %ymm6, %ymm10, %ymm10
; AVX2-NEXT: vblendps {{.*#+}} ymm9 = ymm10[0,1],ymm9[2,3],ymm10[4,5,6,7]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm10 = [17179869184,17179869184,17179869184,17179869184]
; AVX2-NEXT: vpermps %ymm8, %ymm10, %ymm10
; AVX2-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1,2,3],ymm10[4,5],ymm9[6,7]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm10 = [25769803778,25769803778,25769803778,25769803778]
; AVX2-NEXT: vpermps %ymm6, %ymm10, %ymm6
; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm10 = [1,5,2,6,1,5,2,6]
; AVX2-NEXT: # ymm10 = mem[0,1,0,1]
; AVX2-NEXT: vpermps %ymm7, %ymm10, %ymm7
; AVX2-NEXT: vblendps {{.*#+}} ymm6 = ymm7[0,1,2,3],ymm6[4,5],ymm7[6,7]
; AVX2-NEXT: vunpcklps {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1]
; AVX2-NEXT: vblendps {{.*#+}} ymm4 = ymm6[0,1],ymm4[2,3],ymm6[4,5,6,7]
; AVX2-NEXT: vinsertf128 $1, %xmm3, %ymm1, %ymm1
; AVX2-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX2-NEXT: vunpckhps {{.*#+}} ymm0 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
; AVX2-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2,6,3,7,2,6,3,7]
; AVX2-NEXT: # ymm1 = mem[0,1,0,1]
; AVX2-NEXT: vpermps %ymm8, %ymm1, %ymm1
; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5],ymm1[6,7]
; AVX2-NEXT: vmovaps %ymm0, 64(%rax)
; AVX2-NEXT: vmovaps %ymm4, 32(%rax)
; AVX2-NEXT: vmovaps %ymm9, (%rax)
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512-LABEL: store_i32_stride6_vf4:
; AVX512: # %bb.0:
; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX512-NEXT: vmovdqa (%rdi), %xmm0
; AVX512-NEXT: vmovdqa (%rdx), %xmm1
; AVX512-NEXT: vmovdqa (%r8), %xmm2
; AVX512-NEXT: vinserti128 $1, (%rcx), %ymm1, %ymm1
; AVX512-NEXT: vinserti128 $1, (%rsi), %ymm0, %ymm0
; AVX512-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-NEXT: vinserti32x4 $1, (%r9), %zmm2, %zmm1
; AVX512-NEXT: vmovdqa {{.*#+}} ymm2 = [18,22,3,7,11,15,19,23]
; AVX512-NEXT: vpermi2d %zmm1, %zmm0, %zmm2
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm3 = [0,4,8,12,16,20,1,5,9,13,17,21,2,6,10,14]
; AVX512-NEXT: vpermi2d %zmm1, %zmm0, %zmm3
; AVX512-NEXT: vmovdqu64 %zmm3, (%rax)
; AVX512-NEXT: vmovdqa %ymm2, 64(%rax)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%in.vec0 = load <4 x i32>, ptr %in.vecptr0, align 32
%in.vec1 = load <4 x i32>, ptr %in.vecptr1, align 32
%in.vec2 = load <4 x i32>, ptr %in.vecptr2, align 32
%in.vec3 = load <4 x i32>, ptr %in.vecptr3, align 32
%in.vec4 = load <4 x i32>, ptr %in.vecptr4, align 32
%in.vec5 = load <4 x i32>, ptr %in.vecptr5, align 32
%concat01 = shufflevector <4 x i32> %in.vec0, <4 x i32> %in.vec1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%concat23 = shufflevector <4 x i32> %in.vec2, <4 x i32> %in.vec3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%concat45 = shufflevector <4 x i32> %in.vec4, <4 x i32> %in.vec5, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%concat0123 = shufflevector <8 x i32> %concat01, <8 x i32> %concat23, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
%concat45uu = shufflevector <8 x i32> %concat45, <8 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
%concat012345 = shufflevector <16 x i32> %concat0123, <16 x i32> %concat45uu, <24 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
%interleaved.vec = shufflevector <24 x i32> %concat012345, <24 x i32> poison, <24 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 1, i32 5, i32 9, i32 13, i32 17, i32 21, i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 3, i32 7, i32 11, i32 15, i32 19, i32 23>
store <24 x i32> %interleaved.vec, ptr %out.vec, align 32
ret void
}
define void @store_i32_stride6_vf8(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vecptr2, ptr %in.vecptr3, ptr %in.vecptr4, ptr %in.vecptr5, ptr %out.vec) nounwind {
; SSE-LABEL: store_i32_stride6_vf8:
; SSE: # %bb.0:
; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT: movaps (%rdi), %xmm2
; SSE-NEXT: movaps 16(%rdi), %xmm8
; SSE-NEXT: movaps (%rsi), %xmm0
; SSE-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movaps 16(%rsi), %xmm13
; SSE-NEXT: movaps (%rdx), %xmm4
; SSE-NEXT: movaps 16(%rdx), %xmm1
; SSE-NEXT: movaps (%rcx), %xmm11
; SSE-NEXT: movaps 16(%rcx), %xmm15
; SSE-NEXT: movaps (%r8), %xmm10
; SSE-NEXT: movaps 16(%r8), %xmm0
; SSE-NEXT: movaps (%r9), %xmm7
; SSE-NEXT: movaps 16(%r9), %xmm5
; SSE-NEXT: movaps %xmm0, %xmm6
; SSE-NEXT: shufps {{.*#+}} xmm6 = xmm6[3,3],xmm5[3,3]
; SSE-NEXT: movaps %xmm1, %xmm9
; SSE-NEXT: unpckhps {{.*#+}} xmm9 = xmm9[2],xmm15[2],xmm9[3],xmm15[3]
; SSE-NEXT: shufps {{.*#+}} xmm9 = xmm9[2,3],xmm6[0,2]
; SSE-NEXT: movaps %xmm8, %xmm14
; SSE-NEXT: unpckhps {{.*#+}} xmm14 = xmm14[2],xmm13[2],xmm14[3],xmm13[3]
; SSE-NEXT: movaps %xmm0, %xmm12
; SSE-NEXT: unpckhpd {{.*#+}} xmm12 = xmm12[1],xmm5[1]
; SSE-NEXT: shufps {{.*#+}} xmm12 = xmm12[0,2],xmm14[2,3]
; SSE-NEXT: movaps %xmm15, %xmm3
; SSE-NEXT: unpckhpd {{.*#+}} xmm3 = xmm3[1],xmm1[1]
; SSE-NEXT: shufps {{.*#+}} xmm14 = xmm14[0,1],xmm3[2,0]
; SSE-NEXT: movaps %xmm0, %xmm3
; SSE-NEXT: shufps {{.*#+}} xmm3 = xmm3[1,1],xmm5[1,1]
; SSE-NEXT: unpcklps {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1]
; SSE-NEXT: unpcklps {{.*#+}} xmm8 = xmm8[0],xmm13[0],xmm8[1],xmm13[1]
; SSE-NEXT: movlhps {{.*#+}} xmm5 = xmm5[0],xmm0[0]
; SSE-NEXT: shufps {{.*#+}} xmm5 = xmm5[2,0],xmm8[2,3]
; SSE-NEXT: movlhps {{.*#+}} xmm8 = xmm8[0],xmm1[0]
; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[2,3],xmm3[0,2]
; SSE-NEXT: movaps %xmm10, %xmm0
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[3,3],xmm7[3,3]
; SSE-NEXT: movaps %xmm4, %xmm13
; SSE-NEXT: unpckhps {{.*#+}} xmm13 = xmm13[2],xmm11[2],xmm13[3],xmm11[3]
; SSE-NEXT: shufps {{.*#+}} xmm13 = xmm13[2,3],xmm0[0,2]
; SSE-NEXT: movaps %xmm2, %xmm0
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 16-byte Reload
; SSE-NEXT: unpckhps {{.*#+}} xmm0 = xmm0[2],xmm15[2],xmm0[3],xmm15[3]
; SSE-NEXT: movaps %xmm10, %xmm3
; SSE-NEXT: unpckhpd {{.*#+}} xmm3 = xmm3[1],xmm7[1]
; SSE-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,2],xmm0[2,3]
; SSE-NEXT: movaps %xmm11, %xmm6
; SSE-NEXT: unpckhpd {{.*#+}} xmm6 = xmm6[1],xmm4[1]
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm6[2,0]
; SSE-NEXT: movaps %xmm10, %xmm6
; SSE-NEXT: shufps {{.*#+}} xmm6 = xmm6[1,1],xmm7[1,1]
; SSE-NEXT: unpcklps {{.*#+}} xmm4 = xmm4[0],xmm11[0],xmm4[1],xmm11[1]
; SSE-NEXT: unpcklps {{.*#+}} xmm2 = xmm2[0],xmm15[0],xmm2[1],xmm15[1]
; SSE-NEXT: movlhps {{.*#+}} xmm7 = xmm7[0],xmm10[0]
; SSE-NEXT: shufps {{.*#+}} xmm7 = xmm7[2,0],xmm2[2,3]
; SSE-NEXT: movlhps {{.*#+}} xmm2 = xmm2[0],xmm4[0]
; SSE-NEXT: shufps {{.*#+}} xmm4 = xmm4[2,3],xmm6[0,2]
; SSE-NEXT: movaps %xmm2, (%rax)
; SSE-NEXT: movaps %xmm7, 16(%rax)
; SSE-NEXT: movaps %xmm4, 32(%rax)
; SSE-NEXT: movaps %xmm0, 48(%rax)
; SSE-NEXT: movaps %xmm3, 64(%rax)
; SSE-NEXT: movaps %xmm13, 80(%rax)
; SSE-NEXT: movaps %xmm8, 96(%rax)
; SSE-NEXT: movaps %xmm5, 112(%rax)
; SSE-NEXT: movaps %xmm1, 128(%rax)
; SSE-NEXT: movaps %xmm14, 144(%rax)
; SSE-NEXT: movaps %xmm12, 160(%rax)
; SSE-NEXT: movaps %xmm9, 176(%rax)
; SSE-NEXT: retq
;
; AVX1-LABEL: store_i32_stride6_vf8:
; AVX1: # %bb.0:
; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX1-NEXT: vmovaps (%rdi), %ymm8
; AVX1-NEXT: vmovaps (%rsi), %ymm11
; AVX1-NEXT: vmovaps (%rdx), %ymm5
; AVX1-NEXT: vmovaps (%rcx), %ymm6
; AVX1-NEXT: vmovaps (%r8), %ymm3
; AVX1-NEXT: vmovaps (%rcx), %xmm13
; AVX1-NEXT: vmovaps (%rdx), %xmm14
; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm14[1,2],xmm13[1,2]
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm9
; AVX1-NEXT: vmovaps (%rsi), %xmm0
; AVX1-NEXT: vmovaps (%rdi), %xmm7
; AVX1-NEXT: vunpckhps {{.*#+}} xmm4 = xmm7[2],xmm0[2],xmm7[3],xmm0[3]
; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm0, %ymm10
; AVX1-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1,2,3],ymm10[4,5],ymm9[6,7]
; AVX1-NEXT: vbroadcastss 4(%r8), %xmm10
; AVX1-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1],ymm10[2,3],ymm9[4,5,6,7]
; AVX1-NEXT: vbroadcastss 4(%r9), %ymm10
; AVX1-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1,2],ymm10[3],ymm9[4,5,6,7]
; AVX1-NEXT: vunpcklps {{.*#+}} ymm10 = ymm8[0],ymm11[0],ymm8[1],ymm11[1],ymm8[4],ymm11[4],ymm8[5],ymm11[5]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm10 = ymm10[2,3,2,3]
; AVX1-NEXT: vunpcklpd {{.*#+}} ymm12 = ymm6[0],ymm5[0],ymm6[2],ymm5[2]
; AVX1-NEXT: vpermilps {{.*#+}} ymm12 = ymm12[0,1,2,0,4,5,6,4]
; AVX1-NEXT: vextractf128 $1, %ymm12, %xmm12
; AVX1-NEXT: vblendps {{.*#+}} ymm10 = ymm10[0,1],ymm12[2,3],ymm10[4,5,6,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm10 = ymm10[0,1,2,3],ymm3[4,5],ymm10[6,7]
; AVX1-NEXT: vbroadcastss 16(%r9), %ymm12
; AVX1-NEXT: vblendps {{.*#+}} ymm10 = ymm10[0,1,2,3,4],ymm12[5],ymm10[6,7]
; AVX1-NEXT: vunpckhps {{.*#+}} ymm11 = ymm8[2],ymm11[2],ymm8[3],ymm11[3],ymm8[6],ymm11[6],ymm8[7],ymm11[7]
; AVX1-NEXT: vshufps {{.*#+}} ymm8 = ymm5[1,2],ymm6[1,2],ymm5[5,6],ymm6[5,6]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm8 = ymm8[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[0,2,1,3,4,6,5,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1,2,3],ymm11[4,5],ymm8[6,7]
; AVX1-NEXT: vbroadcastss 20(%r8), %xmm12
; AVX1-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1],ymm12[2,3],ymm8[4,5,6,7]
; AVX1-NEXT: vbroadcastss 20(%r9), %ymm12
; AVX1-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1,2],ymm12[3],ymm8[4,5,6,7]
; AVX1-NEXT: vbroadcastss (%rcx), %xmm1
; AVX1-NEXT: vbroadcastss (%rdx), %xmm2
; AVX1-NEXT: vunpcklps {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; AVX1-NEXT: vunpcklps {{.*#+}} xmm0 = xmm7[0],xmm0[0],xmm7[1],xmm0[1]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm2
; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1],ymm1[2,3],ymm2[4,5,6,7]
; AVX1-NEXT: vinsertf128 $1, (%r8), %ymm0, %ymm0
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
; AVX1-NEXT: vbroadcastss (%r9), %ymm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5],ymm0[6,7]
; AVX1-NEXT: vshufps {{.*#+}} ymm1 = ymm6[3,0],ymm5[3,0],ymm6[7,4],ymm5[7,4]
; AVX1-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[2,0,2,3,6,4,6,7]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm11[2,3],ymm1[2,3]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = ymm3[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm2 = ymm2[2,1,3,3,6,5,7,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1],ymm1[2,3,4,5],ymm2[6,7]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = mem[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm2[1],ymm1[2,3,4,5,6],ymm2[7]
; AVX1-NEXT: vunpckhps {{.*#+}} xmm2 = xmm14[2],xmm13[2],xmm14[3],xmm13[3]
; AVX1-NEXT: vpermilps {{.*#+}} xmm2 = xmm2[2,3,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm4, %ymm2
; AVX1-NEXT: vpermilps {{.*#+}} xmm3 = mem[2,1,3,3]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm3, %ymm3
; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3,4,5],ymm3[6,7]
; AVX1-NEXT: vmovaps (%r9), %xmm3
; AVX1-NEXT: vpermilps {{.*#+}} xmm4 = xmm3[0,2,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm4, %ymm3
; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0],ymm3[1],ymm2[2,3,4,5,6],ymm3[7]
; AVX1-NEXT: vmovaps %ymm2, 64(%rax)
; AVX1-NEXT: vmovaps %ymm1, 160(%rax)
; AVX1-NEXT: vmovaps %ymm0, (%rax)
; AVX1-NEXT: vmovaps %ymm8, 128(%rax)
; AVX1-NEXT: vmovaps %ymm10, 96(%rax)
; AVX1-NEXT: vmovaps %ymm9, 32(%rax)
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-SLOW-LABEL: store_i32_stride6_vf8:
; AVX2-SLOW: # %bb.0:
; AVX2-SLOW-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-SLOW-NEXT: vmovdqa (%rdi), %ymm10
; AVX2-SLOW-NEXT: vmovdqa (%rsi), %ymm12
; AVX2-SLOW-NEXT: vmovdqa (%rdx), %ymm3
; AVX2-SLOW-NEXT: vmovdqa (%rcx), %ymm4
; AVX2-SLOW-NEXT: vmovdqa (%r8), %ymm13
; AVX2-SLOW-NEXT: vmovdqa (%rsi), %xmm1
; AVX2-SLOW-NEXT: vmovdqa (%rdi), %xmm2
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} xmm14 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
; AVX2-SLOW-NEXT: vinserti128 $1, %xmm14, %ymm0, %ymm0
; AVX2-SLOW-NEXT: vmovdqa (%rcx), %xmm15
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm6 = xmm15[0,1,2,2]
; AVX2-SLOW-NEXT: vmovdqa (%rdx), %xmm5
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm7 = xmm5[1,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} xmm6 = xmm7[0],xmm6[1],xmm7[2],xmm6[3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm6 = ymm6[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1,2,3],ymm0[4,5],ymm6[6,7]
; AVX2-SLOW-NEXT: vmovdqa (%r8), %xmm0
; AVX2-SLOW-NEXT: vpmovzxdq {{.*#+}} xmm7 = xmm0[0],zero,xmm0[1],zero
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1],ymm7[2,3],ymm6[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 4(%r9), %ymm7
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm8 = ymm6[0,1,2],ymm7[3],ymm6[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastd (%rcx), %xmm6
; AVX2-SLOW-NEXT: vpbroadcastd (%rdx), %xmm7
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm6 = xmm7[0],xmm6[0],xmm7[1],xmm6[1]
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm6[2,3],ymm1[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastq %xmm0, %ymm2
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6,7]
; AVX2-SLOW-NEXT: vmovdqa (%r9), %xmm2
; AVX2-SLOW-NEXT: vpbroadcastd %xmm2, %ymm6
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm9 = ymm1[0,1,2,3,4],ymm6[5],ymm1[6,7]
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} ymm1 = ymm10[2],ymm12[2],ymm10[3],ymm12[3],ymm10[6],ymm12[6],ymm10[7],ymm12[7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm6 = ymm4[0,1,2,2,4,5,6,6]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm7 = ymm3[1,1,2,3,5,5,6,7]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm6 = ymm7[0],ymm6[1],ymm7[2],ymm6[3],ymm7[4],ymm6[5],ymm7[6],ymm6[7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1,2,3],ymm1[4,5],ymm6[6,7]
; AVX2-SLOW-NEXT: vpmovzxdq {{.*#+}} xmm7 = mem[0],zero,mem[1],zero
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1],ymm7[2,3],ymm6[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 20(%r9), %ymm7
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm11 = ymm6[0,1,2],ymm7[3],ymm6[4,5,6,7]
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} ymm6 = ymm3[2],ymm4[2],ymm3[3],ymm4[3],ymm3[6],ymm4[6],ymm3[7],ymm4[7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm6 = ymm6[2,3,2,3,6,7,6,7]
; AVX2-SLOW-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm1[2,3],ymm6[2,3]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm6 = ymm13[2,1,3,3,6,5,7,7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm6[0,1],ymm1[2,3,4,5],ymm6[6,7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm6 = mem[0,2,2,3,4,6,6,7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0],ymm6[1],ymm1[2,3,4,5,6],ymm6[7]
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} xmm5 = xmm5[2],xmm15[2],xmm5[3],xmm15[3]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm5[2,3,2,3]
; AVX2-SLOW-NEXT: vinserti128 $1, %xmm5, %ymm14, %ymm5
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm5[2,3,4,5],ymm0[6,7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[2,2,3,3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm2 = ymm2[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm2[1],ymm0[2,3,4,5,6],ymm2[7]
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} ymm2 = ymm3[0],ymm4[0],ymm3[1],ymm4[1],ymm3[4],ymm4[4],ymm3[5],ymm4[5]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm2 = ymm2[2,2,2,2]
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} ymm3 = ymm10[0],ymm12[0],ymm10[1],ymm12[1],ymm10[4],ymm12[4],ymm10[5],ymm12[5]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm3 = ymm3[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5,6,7]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm13[4,5],ymm2[6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 16(%r9), %ymm3
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4],ymm3[5],ymm2[6,7]
; AVX2-SLOW-NEXT: vmovdqa %ymm2, 96(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm0, 64(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm1, 160(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm11, 128(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm9, (%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm8, 32(%rax)
; AVX2-SLOW-NEXT: vzeroupper
; AVX2-SLOW-NEXT: retq
;
; AVX2-FAST-ALL-LABEL: store_i32_stride6_vf8:
; AVX2-FAST-ALL: # %bb.0:
; AVX2-FAST-ALL-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-FAST-ALL-NEXT: vmovdqa (%rdi), %ymm9
; AVX2-FAST-ALL-NEXT: vmovdqa (%rsi), %ymm11
; AVX2-FAST-ALL-NEXT: vmovdqa (%rdx), %ymm3
; AVX2-FAST-ALL-NEXT: vmovdqa (%rcx), %ymm4
; AVX2-FAST-ALL-NEXT: vmovdqa (%r8), %ymm2
; AVX2-FAST-ALL-NEXT: vmovdqa (%r9), %ymm12
; AVX2-FAST-ALL-NEXT: vmovdqa (%rsi), %xmm1
; AVX2-FAST-ALL-NEXT: vmovdqa (%rdi), %xmm5
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} xmm13 = xmm5[2],xmm1[2],xmm5[3],xmm1[3]
; AVX2-FAST-ALL-NEXT: vinserti128 $1, %xmm13, %ymm0, %ymm8
; AVX2-FAST-ALL-NEXT: vmovdqa (%rcx), %xmm14
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm7 = xmm14[0,1,2,2]
; AVX2-FAST-ALL-NEXT: vmovdqa (%rdx), %xmm15
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm6 = xmm15[1,1,2,3]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} xmm6 = xmm6[0],xmm7[1],xmm6[2],xmm7[3]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm6 = ymm6[0,1,2,1]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1,2,3],ymm8[4,5],ymm6[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqa (%r8), %xmm7
; AVX2-FAST-ALL-NEXT: vpmovzxdq {{.*#+}} xmm8 = xmm7[0],zero,xmm7[1],zero
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1],ymm8[2,3],ymm6[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 4(%r9), %ymm8
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm8 = ymm6[0,1,2],ymm8[3],ymm6[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd (%rcx), %xmm6
; AVX2-FAST-ALL-NEXT: vpbroadcastd (%rdx), %xmm0
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1]
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm5[0],xmm1[0],xmm5[1],xmm1[1]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastq %xmm7, %ymm1
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd (%r9), %ymm1
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm10 = ymm0[0,1,2,3,4],ymm1[5],ymm0[6,7]
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} ymm0 = ymm9[2],ymm11[2],ymm9[3],ymm11[3],ymm9[6],ymm11[6],ymm9[7],ymm11[7]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm1 = ymm4[0,1,2,2,4,5,6,6]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm5 = ymm3[1,1,2,3,5,5,6,7]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm5[0],ymm1[1],ymm5[2],ymm1[3],ymm5[4],ymm1[5],ymm5[6],ymm1[7]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[2,1,2,3]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
; AVX2-FAST-ALL-NEXT: vpmovzxdq {{.*#+}} xmm5 = mem[0],zero,mem[1],zero
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm5[2,3],ymm1[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 20(%r9), %ymm5
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm5[3],ymm1[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} ymm5 = ymm3[2],ymm4[2],ymm3[3],ymm4[3],ymm3[6],ymm4[6],ymm3[7],ymm4[7]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm5 = ymm5[2,3,2,3,6,7,6,7]
; AVX2-FAST-ALL-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],ymm5[2,3]
; AVX2-FAST-ALL-NEXT: vmovdqa {{.*#+}} ymm5 = <6,u,u,u,u,u,7,u>
; AVX2-FAST-ALL-NEXT: vpermd %ymm2, %ymm5, %ymm5
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm5[0,1],ymm0[2,3,4,5],ymm5[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqa {{.*#+}} ymm5 = <u,6,u,u,u,u,u,7>
; AVX2-FAST-ALL-NEXT: vpermd %ymm12, %ymm5, %ymm5
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm5[1],ymm0[2,3,4,5,6],ymm5[7]
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} xmm5 = xmm15[2],xmm14[2],xmm15[3],xmm14[3]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm5 = xmm5[2,3,2,3]
; AVX2-FAST-ALL-NEXT: vinserti128 $1, %xmm5, %ymm13, %ymm5
; AVX2-FAST-ALL-NEXT: vbroadcasti128 {{.*#+}} ymm6 = [2,2,3,3,2,2,3,3]
; AVX2-FAST-ALL-NEXT: # ymm6 = mem[0,1,0,1]
; AVX2-FAST-ALL-NEXT: vpermd %ymm2, %ymm6, %ymm7
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm5 = ymm7[0,1],ymm5[2,3,4,5],ymm7[6,7]
; AVX2-FAST-ALL-NEXT: vpermd %ymm12, %ymm6, %ymm6
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm5 = ymm5[0],ymm6[1],ymm5[2,3,4,5,6],ymm6[7]
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} ymm3 = ymm3[0],ymm4[0],ymm3[1],ymm4[1],ymm3[4],ymm4[4],ymm3[5],ymm4[5]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm3 = ymm3[2,2,2,2]
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} ymm4 = ymm9[0],ymm11[0],ymm9[1],ymm11[1],ymm9[4],ymm11[4],ymm9[5],ymm11[5]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm4 = ymm4[2,1,2,3]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm3 = ymm4[0,1],ymm3[2,3],ymm4[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm2 = ymm3[0,1,2,3],ymm2[4,5],ymm3[6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 16(%r9), %ymm3
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4],ymm3[5],ymm2[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm2, 96(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm5, 64(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm0, 160(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm1, 128(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm10, (%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm8, 32(%rax)
; AVX2-FAST-ALL-NEXT: vzeroupper
; AVX2-FAST-ALL-NEXT: retq
;
; AVX2-FAST-PERLANE-LABEL: store_i32_stride6_vf8:
; AVX2-FAST-PERLANE: # %bb.0:
; AVX2-FAST-PERLANE-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rdi), %ymm10
