--- /dev/null
+; 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 loads.
+
+define void @load_i32_stride6_vf2(<12 x i32>* %in.vec, <2 x i32>* %out.vec0, <2 x i32>* %out.vec1, <2 x i32>* %out.vec2, <2 x i32>* %out.vec3, <2 x i32>* %out.vec4, <2 x i32>* %out.vec5) nounwind {
+; SSE-LABEL: load_i32_stride6_vf2:
+; SSE: # %bb.0:
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: movdqa (%rdi), %xmm0
+; SSE-NEXT: movdqa 16(%rdi), %xmm1
+; SSE-NEXT: movdqa 32(%rdi), %xmm2
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm1[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm0[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm5 = xmm0[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm0[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm1[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm5 = xmm5[0],xmm2[0],xmm5[1],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm2[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm2[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm2[3,3,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm7 = xmm1[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm7 = xmm7[0],xmm2[0],xmm7[1],xmm2[1]
+; SSE-NEXT: movq %xmm0, (%rsi)
+; SSE-NEXT: movq %xmm4, (%rdx)
+; SSE-NEXT: movq %xmm5, (%rcx)
+; SSE-NEXT: movq %xmm6, (%r8)
+; SSE-NEXT: movq %xmm1, (%r9)
+; SSE-NEXT: movq %xmm7, (%rax)
+; SSE-NEXT: retq
+;
+; AVX1-LABEL: load_i32_stride6_vf2:
+; AVX1: # %bb.0:
+; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX1-NEXT: vmovaps (%rdi), %xmm0
+; AVX1-NEXT: vmovaps 16(%rdi), %xmm1
+; AVX1-NEXT: vmovaps 32(%rdi), %xmm2
+; AVX1-NEXT: vpermilps {{.*#+}} xmm3 = xmm1[2,2,3,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm3 = xmm0[0],xmm3[1],xmm0[2,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm4 = xmm0[0,1],xmm1[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm4 = xmm4[1,3,2,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm5 = xmm2[0,1],xmm0[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm5 = xmm5[2,0,2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[3,3,3,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm6 = xmm2[2,2,3,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm6 = xmm1[0],xmm6[1],xmm1[2,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm1[1,3,2,3]
+; AVX1-NEXT: vmovlps %xmm3, (%rsi)
+; AVX1-NEXT: vmovlps %xmm4, (%rdx)
+; AVX1-NEXT: vmovlps %xmm5, (%rcx)
+; AVX1-NEXT: vmovlps %xmm0, (%r8)
+; AVX1-NEXT: vmovlps %xmm6, (%r9)
+; AVX1-NEXT: vmovlps %xmm1, (%rax)
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: load_i32_stride6_vf2:
+; AVX2: # %bb.0:
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: vmovaps 32(%rdi), %ymm0
+; AVX2-NEXT: vmovaps (%rdi), %xmm1
+; AVX2-NEXT: vmovaps 16(%rdi), %xmm2
+; AVX2-NEXT: vmovaps 32(%rdi), %xmm3
+; AVX2-NEXT: vpermilps {{.*#+}} xmm4 = xmm2[2,2,3,3]
+; AVX2-NEXT: vblendps {{.*#+}} xmm4 = xmm1[0],xmm4[1],xmm1[2,3]
+; AVX2-NEXT: vblendps {{.*#+}} xmm2 = xmm1[0,1],xmm2[2,3]
+; AVX2-NEXT: vpermilps {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; AVX2-NEXT: vblendps {{.*#+}} xmm5 = xmm3[0,1],xmm1[2,3]
+; AVX2-NEXT: vpermilps {{.*#+}} xmm5 = xmm5[2,0,2,3]
+; AVX2-NEXT: vpermilps {{.*#+}} xmm1 = xmm1[3,3,3,3]
+; AVX2-NEXT: vblendps {{.*#+}} xmm1 = xmm1[0],xmm3[1],xmm1[2,3]
+; AVX2-NEXT: vmovaps {{.*#+}} xmm3 = <4,2,u,u>
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],mem[4,5],ymm0[6,7]
+; AVX2-NEXT: vpermps %ymm0, %ymm3, %ymm3
+; AVX2-NEXT: vmovaps {{.*#+}} xmm6 = <5,3,u,u>
+; AVX2-NEXT: vpermps %ymm0, %ymm6, %ymm0
+; AVX2-NEXT: vmovlps %xmm4, (%rsi)
+; AVX2-NEXT: vmovlps %xmm2, (%rdx)
+; AVX2-NEXT: vmovlps %xmm5, (%rcx)
+; AVX2-NEXT: vmovlps %xmm1, (%r8)
+; AVX2-NEXT: vmovlps %xmm3, (%r9)
+; AVX2-NEXT: vmovlps %xmm0, (%rax)
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_i32_stride6_vf2:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: vmovaps 32(%rdi), %ymm0
+; AVX512-NEXT: vmovaps (%rdi), %xmm1
+; AVX512-NEXT: vmovaps 16(%rdi), %xmm2
+; AVX512-NEXT: vmovaps 32(%rdi), %xmm3
+; AVX512-NEXT: vpermilps {{.*#+}} xmm4 = xmm2[2,2,3,3]
+; AVX512-NEXT: vblendps {{.*#+}} xmm4 = xmm1[0],xmm4[1],xmm1[2,3]
+; AVX512-NEXT: vblendps {{.*#+}} xmm2 = xmm1[0,1],xmm2[2,3]
+; AVX512-NEXT: vpermilps {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; AVX512-NEXT: vblendps {{.*#+}} xmm5 = xmm3[0,1],xmm1[2,3]
+; AVX512-NEXT: vpermilps {{.*#+}} xmm5 = xmm5[2,0,2,3]
+; AVX512-NEXT: vpermilps {{.*#+}} xmm1 = xmm1[3,3,3,3]
+; AVX512-NEXT: vblendps {{.*#+}} xmm1 = xmm1[0],xmm3[1],xmm1[2,3]
+; AVX512-NEXT: vmovaps {{.*#+}} xmm3 = <4,2,u,u>
+; AVX512-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],mem[4,5],ymm0[6,7]
+; AVX512-NEXT: vpermps %ymm0, %ymm3, %ymm3
+; AVX512-NEXT: vmovaps {{.*#+}} xmm6 = <5,3,u,u>
+; AVX512-NEXT: vpermps %ymm0, %ymm6, %ymm0
+; AVX512-NEXT: vmovlps %xmm4, (%rsi)
+; AVX512-NEXT: vmovlps %xmm2, (%rdx)
+; AVX512-NEXT: vmovlps %xmm5, (%rcx)
+; AVX512-NEXT: vmovlps %xmm1, (%r8)
+; AVX512-NEXT: vmovlps %xmm3, (%r9)
+; AVX512-NEXT: vmovlps %xmm0, (%rax)
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %wide.vec = load <12 x i32>, <12 x i32>* %in.vec, align 32
+
+ %