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rsi), %ymm12
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rdx), %ymm3
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rcx), %ymm4
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%r8), %ymm13
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rsi), %xmm1
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rdi), %xmm2
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} xmm14 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
; AVX2-FAST-PERLANE-NEXT: vinserti128 $1, %xmm14, %ymm0, %ymm0
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rcx), %xmm15
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm6 = xmm15[0,1,2,2]
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rdx), %xmm5
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm7 = xmm5[1,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} xmm6 = xmm7[0],xmm6[1],xmm7[2],xmm6[3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm6 = ymm6[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1,2,3],ymm0[4,5],ymm6[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%r8), %xmm0
; AVX2-FAST-PERLANE-NEXT: vpmovzxdq {{.*#+}} xmm7 = xmm0[0],zero,xmm0[1],zero
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1],ymm7[2,3],ymm6[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 4(%r9), %ymm7
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm8 = ymm6[0,1,2],ymm7[3],ymm6[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd (%rcx), %xmm6
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd (%rdx), %xmm7
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} xmm6 = xmm7[0],xmm6[0],xmm7[1],xmm6[1]
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm6[2,3],ymm1[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastq %xmm0, %ymm2
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%r9), %xmm2
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd %xmm2, %ymm6
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm9 = ymm1[0,1,2,3,4],ymm6[5],ymm1[6,7]
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} ymm1 = ymm10[2],ymm12[2],ymm10[3],ymm12[3],ymm10[6],ymm12[6],ymm10[7],ymm12[7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm6 = ymm4[0,1,2,2,4,5,6,6]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm7 = ymm3[1,1,2,3,5,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm6 = ymm7[0],ymm6[1],ymm7[2],ymm6[3],ymm7[4],ymm6[5],ymm7[6],ymm6[7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1,2,3],ymm1[4,5],ymm6[6,7]
; AVX2-FAST-PERLANE-NEXT: vpmovzxdq {{.*#+}} xmm7 = mem[0],zero,mem[1],zero
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1],ymm7[2,3],ymm6[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 20(%r9), %ymm7
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm11 = ymm6[0,1,2],ymm7[3],ymm6[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} ymm6 = ymm3[2],ymm4[2],ymm3[3],ymm4[3],ymm3[6],ymm4[6],ymm3[7],ymm4[7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm6 = ymm6[2,3,2,3,6,7,6,7]
; AVX2-FAST-PERLANE-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm1[2,3],ymm6[2,3]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm6 = ymm13[2,1,3,3,6,5,7,7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm6[0,1],ymm1[2,3,4,5],ymm6[6,7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm6 = mem[0,2,2,3,4,6,6,7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0],ymm6[1],ymm1[2,3,4,5,6],ymm6[7]
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} xmm5 = xmm5[2],xmm15[2],xmm5[3],xmm15[3]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm5 = xmm5[2,3,2,3]
; AVX2-FAST-PERLANE-NEXT: vinserti128 $1, %xmm5, %ymm14, %ymm5
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm5[2,3,4,5],ymm0[6,7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[2,2,3,3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm2 = ymm2[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm2[1],ymm0[2,3,4,5,6],ymm2[7]
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} ymm2 = ymm3[0],ymm4[0],ymm3[1],ymm4[1],ymm3[4],ymm4[4],ymm3[5],ymm4[5]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm2 = ymm2[2,2,2,2]
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} ymm3 = ymm10[0],ymm12[0],ymm10[1],ymm12[1],ymm10[4],ymm12[4],ymm10[5],ymm12[5]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm3 = ymm3[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm13[4,5],ymm2[6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 16(%r9), %ymm3
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4],ymm3[5],ymm2[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm2, 96(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm0, 64(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm1, 160(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm11, 128(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm9, (%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm8, 32(%rax)
; AVX2-FAST-PERLANE-NEXT: vzeroupper
; AVX2-FAST-PERLANE-NEXT: retq
;
; AVX512-LABEL: store_i32_stride6_vf8:
; AVX512: # %bb.0:
; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX512-NEXT: vmovdqa (%rdi), %ymm0
; AVX512-NEXT: vmovdqa (%rdx), %ymm1
; AVX512-NEXT: vmovdqa (%r8), %ymm2
; AVX512-NEXT: vinserti64x4 $1, (%rsi), %zmm0, %zmm0
; AVX512-NEXT: vinserti64x4 $1, (%rcx), %zmm1, %zmm1
; AVX512-NEXT: vinserti64x4 $1, (%r9), %zmm2, %zmm2
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm3 = <0,8,16,24,u,u,1,9,17,25,u,u,2,10,18,26>
; AVX512-NEXT: vpermi2d %zmm1, %zmm0, %zmm3
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm4 = [0,1,2,3,16,24,6,7,8,9,17,25,12,13,14,15]
; AVX512-NEXT: vpermi2d %zmm2, %zmm3, %zmm4
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm3 = <u,u,3,11,19,27,u,u,4,12,20,28,u,u,5,13>
; AVX512-NEXT: vpermi2d %zmm1, %zmm0, %zmm3
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm5 = [18,26,2,3,4,5,19,27,8,9,10,11,20,28,14,15]
; AVX512-NEXT: vpermi2d %zmm2, %zmm3, %zmm5
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm3 = <5,13,u,u,22,30,6,14,u,u,23,31,7,15,u,u>
; AVX512-NEXT: vpermi2d %zmm0, %zmm1, %zmm3
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm0 = [0,1,21,29,4,5,6,7,22,30,10,11,12,13,23,31]
; AVX512-NEXT: vpermi2d %zmm2, %zmm3, %zmm0
; AVX512-NEXT: vmovdqu64 %zmm0, 128(%rax)
; AVX512-NEXT: vmovdqu64 %zmm5, 64(%rax)
; AVX512-NEXT: vmovdqu64 %zmm4, (%rax)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%in.vec0 = load <8 x i32>, ptr %in.vecptr0, align 32
%in.vec1 = load <8 x i32>, ptr %in.vecptr1, align 32
%in.vec2 = load <8 x i32>, ptr %in.vecptr2, align 32
%in.vec3 = load <8 x i32>, ptr %in.vecptr3, align 32
%in.vec4 = load <8 x i32>, ptr %in.vecptr4, align 32
%in.vec5 = load <8 x i32>, ptr %in.vecptr5, align 32
%concat01 = shufflevector <8 x i32> %in.vec0, <8 x i32> %in.vec1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
%concat23 = shufflevector <8 x i32> %in.vec2, <8 x i32> %in.vec3, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
%concat45 = shufflevector <8 x i32> %in.vec4, <8 x i32> %in.vec5, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
%concat0123 = shufflevector <16 x i32> %concat01, <16 x i32> %concat23, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
%concat45uu = shufflevector <16 x i32> %concat45, <16 x i32> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
%concat012345 = shufflevector <32 x i32> %concat0123, <32 x i32> %concat45uu, <48 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47>
%interleaved.vec = shufflevector <48 x i32> %concat012345, <48 x i32> poison, <48 x i32> <i32 0, i32 8, i32 16, i32 24, i32 32, i32 40, i32 1, i32 9, i32 17, i32 25, i32 33, i32 41, i32 2, i32 10, i32 18, i32 26, i32 34, i32 42, i32 3, i32 11, i32 19, i32 27, i32 35, i32 43, i32 4, i32 12, i32 20, i32 28, i32 36, i32 44, i32 5, i32 13, i32 21, i32 29, i32 37, i32 45, i32 6, i32 14, i32 22, i32 30, i32 38, i32 46, i32 7, i32 15, i32 23, i32 31, i32 39, i32 47>
store <48 x i32> %interleaved.vec, ptr %out.vec, align 32
ret void
}
define void @store_i32_stride6_vf16(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vecptr2, ptr %in.vecptr3, ptr %in.vecptr4, ptr %in.vecptr5, ptr %out.vec) nounwind {
; SSE-LABEL: store_i32_stride6_vf16:
; SSE: # %bb.0:
; SSE-NEXT: subq $72, %rsp
; SSE-NEXT: movaps (%rdi), %xmm0
; SSE-NEXT: movaps 16(%rdi), %xmm10
; SSE-NEXT: movaps (%rsi), %xmm11
; SSE-NEXT: movaps 16(%rsi), %xmm8
; SSE-NEXT: movaps (%rdx), %xmm13
; SSE-NEXT: movaps 16(%rdx), %xmm12
; SSE-NEXT: movaps (%rcx), %xmm3
; SSE-NEXT: movaps 16(%rcx), %xmm9
; SSE-NEXT: movaps (%r8), %xmm2
; SSE-NEXT: movaps 16(%r8), %xmm7
; SSE-NEXT: movaps (%r9), %xmm1
; SSE-NEXT: movaps 16(%r9), %xmm14
; SSE-NEXT: movaps %xmm13, %xmm4
; SSE-NEXT: unpcklps {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; SSE-NEXT: movaps %xmm0, %xmm5
; SSE-NEXT: unpcklps {{.*#+}} xmm5 = xmm5[0],xmm11[0],xmm5[1],xmm11[1]
; SSE-NEXT: movaps %xmm1, %xmm6
; SSE-NEXT: movlhps {{.*#+}} xmm6 = xmm6[0],xmm2[0]
; SSE-NEXT: shufps {{.*#+}} xmm6 = xmm6[2,0],xmm5[2,3]
; SSE-NEXT: movaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movlhps {{.*#+}} xmm5 = xmm5[0],xmm4[0]
; SSE-NEXT: movaps %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movaps %xmm2, %xmm5
; SSE-NEXT: shufps {{.*#+}} xmm5 = xmm5[1,1],xmm1[1,1]
; SSE-NEXT: shufps {{.*#+}} xmm4 = xmm4[2,3],xmm5[0,2]
; SSE-NEXT: movaps %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movaps %xmm3, %xmm5
; SSE-NEXT: unpckhpd {{.*#+}} xmm5 = xmm5[1],xmm13[1]
; SSE-NEXT: unpckhps {{.*#+}} xmm0 = xmm0[2],xmm11[2],xmm0[3],xmm11[3]
; SSE-NEXT: movaps %xmm2, %xmm4
; SSE-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1],xmm1[1]
; SSE-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,2],xmm0[2,3]
; SSE-NEXT: movaps %xmm4, (%rsp) # 16-byte Spill
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm5[2,0]
; SSE-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: shufps {{.*#+}} xmm2 = xmm2[3,3],xmm1[3,3]
; SSE-NEXT: unpckhps {{.*#+}} xmm13 = xmm13[2],xmm3[2],xmm13[3],xmm3[3]
; SSE-NEXT: shufps {{.*#+}} xmm13 = xmm13[2,3],xmm2[0,2]
; SSE-NEXT: movaps %xmm13, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movaps %xmm12, %xmm1
; SSE-NEXT: unpcklps {{.*#+}} xmm1 = xmm1[0],xmm9[0],xmm1[1],xmm9[1]
; SSE-NEXT: movaps %xmm10, %xmm0
; SSE-NEXT: unpcklps {{.*#+}} xmm0 = xmm0[0],xmm8[0],xmm0[1],xmm8[1]
; SSE-NEXT: movaps %xmm14, %xmm2
; SSE-NEXT: movlhps {{.*#+}} xmm2 = xmm2[0],xmm7[0]
; SSE-NEXT: shufps {{.*#+}} xmm2 = xmm2[2,0],xmm0[2,3]
; SSE-NEXT: movaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; SSE-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movaps %xmm7, %xmm2
; SSE-NEXT: shufps {{.*#+}} xmm2 = xmm2[1,1],xmm14[1,1]
; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[2,3],xmm2[0,2]
; SSE-NEXT: movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movaps %xmm9, %xmm2
; SSE-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1],xmm12[1]
; SSE-NEXT: unpckhps {{.*#+}} xmm10 = xmm10[2],xmm8[2],xmm10[3],xmm8[3]
; SSE-NEXT: movaps %xmm7, %xmm0
; SSE-NEXT: unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm14[1]
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm10[2,3]
; SSE-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: shufps {{.*#+}} xmm10 = xmm10[0,1],xmm2[2,0]
; SSE-NEXT: movaps %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movaps 32(%rdi), %xmm10
; SSE-NEXT: shufps {{.*#+}} xmm7 = xmm7[3,3],xmm14[3,3]
; SSE-NEXT: movaps 32(%rdx), %xmm11
; SSE-NEXT: unpckhps {{.*#+}} xmm12 = xmm12[2],xmm9[2],xmm12[3],xmm9[3]
; SSE-NEXT: movaps 32(%rcx), %xmm0
; SSE-NEXT: shufps {{.*#+}} xmm12 = xmm12[2,3],xmm7[0,2]
; SSE-NEXT: movaps %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE-NEXT: movaps %xmm11, %xmm12
; SSE-NEXT: unpcklps {{.*#+}} xmm12 = xmm12[0],xmm0[0],xmm12[1],xmm0[1]
; SSE-NEXT: movaps 32(%rsi), %xmm1
; SSE-NEXT: movaps %xmm10, %xmm13
; SSE-NEXT: unpcklps {{.*#+}} xmm13 = xmm13[0],xmm1[0],xmm13[1],xmm1[1]
; SSE-NEXT: movaps 32(%r8), %xmm2
; SSE-NEXT: movaps 32(%r9), %xmm8
; SSE-NEXT: movaps %xmm8, %xmm15
; SSE-NEXT: movlhps {{.*#+}} xmm15 = xmm15[0],xmm2[0]
; SSE-NEXT: shufps {{.*#+}} xmm15 = xmm15[2,0],xmm13[2,3]
; SSE-NEXT: movlhps {{.*#+}} xmm13 = xmm13[0],xmm12[0]
; SSE-NEXT: movaps %xmm2, %xmm3
; SSE-NEXT: shufps {{.*#+}} xmm3 = xmm3[1,1],xmm8[1,1]
; SSE-NEXT: shufps {{.*#+}} xmm12 = xmm12[2,3],xmm3[0,2]
; SSE-NEXT: unpckhps {{.*#+}} xmm10 = xmm10[2],xmm1[2],xmm10[3],xmm1[3]
; SSE-NEXT: movaps %xmm0, %xmm1
; SSE-NEXT: unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm11[1]
; SSE-NEXT: movaps %xmm2, %xmm9
; SSE-NEXT: unpckhpd {{.*#+}} xmm9 = xmm9[1],xmm8[1]
; SSE-NEXT: shufps {{.*#+}} xmm9 = xmm9[0,2],xmm10[2,3]
; SSE-NEXT: shufps {{.*#+}} xmm10 = xmm10[0,1],xmm1[2,0]
; SSE-NEXT: shufps {{.*#+}} xmm2 = xmm2[3,3],xmm8[3,3]
; SSE-NEXT: unpckhps {{.*#+}} xmm11 = xmm11[2],xmm0[2],xmm11[3],xmm0[3]
; SSE-NEXT: shufps {{.*#+}} xmm11 = xmm11[2,3],xmm2[0,2]
; SSE-NEXT: movaps 48(%rdx), %xmm2
; SSE-NEXT: movaps 48(%rcx), %xmm8
; SSE-NEXT: movaps %xmm2, %xmm4
; SSE-NEXT: unpcklps {{.*#+}} xmm4 = xmm4[0],xmm8[0],xmm4[1],xmm8[1]
; SSE-NEXT: movaps 48(%rdi), %xmm0
; SSE-NEXT: movaps 48(%rsi), %xmm14
; SSE-NEXT: movaps %xmm0, %xmm5
; SSE-NEXT: unpcklps {{.*#+}} xmm5 = xmm5[0],xmm14[0],xmm5[1],xmm14[1]
; SSE-NEXT: movaps 48(%r8), %xmm3
; SSE-NEXT: movaps 48(%r9), %xmm7
; SSE-NEXT: movaps %xmm7, %xmm6
; SSE-NEXT: movlhps {{.*#+}} xmm6 = xmm6[0],xmm3[0]
; SSE-NEXT: shufps {{.*#+}} xmm6 = xmm6[2,0],xmm5[2,3]
; SSE-NEXT: movlhps {{.*#+}} xmm5 = xmm5[0],xmm4[0]
; SSE-NEXT: movaps %xmm3, %xmm1
; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,1],xmm7[1,1]
; SSE-NEXT: shufps {{.*#+}} xmm4 = xmm4[2,3],xmm1[0,2]
; SSE-NEXT: unpckhps {{.*#+}} xmm0 = xmm0[2],xmm14[2],xmm0[3],xmm14[3]
; SSE-NEXT: movaps %xmm8, %xmm14
; SSE-NEXT: unpckhpd {{.*#+}} xmm14 = xmm14[1],xmm2[1]
; SSE-NEXT: movaps %xmm3, %xmm1
; SSE-NEXT: unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm7[1]
; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[2,3]
; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm14[2,0]
; SSE-NEXT: shufps {{.*#+}} xmm3 = xmm3[3,3],xmm7[3,3]
; SSE-NEXT: unpckhps {{.*#+}} xmm2 = xmm2[2],xmm8[2],xmm2[3],xmm8[3]
; SSE-NEXT: shufps {{.*#+}} xmm2 = xmm2[2,3],xmm3[0,2]
; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
; SSE-NEXT: movaps %xmm2, 368(%rax)
; SSE-NEXT: movaps %xmm1, 352(%rax)
; SSE-NEXT: movaps %xmm0, 336(%rax)
; SSE-NEXT: movaps %xmm4, 320(%rax)
; SSE-NEXT: movaps %xmm6, 304(%rax)
; SSE-NEXT: movaps %xmm5, 288(%rax)
; SSE-NEXT: movaps %xmm11, 272(%rax)
; SSE-NEXT: movaps %xmm9, 256(%rax)
; SSE-NEXT: movaps %xmm10, 240(%rax)
; SSE-NEXT: movaps %xmm12, 224(%rax)
; SSE-NEXT: movaps %xmm15, 208(%rax)
; SSE-NEXT: movaps %xmm13, 192(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 176(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 160(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 144(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 128(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 112(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 96(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 80(%rax)
; SSE-NEXT: movaps (%rsp), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 64(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 48(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 32(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, 16(%rax)
; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE-NEXT: movaps %xmm0, (%rax)
; SSE-NEXT: addq $72, %rsp
; SSE-NEXT: retq
;
; AVX1-LABEL: store_i32_stride6_vf16:
; AVX1: # %bb.0:
; AVX1-NEXT: subq $136, %rsp
; AVX1-NEXT: vmovaps (%rdi), %ymm8
; AVX1-NEXT: vmovaps 32(%rdi), %ymm4
; AVX1-NEXT: vmovaps (%rsi), %ymm5
; AVX1-NEXT: vmovaps 32(%rsi), %ymm2
; AVX1-NEXT: vmovaps (%rdx), %ymm15
; AVX1-NEXT: vmovaps 32(%rdx), %ymm10
; AVX1-NEXT: vmovaps 32(%rcx), %ymm13
; AVX1-NEXT: vmovaps 32(%r8), %ymm3
; AVX1-NEXT: vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT: vunpcklps {{.*#+}} ymm0 = ymm4[0],ymm2[0],ymm4[1],ymm2[1],ymm4[4],ymm2[4],ymm4[5],ymm2[5]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: vunpcklpd {{.*#+}} ymm1 = ymm13[0],ymm10[0],ymm13[2],ymm10[2]
; AVX1-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,1,2,0,4,5,6,4]
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm3[4,5],ymm0[6,7]
; AVX1-NEXT: vbroadcastss 48(%r9), %ymm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5],ymm0[6,7]
; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT: vmovaps 32(%rcx), %xmm11
; AVX1-NEXT: vmovaps 32(%rdx), %xmm7
; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm7[1,2],xmm11[1,2]
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: vmovaps 32(%rsi), %xmm1
; AVX1-NEXT: vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT: vmovaps 32(%rdi), %xmm3
; AVX1-NEXT: vmovaps %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT: vunpckhps {{.*#+}} xmm6 = xmm3[2],xmm1[2],xmm3[3],xmm1[3]
; AVX1-NEXT: vinsertf128 $1, %xmm6, %ymm0, %ymm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
; AVX1-NEXT: vbroadcastss 36(%r8), %xmm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
; AVX1-NEXT: vbroadcastss 36(%r9), %ymm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3],ymm0[4,5,6,7]
; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT: vunpcklps {{.*#+}} ymm0 = ymm8[0],ymm5[0],ymm8[1],ymm5[1],ymm8[4],ymm5[4],ymm8[5],ymm5[5]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: vmovaps (%rcx), %ymm9
; AVX1-NEXT: vunpcklpd {{.*#+}} ymm1 = ymm9[0],ymm15[0],ymm9[2],ymm15[2]
; AVX1-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,1,2,0,4,5,6,4]
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
; AVX1-NEXT: vmovaps (%r8), %ymm1
; AVX1-NEXT: vmovups %ymm1, (%rsp) # 32-byte Spill
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
; AVX1-NEXT: vbroadcastss 16(%r9), %ymm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5],ymm0[6,7]
; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT: vmovaps (%rcx), %xmm0
; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT: vmovaps (%rdx), %xmm1
; AVX1-NEXT: vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm1[1,2],xmm0[1,2]
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,1,3]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: vmovaps (%rsi), %xmm3
; AVX1-NEXT: vmovaps (%rdi), %xmm1
; AVX1-NEXT: vunpckhps {{.*#+}} xmm14 = xmm1[2],xmm3[2],xmm1[3],xmm3[3]
; AVX1-NEXT: vinsertf128 $1, %xmm14, %ymm0, %ymm12
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm12[4,5],ymm0[6,7]
; AVX1-NEXT: vbroadcastss 4(%r8), %xmm12
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm12[2,3],ymm0[4,5,6,7]
; AVX1-NEXT: vbroadcastss 4(%r9), %ymm12
; AVX1-NEXT: vblendps {{.*#+}} ymm12 = ymm0[0,1,2],ymm12[3],ymm0[4,5,6,7]
; AVX1-NEXT: vunpckhps {{.*#+}} ymm2 = ymm4[2],ymm2[2],ymm4[3],ymm2[3],ymm4[6],ymm2[6],ymm4[7],ymm2[7]
; AVX1-NEXT: vshufps {{.*#+}} ymm0 = ymm10[1,2],ymm13[1,2],ymm10[5,6],ymm13[5,6]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,1,3,4,6,5,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm2[4,5],ymm0[6,7]
; AVX1-NEXT: vbroadcastss 52(%r8), %xmm4
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm4[2,3],ymm0[4,5,6,7]
; AVX1-NEXT: vbroadcastss 52(%r9), %ymm4
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm4[3],ymm0[4,5,6,7]
; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX1-NEXT: vunpckhps {{.*#+}} ymm4 = ymm8[2],ymm5[2],ymm8[3],ymm5[3],ymm8[6],ymm5[6],ymm8[7],ymm5[7]
; AVX1-NEXT: vshufps {{.*#+}} ymm0 = ymm15[1,2],ymm9[1,2],ymm15[5,6],ymm9[5,6]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,1,3,4,6,5,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm4[4,5],ymm0[6,7]
; AVX1-NEXT: vbroadcastss 20(%r8), %xmm5
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm5[2,3],ymm0[4,5,6,7]
; AVX1-NEXT: vbroadcastss 20(%r9), %ymm5
; AVX1-NEXT: vblendps {{.*#+}} ymm8 = ymm0[0,1,2],ymm5[3],ymm0[4,5,6,7]
; AVX1-NEXT: vbroadcastss (%rcx), %xmm0
; AVX1-NEXT: vbroadcastss (%rdx), %xmm5
; AVX1-NEXT: vunpcklps {{.*#+}} xmm0 = xmm5[0],xmm0[0],xmm5[1],xmm0[1]
; AVX1-NEXT: vunpcklps {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm3
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm3[0,1],ymm0[2,3],ymm3[4,5,6,7]
; AVX1-NEXT: vinsertf128 $1, (%r8), %ymm1, %ymm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
; AVX1-NEXT: vbroadcastss (%r9), %ymm1
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5],ymm0[6,7]
; AVX1-NEXT: vshufps {{.*#+}} ymm1 = ymm13[3,0],ymm10[3,0],ymm13[7,4],ymm10[7,4]
; AVX1-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[2,0,2,3,6,4,6,7]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm2[2,3],ymm1[2,3]
; AVX1-NEXT: vperm2f128 $51, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm2 # 32-byte Folded Reload
; AVX1-NEXT: # ymm2 = mem[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm2 = ymm2[2,1,3,3,6,5,7,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1],ymm1[2,3,4,5],ymm2[6,7]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm2 = mem[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm1[0],ymm2[1],ymm1[2,3,4,5,6],ymm2[7]
; AVX1-NEXT: vunpckhps {{.*#+}} xmm1 = xmm7[2],xmm11[2],xmm7[3],xmm11[3]
; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm1[2,3,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm6, %ymm1
; AVX1-NEXT: vpermilps {{.*#+}} xmm3 = mem[2,1,3,3]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm3, %ymm3
; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm3[0,1],ymm1[2,3,4,5],ymm3[6,7]
; AVX1-NEXT: vmovaps 32(%r9), %xmm3
; AVX1-NEXT: vpermilps {{.*#+}} xmm5 = xmm3[0,2,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm5, %ymm3
; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm1[0],ymm3[1],ymm1[2,3,4,5,6],ymm3[7]
; AVX1-NEXT: vbroadcastss 32(%rcx), %xmm1
; AVX1-NEXT: vbroadcastss 32(%rdx), %xmm5
; AVX1-NEXT: vunpcklps {{.*#+}} xmm1 = xmm5[0],xmm1[0],xmm5[1],xmm1[1]
; AVX1-NEXT: vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; AVX1-NEXT: vunpcklps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5, %xmm5 # 16-byte Folded Reload
; AVX1-NEXT: # xmm5 = xmm5[0],mem[0],xmm5[1],mem[1]
; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm0, %ymm6
; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm6[0,1],ymm1[2,3],ymm6[4,5,6,7]
; AVX1-NEXT: vinsertf128 $1, 32(%r8), %ymm5, %ymm5
; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm5[0,1],ymm1[2,3],ymm5[4,5],ymm1[6,7]
; AVX1-NEXT: vbroadcastss 32(%r9), %ymm5
; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm5[5],ymm1[6,7]
; AVX1-NEXT: vshufps {{.*#+}} ymm5 = ymm9[3,0],ymm15[3,0],ymm9[7,4],ymm15[7,4]
; AVX1-NEXT: vpermilps {{.*#+}} ymm5 = ymm5[2,0,2,3,6,4,6,7]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm4 = ymm4[2,3],ymm5[2,3]
; AVX1-NEXT: vperm2f128 $51, (%rsp), %ymm0, %ymm5 # 32-byte Folded Reload
; AVX1-NEXT: # ymm5 = mem[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm5 = ymm5[2,1,3,3,6,5,7,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm5[0,1],ymm4[2,3,4,5],ymm5[6,7]
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm5 = mem[2,3,2,3]
; AVX1-NEXT: vpermilps {{.*#+}} ymm5 = ymm5[0,2,2,3,4,6,6,7]
; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0],ymm5[1],ymm4[2,3,4,5,6],ymm5[7]
; AVX1-NEXT: vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
; AVX1-NEXT: vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm5, %xmm5 # 16-byte Folded Reload
; AVX1-NEXT: # xmm5 = xmm5[2],mem[2],xmm5[3],mem[3]
; AVX1-NEXT: vmovaps (%r9), %xmm6
; AVX1-NEXT: vpermilps {{.*#+}} xmm5 = xmm5[2,3,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm14, %ymm5
; AVX1-NEXT: vpermilps {{.*#+}} xmm7 = mem[2,1,3,3]
; AVX1-NEXT: vinsertf128 $1, %xmm7, %ymm7, %ymm7
; AVX1-NEXT: vblendps {{.*#+}} ymm5 = ymm7[0,1],ymm5[2,3,4,5],ymm7[6,7]
; AVX1-NEXT: vpermilps {{.*#+}} xmm7 = xmm6[0,2,2,3]
; AVX1-NEXT: vinsertf128 $1, %xmm6, %ymm7, %ymm6
; AVX1-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0],ymm6[1],ymm5[2,3,4,5,6],ymm6[7]
; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX1-NEXT: vmovaps %ymm5, 64(%rax)
; AVX1-NEXT: vmovaps %ymm4, 160(%rax)
; AVX1-NEXT: vmovaps %ymm1, 192(%rax)
; AVX1-NEXT: vmovaps %ymm3, 256(%rax)
; AVX1-NEXT: vmovaps %ymm2, 352(%rax)
; AVX1-NEXT: vmovaps %ymm0, (%rax)
; AVX1-NEXT: vmovaps %ymm8, 128(%rax)
; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-NEXT: vmovaps %ymm0, 320(%rax)
; AVX1-NEXT: vmovaps %ymm12, 32(%rax)
; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-NEXT: vmovaps %ymm0, 96(%rax)
; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-NEXT: vmovaps %ymm0, 224(%rax)
; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX1-NEXT: vmovaps %ymm0, 288(%rax)
; AVX1-NEXT: addq $136, %rsp
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-SLOW-LABEL: store_i32_stride6_vf16:
; AVX2-SLOW: # %bb.0:
; AVX2-SLOW-NEXT: subq $200, %rsp
; AVX2-SLOW-NEXT: vmovdqa (%rsi), %xmm0
; AVX2-SLOW-NEXT: vmovdqa 32(%rsi), %xmm8
; AVX2-SLOW-NEXT: vmovdqa (%rdi), %xmm2
; AVX2-SLOW-NEXT: vmovdqa 32(%rdi), %xmm10
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm10[2],xmm8[2],xmm10[3],xmm8[3]
; AVX2-SLOW-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-SLOW-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX2-SLOW-NEXT: vmovdqa (%rcx), %xmm6
; AVX2-SLOW-NEXT: vmovdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-SLOW-NEXT: vmovdqa 32(%rcx), %xmm4
; AVX2-SLOW-NEXT: vmovdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm4 = xmm4[0,1,2,2]
; AVX2-SLOW-NEXT: vmovdqa (%rdx), %xmm7
; AVX2-SLOW-NEXT: vmovdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-SLOW-NEXT: vmovdqa 32(%rdx), %xmm1
; AVX2-SLOW-NEXT: vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm1[1,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} xmm4 = xmm5[0],xmm4[1],xmm5[2],xmm4[3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm4 = ymm4[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm3[4,5],ymm4[6,7]
; AVX2-SLOW-NEXT: vmovdqa (%r8), %xmm12
; AVX2-SLOW-NEXT: vmovdqa 32(%r8), %xmm11
; AVX2-SLOW-NEXT: vpmovzxdq {{.*#+}} xmm4 = xmm11[0],zero,xmm11[1],zero
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3],ymm3[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 36(%r9), %ymm4
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm4[3],ymm3[4,5,6,7]
; AVX2-SLOW-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; AVX2-SLOW-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-SLOW-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm4 = xmm6[0,1,2,2]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm7[1,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} xmm4 = xmm5[0],xmm4[1],xmm5[2],xmm4[3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm4 = ymm4[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm3[4,5],ymm4[6,7]
; AVX2-SLOW-NEXT: vpmovzxdq {{.*#+}} xmm4 = xmm12[0],zero,xmm12[1],zero
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3],ymm3[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 4(%r9), %ymm4
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm4[3],ymm3[4,5,6,7]
; AVX2-SLOW-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-SLOW-NEXT: vpbroadcastd (%rcx), %xmm3
; AVX2-SLOW-NEXT: vpbroadcastd (%rdx), %xmm4
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; AVX2-SLOW-NEXT: vmovdqa 32(%rdx), %ymm5
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm4[2,3],ymm0[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastq %xmm12, %ymm2
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm2[4,5],ymm0[6,7]
; AVX2-SLOW-NEXT: vmovdqa (%r9), %xmm15
; AVX2-SLOW-NEXT: vpbroadcastd %xmm15, %ymm2
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm2[5],ymm0[6,7]
; AVX2-SLOW-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-SLOW-NEXT: vmovdqa 32(%rcx), %ymm3
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm2 = ymm3[0,1,2,2,4,5,6,6]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm4 = ymm5[1,1,2,3,5,5,6,7]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm2 = ymm4[0],ymm2[1],ymm4[2],ymm2[3],ymm4[4],ymm2[5],ymm4[6],ymm2[7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm2 = ymm2[2,1,2,3]
; AVX2-SLOW-NEXT: vmovdqa 32(%rdi), %ymm9
; AVX2-SLOW-NEXT: vmovdqa 32(%rsi), %ymm7
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} ymm13 = ymm9[2],ymm7[2],ymm9[3],ymm7[3],ymm9[6],ymm7[6],ymm9[7],ymm7[7]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm13[4,5],ymm2[6,7]
; AVX2-SLOW-NEXT: vpmovzxdq {{.*#+}} xmm4 = mem[0],zero,mem[1],zero
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1],ymm4[2,3],ymm2[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 52(%r9), %ymm4
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm2[0,1,2],ymm4[3],ymm2[4,5,6,7]
; AVX2-SLOW-NEXT: vmovdqu %ymm0, (%rsp) # 32-byte Spill
; AVX2-SLOW-NEXT: vpbroadcastd 32(%rcx), %xmm2
; AVX2-SLOW-NEXT: vpbroadcastd 32(%rdx), %xmm4
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm4[0],xmm2[0],xmm4[1],xmm2[1]
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm10[0],xmm8[0],xmm10[1],xmm8[1]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm2[2,3],ymm1[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastq %xmm11, %ymm2
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6,7]
; AVX2-SLOW-NEXT: vmovdqa 32(%r9), %xmm6
; AVX2-SLOW-NEXT: vpbroadcastd %xmm6, %ymm2
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1,2,3,4],ymm2[5],ymm1[6,7]
; AVX2-SLOW-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-SLOW-NEXT: vmovdqa (%rdx), %ymm2
; AVX2-SLOW-NEXT: vmovdqa (%rcx), %ymm1
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm4 = ymm1[0,1,2,2,4,5,6,6]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm8 = ymm2[1,1,2,3,5,5,6,7]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm4 = ymm8[0],ymm4[1],ymm8[2],ymm4[3],ymm8[4],ymm4[5],ymm8[6],ymm4[7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm4[2,1,2,3]
; AVX2-SLOW-NEXT: vmovdqa (%rdi), %ymm10
; AVX2-SLOW-NEXT: vmovdqa (%rsi), %ymm8
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} ymm4 = ymm10[2],ymm8[2],ymm10[3],ymm8[3],ymm10[6],ymm8[6],ymm10[7],ymm8[7]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm4[4,5],ymm0[6,7]
; AVX2-SLOW-NEXT: vpmovzxdq {{.*#+}} xmm14 = mem[0],zero,mem[1],zero
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm14[2,3],ymm0[4,5,6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 20(%r9), %ymm14
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm14[3],ymm0[4,5,6,7]
; AVX2-SLOW-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} ymm0 = ymm5[2],ymm3[2],ymm5[3],ymm3[3],ymm5[6],ymm3[6],ymm5[7],ymm3[7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
; AVX2-SLOW-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm13[2,3],ymm0[2,3]
; AVX2-SLOW-NEXT: vmovdqa 32(%r8), %ymm14
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm13 = ymm14[2,1,3,3,6,5,7,7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm13 = ymm13[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm13[0,1],ymm0[2,3,4,5],ymm13[6,7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm13 = mem[0,2,2,3,4,6,6,7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm13 = ymm13[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm13 = ymm0[0],ymm13[1],ymm0[2,3,4,5,6],ymm13[7]
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} ymm0 = ymm5[0],ymm3[0],ymm5[1],ymm3[1],ymm5[4],ymm3[4],ymm5[5],ymm3[5]
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} ymm3 = ymm9[0],ymm7[0],ymm9[1],ymm7[1],ymm9[4],ymm7[4],ymm9[5],ymm7[5]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,2]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm3 = ymm3[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm3[0,1],ymm0[2,3],ymm3[4,5,6,7]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm14[4,5],ymm0[6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 48(%r9), %ymm3
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm3[5],ymm0[6,7]
; AVX2-SLOW-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; AVX2-SLOW-NEXT: vpunpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 16-byte Folded Reload
; AVX2-SLOW-NEXT: # xmm3 = xmm3[2],mem[2],xmm3[3],mem[3]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[2,3,2,3]
; AVX2-SLOW-NEXT: vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-SLOW-NEXT: vinserti128 $1, %xmm3, %ymm5, %ymm3
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm11[2,2,3,3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm3 = ymm5[0,1],ymm3[2,3,4,5],ymm5[6,7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm6[2,2,3,3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0],ymm5[1],ymm3[2,3,4,5,6],ymm5[7]
; AVX2-SLOW-NEXT: vpunpckhdq {{.*#+}} ymm5 = ymm2[2],ymm1[2],ymm2[3],ymm1[3],ymm2[6],ymm1[6],ymm2[7],ymm1[7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm5 = ymm5[2,3,2,3,6,7,6,7]
; AVX2-SLOW-NEXT: vperm2i128 {{.*#+}} ymm4 = ymm4[2,3],ymm5[2,3]
; AVX2-SLOW-NEXT: vmovdqa (%r8), %ymm5
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm6 = ymm5[2,1,3,3,6,5,7,7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm4 = ymm6[0,1],ymm4[2,3,4,5],ymm6[6,7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} ymm6 = mem[0,2,2,3,4,6,6,7]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm4 = ymm4[0],ymm6[1],ymm4[2,3,4,5,6],ymm6[7]
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} ymm1 = ymm2[0],ymm1[0],ymm2[1],ymm1[1],ymm2[4],ymm1[4],ymm2[5],ymm1[5]
; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} ymm2 = ymm10[0],ymm8[0],ymm10[1],ymm8[1],ymm10[4],ymm8[4],ymm10[5],ymm8[5]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm1 = ymm1[2,2,2,2]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm2 = ymm2[2,1,2,3]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm2[0,1],ymm1[2,3],ymm2[4,5,6,7]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm5[4,5],ymm1[6,7]
; AVX2-SLOW-NEXT: vpbroadcastd 16(%r9), %ymm2
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm2[5],ymm1[6,7]
; AVX2-SLOW-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX2-SLOW-NEXT: vpunpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 16-byte Folded Reload
; AVX2-SLOW-NEXT: # xmm2 = xmm2[2],mem[2],xmm2[3],mem[3]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[2,3,2,3]
; AVX2-SLOW-NEXT: vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-SLOW-NEXT: vinserti128 $1, %xmm2, %ymm5, %ymm2
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm12[2,2,3,3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm2 = ymm5[0,1],ymm2[2,3,4,5],ymm5[6,7]
; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm5 = xmm15[2,2,3,3]
; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,1,2,1]
; AVX2-SLOW-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0],ymm5[1],ymm2[2,3,4,5,6],ymm5[7]
; AVX2-SLOW-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-SLOW-NEXT: vmovdqa %ymm2, 64(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm1, 96(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm4, 160(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm3, 256(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm0, 288(%rax)
; AVX2-SLOW-NEXT: vmovdqa %ymm13, 352(%rax)
; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-SLOW-NEXT: vmovaps %ymm0, 128(%rax)
; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-SLOW-NEXT: vmovaps %ymm0, 192(%rax)
; AVX2-SLOW-NEXT: vmovups (%rsp), %ymm0 # 32-byte Reload
; AVX2-SLOW-NEXT: vmovaps %ymm0, 320(%rax)
; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-SLOW-NEXT: vmovaps %ymm0, (%rax)
; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-SLOW-NEXT: vmovaps %ymm0, 32(%rax)
; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-SLOW-NEXT: vmovaps %ymm0, 224(%rax)
; AVX2-SLOW-NEXT: addq $200, %rsp
; AVX2-SLOW-NEXT: vzeroupper
; AVX2-SLOW-NEXT: retq
;
; AVX2-FAST-ALL-LABEL: store_i32_stride6_vf16:
; AVX2-FAST-ALL: # %bb.0:
; AVX2-FAST-ALL-NEXT: subq $184, %rsp
; AVX2-FAST-ALL-NEXT: vmovdqa (%rsi), %xmm9
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%rsi), %xmm10
; AVX2-FAST-ALL-NEXT: vmovdqa (%rdi), %xmm2
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%rdi), %xmm8
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} xmm0 = xmm8[2],xmm10[2],xmm8[3],xmm10[3]
; AVX2-FAST-ALL-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-ALL-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX2-FAST-ALL-NEXT: vmovdqa (%rcx), %xmm7
; AVX2-FAST-ALL-NEXT: vmovdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%rcx), %xmm1
; AVX2-FAST-ALL-NEXT: vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm4 = xmm1[0,1,2,2]
; AVX2-FAST-ALL-NEXT: vmovdqa (%rdx), %xmm1
; AVX2-FAST-ALL-NEXT: vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%rdx), %xmm11
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm5 = xmm11[1,1,2,3]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} xmm4 = xmm5[0],xmm4[1],xmm5[2],xmm4[3]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm4 = ymm4[0,1,2,1]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm4 = ymm4[0,1,2,3],ymm0[4,5],ymm4[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqa (%r8), %xmm5
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%r8), %xmm0
; AVX2-FAST-ALL-NEXT: vpmovzxdq {{.*#+}} xmm6 = xmm0[0],zero,xmm0[1],zero
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm4 = ymm4[0,1],ymm6[2,3],ymm4[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 36(%r9), %ymm6
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm3 = ymm4[0,1,2],ymm6[3],ymm4[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm2[2],xmm9[2],xmm2[3],xmm9[3]
; AVX2-FAST-ALL-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-ALL-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm4
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm6 = xmm7[0,1,2,2]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm3 = xmm1[1,1,2,3]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} xmm3 = xmm3[0],xmm6[1],xmm3[2],xmm6[3]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm3 = ymm3[0,1,2,1]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
; AVX2-FAST-ALL-NEXT: vpmovzxdq {{.*#+}} xmm4 = xmm5[0],zero,xmm5[1],zero
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3],ymm3[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 4(%r9), %ymm4
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm3[0,1,2],ymm4[3],ymm3[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-ALL-NEXT: vpbroadcastd (%rcx), %xmm3
; AVX2-FAST-ALL-NEXT: vpbroadcastd (%rdx), %xmm4
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%rdx), %ymm13
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm2[0],xmm9[0],xmm2[1],xmm9[1]
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%rcx), %ymm14
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm3[2,3],ymm1[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastq %xmm5, %ymm2
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd (%r9), %ymm2
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm2[5],ymm1[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm1 = ymm14[0,1,2,2,4,5,6,6]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm2 = ymm13[1,1,2,3,5,5,6,7]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm2[0],ymm1[1],ymm2[2],ymm1[3],ymm2[4],ymm1[5],ymm2[6],ymm1[7]
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%rdi), %ymm4
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm1 = ymm1[2,1,2,3]
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%rsi), %ymm2
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} ymm6 = ymm4[2],ymm2[2],ymm4[3],ymm2[3],ymm4[6],ymm2[6],ymm4[7],ymm2[7]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm6[4,5],ymm1[6,7]
; AVX2-FAST-ALL-NEXT: vpmovzxdq {{.*#+}} xmm3 = mem[0],zero,mem[1],zero
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm3[2,3],ymm1[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 52(%r9), %ymm3
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm3[3],ymm1[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-ALL-NEXT: vpbroadcastd 32(%rcx), %xmm1
; AVX2-FAST-ALL-NEXT: vpbroadcastd 32(%rdx), %xmm3
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm8[0],xmm10[0],xmm8[1],xmm10[1]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm3 = ymm3[0,1,2,1]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm3[0,1],ymm1[2,3],ymm3[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastq %xmm0, %ymm0
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 32(%r9), %ymm1
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5],ymm0[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-ALL-NEXT: vmovdqa (%rdx), %ymm7
; AVX2-FAST-ALL-NEXT: vmovdqa (%rcx), %ymm0
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm1 = ymm0[0,1,2,2,4,5,6,6]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm3 = ymm7[1,1,2,3,5,5,6,7]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm3[0],ymm1[1],ymm3[2],ymm1[3],ymm3[4],ymm1[5],ymm3[6],ymm1[7]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm10 = ymm1[2,1,2,3]
; AVX2-FAST-ALL-NEXT: vmovdqa (%rdi), %ymm5
; AVX2-FAST-ALL-NEXT: vmovdqa (%rsi), %ymm3
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} ymm1 = ymm5[2],ymm3[2],ymm5[3],ymm3[3],ymm5[6],ymm3[6],ymm5[7],ymm3[7]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm10 = ymm10[0,1,2,3],ymm1[4,5],ymm10[6,7]
; AVX2-FAST-ALL-NEXT: vpmovzxdq {{.*#+}} xmm15 = mem[0],zero,mem[1],zero
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm10 = ymm10[0,1],ymm15[2,3],ymm10[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 20(%r9), %ymm15
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm8 = ymm10[0,1,2],ymm15[3],ymm10[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vmovdqu %ymm8, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} ymm15 = ymm13[2],ymm14[2],ymm13[3],ymm14[3],ymm13[6],ymm14[6],ymm13[7],ymm14[7]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm15 = ymm15[2,3,2,3,6,7,6,7]
; AVX2-FAST-ALL-NEXT: vperm2i128 {{.*#+}} ymm6 = ymm6[2,3],ymm15[2,3]
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%r8), %ymm15
; AVX2-FAST-ALL-NEXT: vmovdqa {{.*#+}} ymm9 = [6,5,3,3,6,5,7,7]
; AVX2-FAST-ALL-NEXT: vpermd %ymm15, %ymm9, %ymm12
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm6 = ymm12[0,1],ymm6[2,3,4,5],ymm12[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqa 32(%r9), %ymm12
; AVX2-FAST-ALL-NEXT: vmovdqa {{.*#+}} ymm8 = [4,6,2,3,4,6,6,7]
; AVX2-FAST-ALL-NEXT: vpermd %ymm12, %ymm8, %ymm10
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0],ymm10[1],ymm6[2,3,4,5,6],ymm10[7]
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} ymm10 = ymm13[0],ymm14[0],ymm13[1],ymm14[1],ymm13[4],ymm14[4],ymm13[5],ymm14[5]
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} ymm2 = ymm4[0],ymm2[0],ymm4[1],ymm2[1],ymm4[4],ymm2[4],ymm4[5],ymm2[5]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm4 = ymm10[2,2,2,2]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm2 = ymm2[2,1,2,3]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1],ymm4[2,3],ymm2[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm15[4,5],ymm2[6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 48(%r9), %ymm4
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4],ymm4[5],ymm2[6,7]
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm11, %xmm4 # 16-byte Folded Reload
; AVX2-FAST-ALL-NEXT: # xmm4 = xmm11[2],mem[2],xmm11[3],mem[3]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm4 = xmm4[2,3,2,3]
; AVX2-FAST-ALL-NEXT: vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm10 # 32-byte Reload
; AVX2-FAST-ALL-NEXT: vinserti128 $1, %xmm4, %ymm10, %ymm4
; AVX2-FAST-ALL-NEXT: vbroadcasti128 {{.*#+}} ymm10 = [2,2,3,3,2,2,3,3]
; AVX2-FAST-ALL-NEXT: # ymm10 = mem[0,1,0,1]
; AVX2-FAST-ALL-NEXT: vpermd %ymm15, %ymm10, %ymm11
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm4 = ymm11[0,1],ymm4[2,3,4,5],ymm11[6,7]
; AVX2-FAST-ALL-NEXT: vpermd %ymm12, %ymm10, %ymm11
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm4 = ymm4[0],ymm11[1],ymm4[2,3,4,5,6],ymm11[7]
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{.*#+}} ymm11 = ymm7[2],ymm0[2],ymm7[3],ymm0[3],ymm7[6],ymm0[6],ymm7[7],ymm0[7]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} ymm11 = ymm11[2,3,2,3,6,7,6,7]
; AVX2-FAST-ALL-NEXT: vperm2i128 {{.*#+}} ymm1 = ymm1[2,3],ymm11[2,3]
; AVX2-FAST-ALL-NEXT: vmovdqa (%r8), %ymm11
; AVX2-FAST-ALL-NEXT: vpermd %ymm11, %ymm9, %ymm9
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm9[0,1],ymm1[2,3,4,5],ymm9[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqa (%r9), %ymm9
; AVX2-FAST-ALL-NEXT: vpermd %ymm9, %ymm8, %ymm8
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0],ymm8[1],ymm1[2,3,4,5,6],ymm8[7]
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} ymm0 = ymm7[0],ymm0[0],ymm7[1],ymm0[1],ymm7[4],ymm0[4],ymm7[5],ymm0[5]
; AVX2-FAST-ALL-NEXT: vpunpckldq {{.*#+}} ymm3 = ymm5[0],ymm3[0],ymm5[1],ymm3[1],ymm5[4],ymm3[4],ymm5[5],ymm3[5]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,2]
; AVX2-FAST-ALL-NEXT: vpermq {{.*#+}} ymm3 = ymm3[2,1,2,3]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm3[0,1],ymm0[2,3],ymm3[4,5,6,7]
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm11[4,5],ymm0[6,7]
; AVX2-FAST-ALL-NEXT: vpbroadcastd 16(%r9), %ymm3
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm3[5],ymm0[6,7]
; AVX2-FAST-ALL-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; AVX2-FAST-ALL-NEXT: vpunpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 16-byte Folded Reload
; AVX2-FAST-ALL-NEXT: # xmm3 = xmm3[2],mem[2],xmm3[3],mem[3]
; AVX2-FAST-ALL-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[2,3,2,3]
; AVX2-FAST-ALL-NEXT: vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-FAST-ALL-NEXT: vinserti128 $1, %xmm3, %ymm5, %ymm3
; AVX2-FAST-ALL-NEXT: vpermd %ymm11, %ymm10, %ymm5
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm3 = ymm5[0,1],ymm3[2,3,4,5],ymm5[6,7]
; AVX2-FAST-ALL-NEXT: vpermd %ymm9, %ymm10, %ymm5
; AVX2-FAST-ALL-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0],ymm5[1],ymm3[2,3,4,5,6],ymm5[7]
; AVX2-FAST-ALL-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm3, 64(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm0, 96(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm1, 160(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm4, 256(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm2, 288(%rax)
; AVX2-FAST-ALL-NEXT: vmovdqa %ymm6, 352(%rax)
; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, 128(%rax)
; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, 192(%rax)
; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, 320(%rax)
; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, (%rax)
; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, 32(%rax)
; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, 224(%rax)
; AVX2-FAST-ALL-NEXT: addq $184, %rsp
; AVX2-FAST-ALL-NEXT: vzeroupper
; AVX2-FAST-ALL-NEXT: retq
;
; AVX2-FAST-PERLANE-LABEL: store_i32_stride6_vf16:
; AVX2-FAST-PERLANE: # %bb.0:
; AVX2-FAST-PERLANE-NEXT: subq $200, %rsp
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rsi), %xmm0
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%rsi), %xmm8
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rdi), %xmm2
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%rdi), %xmm10
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm10[2],xmm8[2],xmm10[3],xmm8[3]
; AVX2-FAST-PERLANE-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-PERLANE-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rcx), %xmm6
; AVX2-FAST-PERLANE-NEXT: vmovdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%rcx), %xmm4
; AVX2-FAST-PERLANE-NEXT: vmovdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm4 = xmm4[0,1,2,2]
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rdx), %xmm7
; AVX2-FAST-PERLANE-NEXT: vmovdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%rdx), %xmm1
; AVX2-FAST-PERLANE-NEXT: vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm5 = xmm1[1,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} xmm4 = xmm5[0],xmm4[1],xmm5[2],xmm4[3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm4 = ymm4[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm3[4,5],ymm4[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%r8), %xmm12
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%r8), %xmm11
; AVX2-FAST-PERLANE-NEXT: vpmovzxdq {{.*#+}} xmm4 = xmm11[0],zero,xmm11[1],zero
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3],ymm3[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 36(%r9), %ymm4
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm4[3],ymm3[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; AVX2-FAST-PERLANE-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-PERLANE-NEXT: vinserti128 $1, %xmm3, %ymm0, %ymm3
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm4 = xmm6[0,1,2,2]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm5 = xmm7[1,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} xmm4 = xmm5[0],xmm4[1],xmm5[2],xmm4[3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm4 = ymm4[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm3[4,5],ymm4[6,7]
; AVX2-FAST-PERLANE-NEXT: vpmovzxdq {{.*#+}} xmm4 = xmm12[0],zero,xmm12[1],zero
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3],ymm3[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 4(%r9), %ymm4
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1,2],ymm4[3],ymm3[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqu %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd (%rcx), %xmm3
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd (%rdx), %xmm4
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%rdx), %ymm5
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm4[2,3],ymm0[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastq %xmm12, %ymm2
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm2[4,5],ymm0[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%r9), %xmm15
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd %xmm15, %ymm2
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm2[5],ymm0[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%rcx), %ymm3
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm2 = ymm3[0,1,2,2,4,5,6,6]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm4 = ymm5[1,1,2,3,5,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm2 = ymm4[0],ymm2[1],ymm4[2],ymm2[3],ymm4[4],ymm2[5],ymm4[6],ymm2[7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm2 = ymm2[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%rdi), %ymm9
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%rsi), %ymm7
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} ymm13 = ymm9[2],ymm7[2],ymm9[3],ymm7[3],ymm9[6],ymm7[6],ymm9[7],ymm7[7]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm13[4,5],ymm2[6,7]
; AVX2-FAST-PERLANE-NEXT: vpmovzxdq {{.*#+}} xmm4 = mem[0],zero,mem[1],zero
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1],ymm4[2,3],ymm2[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 52(%r9), %ymm4
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm2[0,1,2],ymm4[3],ymm2[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqu %ymm0, (%rsp) # 32-byte Spill
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 32(%rcx), %xmm2
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 32(%rdx), %xmm4
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm4[0],xmm2[0],xmm4[1],xmm2[1]
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm10[0],xmm8[0],xmm10[1],xmm8[1]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1],ymm2[2,3],ymm1[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastq %xmm11, %ymm2
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%r9), %xmm6
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd %xmm6, %ymm2
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1,2,3,4],ymm2[5],ymm1[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rdx), %ymm2
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rcx), %ymm1
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm4 = ymm1[0,1,2,2,4,5,6,6]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm8 = ymm2[1,1,2,3,5,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm4 = ymm8[0],ymm4[1],ymm8[2],ymm4[3],ymm8[4],ymm4[5],ymm8[6],ymm4[7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm0 = ymm4[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rdi), %ymm10
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%rsi), %ymm8
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} ymm4 = ymm10[2],ymm8[2],ymm10[3],ymm8[3],ymm10[6],ymm8[6],ymm10[7],ymm8[7]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm4[4,5],ymm0[6,7]
; AVX2-FAST-PERLANE-NEXT: vpmovzxdq {{.*#+}} xmm14 = mem[0],zero,mem[1],zero
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm14[2,3],ymm0[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 20(%r9), %ymm14
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm14[3],ymm0[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} ymm0 = ymm5[2],ymm3[2],ymm5[3],ymm3[3],ymm5[6],ymm3[6],ymm5[7],ymm3[7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
; AVX2-FAST-PERLANE-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm13[2,3],ymm0[2,3]
; AVX2-FAST-PERLANE-NEXT: vmovdqa 32(%r8), %ymm14
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm13 = ymm14[2,1,3,3,6,5,7,7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm13 = ymm13[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm13[0,1],ymm0[2,3,4,5],ymm13[6,7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm13 = mem[0,2,2,3,4,6,6,7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm13 = ymm13[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm13 = ymm0[0],ymm13[1],ymm0[2,3,4,5,6],ymm13[7]
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} ymm0 = ymm5[0],ymm3[0],ymm5[1],ymm3[1],ymm5[4],ymm3[4],ymm5[5],ymm3[5]
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} ymm3 = ymm9[0],ymm7[0],ymm9[1],ymm7[1],ymm9[4],ymm7[4],ymm9[5],ymm7[5]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,2]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm3 = ymm3[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm3[0,1],ymm0[2,3],ymm3[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm14[4,5],ymm0[6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 48(%r9), %ymm3
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm3[5],ymm0[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 16-byte Folded Reload
; AVX2-FAST-PERLANE-NEXT: # xmm3 = xmm3[2],mem[2],xmm3[3],mem[3]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm3 = xmm3[2,3,2,3]
; AVX2-FAST-PERLANE-NEXT: vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-FAST-PERLANE-NEXT: vinserti128 $1, %xmm3, %ymm5, %ymm3
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm5 = xmm11[2,2,3,3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm3 = ymm5[0,1],ymm3[2,3,4,5],ymm5[6,7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm5 = xmm6[2,2,3,3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0],ymm5[1],ymm3[2,3,4,5,6],ymm5[7]
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{.*#+}} ymm5 = ymm2[2],ymm1[2],ymm2[3],ymm1[3],ymm2[6],ymm1[6],ymm2[7],ymm1[7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm5 = ymm5[2,3,2,3,6,7,6,7]
; AVX2-FAST-PERLANE-NEXT: vperm2i128 {{.*#+}} ymm4 = ymm4[2,3],ymm5[2,3]
; AVX2-FAST-PERLANE-NEXT: vmovdqa (%r8), %ymm5
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm6 = ymm5[2,1,3,3,6,5,7,7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm4 = ymm6[0,1],ymm4[2,3,4,5],ymm6[6,7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} ymm6 = mem[0,2,2,3,4,6,6,7]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm4 = ymm4[0],ymm6[1],ymm4[2,3,4,5,6],ymm6[7]
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} ymm1 = ymm2[0],ymm1[0],ymm2[1],ymm1[1],ymm2[4],ymm1[4],ymm2[5],ymm1[5]
; AVX2-FAST-PERLANE-NEXT: vpunpckldq {{.*#+}} ymm2 = ymm10[0],ymm8[0],ymm10[1],ymm8[1],ymm10[4],ymm8[4],ymm10[5],ymm8[5]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm1 = ymm1[2,2,2,2]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm2 = ymm2[2,1,2,3]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm2[0,1],ymm1[2,3],ymm2[4,5,6,7]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm5[4,5],ymm1[6,7]
; AVX2-FAST-PERLANE-NEXT: vpbroadcastd 16(%r9), %ymm2
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm2[5],ymm1[6,7]
; AVX2-FAST-PERLANE-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
; AVX2-FAST-PERLANE-NEXT: vpunpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 16-byte Folded Reload
; AVX2-FAST-PERLANE-NEXT: # xmm2 = xmm2[2],mem[2],xmm2[3],mem[3]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[2,3,2,3]
; AVX2-FAST-PERLANE-NEXT: vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
; AVX2-FAST-PERLANE-NEXT: vinserti128 $1, %xmm2, %ymm5, %ymm2
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm5 = xmm12[2,2,3,3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm2 = ymm5[0,1],ymm2[2,3,4,5],ymm5[6,7]
; AVX2-FAST-PERLANE-NEXT: vpshufd {{.*#+}} xmm5 = xmm15[2,2,3,3]
; AVX2-FAST-PERLANE-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,1,2,1]
; AVX2-FAST-PERLANE-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0],ymm5[1],ymm2[2,3,4,5,6],ymm5[7]
; AVX2-FAST-PERLANE-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm2, 64(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm1, 96(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm4, 160(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm3, 256(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm0, 288(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovdqa %ymm13, 352(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, 128(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, 192(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovups (%rsp), %ymm0 # 32-byte Reload
; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, 320(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, (%rax)
; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, 32(%rax)
; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, 224(%rax)
; AVX2-FAST-PERLANE-NEXT: addq $200, %rsp
; AVX2-FAST-PERLANE-NEXT: vzeroupper
; AVX2-FAST-PERLANE-NEXT: retq
;
; AVX512-LABEL: store_i32_stride6_vf16:
; AVX512: # %bb.0:
; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rax
; AVX512-NEXT: vmovdqu64 (%rdi), %zmm2
; AVX512-NEXT: vmovdqu64 (%rsi), %zmm3
; AVX512-NEXT: vmovdqu64 (%rdx), %zmm4
; AVX512-NEXT: vmovdqu64 (%rcx), %zmm5
; AVX512-NEXT: vmovdqu64 (%r8), %zmm1
; AVX512-NEXT: vmovdqu64 (%r9), %zmm0
; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm6 = [4,20,3,19,0,0,5,21,4,20,3,19,0,0,5,21]
; AVX512-NEXT: # zmm6 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vpermi2d %zmm3, %zmm2, %zmm6
; AVX512-NEXT: vmovdqa (%rdx), %ymm7
; AVX512-NEXT: vmovdqa {{.*#+}} ymm8 = [3,11,0,8,7,15,4,12]
; AVX512-NEXT: vpermi2d (%rcx), %ymm7, %ymm8
; AVX512-NEXT: movb $36, %cl
; AVX512-NEXT: kmovd %ecx, %k1
; AVX512-NEXT: vshufi64x2 {{.*#+}} zmm6 {%k1} = zmm8[0,1,0,1,2,3,0,1]
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm7 = <18,u,2,3,4,5,19,u,8,9,10,11,20,u,14,15>
; AVX512-NEXT: vpermi2d %zmm1, %zmm6, %zmm7
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm6 = [0,18,2,3,4,5,6,19,8,9,10,11,12,20,14,15]
; AVX512-NEXT: vpermi2d %zmm0, %zmm7, %zmm6
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm7 = <u,u,0,16,11,27,u,u,15,31,12,28,u,u,0,16>
; AVX512-NEXT: vpermi2d %zmm5, %zmm4, %zmm7
; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [12,28,11,27,0,0,13,29,12,28,11,27,0,0,13,29]
; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vpermi2d %zmm3, %zmm2, %zmm8
; AVX512-NEXT: vmovdqa64 %zmm7, %zmm8 {%k1}
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm7 = <26,u,2,3,4,5,27,u,8,9,10,11,28,u,14,15>
; AVX512-NEXT: vpermi2d %zmm1, %zmm8, %zmm7
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm8 = [0,26,2,3,4,5,6,27,8,9,10,11,12,28,14,15]
; AVX512-NEXT: vpermi2d %zmm0, %zmm7, %zmm8
; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm7 = [9,25,8,24,0,0,10,26,9,25,8,24,0,0,10,26]
; AVX512-NEXT: # zmm7 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vpermi2d %zmm5, %zmm4, %zmm7
; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm9 = [8,24,0,0,10,26,9,25,8,24,0,0,10,26,9,25]
; AVX512-NEXT: # zmm9 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vpermi2d %zmm3, %zmm2, %zmm9
; AVX512-NEXT: movb $-110, %cl
; AVX512-NEXT: kmovd %ecx, %k2
; AVX512-NEXT: vmovdqa64 %zmm7, %zmm9 {%k2}
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm7 = <0,1,2,3,24,u,6,7,8,9,25,u,12,13,14,15>
; AVX512-NEXT: vpermi2d %zmm1, %zmm9, %zmm7
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm9 = [0,1,2,3,4,24,6,7,8,9,10,25,12,13,14,15]
; AVX512-NEXT: vpermi2d %zmm0, %zmm7, %zmm9
; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm7 = [1,17,0,16,0,0,2,18,1,17,0,16,0,0,2,18]
; AVX512-NEXT: # zmm7 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vpermi2d %zmm5, %zmm4, %zmm7
; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm10 = [0,16,0,0,2,18,1,17,0,16,0,0,2,18,1,17]
; AVX512-NEXT: # zmm10 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vpermi2d %zmm3, %zmm2, %zmm10
; AVX512-NEXT: vmovdqa64 %zmm7, %zmm10 {%k2}
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm7 = <0,1,2,3,16,u,6,7,8,9,17,u,12,13,14,15>
; AVX512-NEXT: vpermi2d %zmm1, %zmm10, %zmm7
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm10 = [0,1,2,3,4,16,6,7,8,9,10,17,12,13,14,15]
; AVX512-NEXT: vpermi2d %zmm0, %zmm7, %zmm10
; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm7 = [5,21,0,0,7,23,6,22,5,21,0,0,7,23,6,22]
; AVX512-NEXT: # zmm7 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vpermi2d %zmm5, %zmm4, %zmm7
; AVX512-NEXT: vmovdqa (%rdi), %ymm11
; AVX512-NEXT: vpunpckhdq {{.*#+}} ymm11 = ymm11[2],mem[2],ymm11[3],mem[3],ymm11[6],mem[6],ymm11[7],mem[7]
; AVX512-NEXT: vshufi64x2 {{.*#+}} zmm7 {%k1} = zmm11[0,1,2,3,2,3,0,1]
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm11 = <0,1,21,u,4,5,6,7,22,u,10,11,12,13,23,u>
; AVX512-NEXT: vpermi2d %zmm1, %zmm7, %zmm11
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm7 = [0,1,2,21,4,5,6,7,8,22,10,11,12,13,14,23]
; AVX512-NEXT: vpermi2d %zmm0, %zmm11, %zmm7
; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm11 = [13,29,0,0,15,31,14,30,13,29,0,0,15,31,14,30]
; AVX512-NEXT: # zmm11 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vpermi2d %zmm5, %zmm4, %zmm11
; AVX512-NEXT: vpunpckhdq {{.*#+}} zmm2 = zmm2[2],zmm3[2],zmm2[3],zmm3[3],zmm2[6],zmm3[6],zmm2[7],zmm3[7],zmm2[10],zmm3[10],zmm2[11],zmm3[11],zmm2[14],zmm3[14],zmm2[15],zmm3[15]
; AVX512-NEXT: vshufi64x2 {{.*#+}} zmm11 {%k1} = zmm2[0,1,6,7,6,7,0,1]
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm2 = <0,1,29,u,4,5,6,7,30,u,10,11,12,13,31,u>
; AVX512-NEXT: vpermi2d %zmm1, %zmm11, %zmm2
; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm1 = [0,1,2,29,4,5,6,7,8,30,10,11,12,13,14,31]
; AVX512-NEXT: vpermi2d %zmm0, %zmm2, %zmm1
; AVX512-NEXT: vmovdqu64 %zmm10, (%rax)
; AVX512-NEXT: vmovdqu64 %zmm9, 192(%rax)
; AVX512-NEXT: vmovdqu64 %zmm1, 320(%rax)
; AVX512-NEXT: vmovdqu64 %zmm8, 256(%rax)
; AVX512-NEXT: vmovdqu64 %zmm7, 128(%rax)
; AVX512-NEXT: vmovdqu64 %zmm6, 64(%rax)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%in.vec0 = load <16 x i32>, ptr %in.vecptr0, align 32
%in.vec1 = load <16 x i32>, ptr %in.vecptr1, align 32
%in.vec2 = load <16 x i32>, ptr %in.vecptr2, align 32
%in.vec3 = load <16 x i32>, ptr %in.vecptr3, align 32
%in.vec4 = load <16 x i32>, ptr %in.vecptr4, align 32
%in.vec5 = load <16 x i32>, ptr %in.vecptr5, align 32
%concat01 = shufflevector <16 x i32> %in.vec0, <16 x i32> %in.vec1, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
%concat23 = shufflevector <16 x i32> %in.vec2, <16 x i32> %in.vec3, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
%concat45 = shufflevector <16 x i32> %in.vec4, <16 x i32> %in.vec5, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
%concat0123 = shufflevector <32 x i32> %concat01, <32 x i32> %concat23, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
%concat45uu = shufflevector <32 x i32> %concat45, <32 x i32> poison, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
%concat012345 = shufflevector <64 x i32> %concat0123, <64 x i32> %concat45uu, <96 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63, i32 64, i32 65, i32 66, i32 67, i32 68, i32 69, i32 70, i32 71, i32 72, i32 73, i32 74, i32 75, i32 76, i32 77, i32 78, i32 79, i32 80, i32 81, i32 82, i32 83, i32 84, i32 85, i32 86, i32 87, i32 88, i32 89, i32 90, i32 91, i32 92, i32 93, i32 94, i32 95>
%interleaved.vec = shufflevector <96 x i32> %concat012345, <96 x i32> poison, <96 x i32> <i32 0, i32 16, i32 32, i32 48, i32 64, i32 80, i32 1, i32 17, i32 33, i32 49, i32 65, i32 81, i32 2, i32 18, i32 34, i32 50, i32 66, i32 82, i32 3, i32 19, i32 35, i32 51, i32 67, i32 83, i32 4, i32 20, i32 36, i32 52, i32 68, i32 84, i32 5, i32 21, i32 37, i32 53, i32 69, i32 85, i32 6, i32 22, i32 38, i32 54, i32 70, i32 86, i32 7, i32 23, i32 39, i32 55, i32 71, i32 87, i32 8, i32 24, i32 40, i32 56, i32 72, i32 88, i32 9, i32 25, i32 41, i32 57, i32 73, i32 89, i32 10, i32 26, i32 42, i32 58, i32 74, i32 90, i32 11, i32 27, i32 43, i32 59, i32 75, i32 91, i32 12, i32 28, i32 44, i32 60, i32 76, i32 92, i32 13, i32 29, i32 45, i32 61, i32 77, i32 93, i32 14, i32 30, i32 46, i32 62, i32 78, i32 94, i32 15, i32 31, i32 47, i32 63, i32 79, i32 95>
store <96 x i32> %interleaved.vec, ptr %out.vec, align 32
ret void
}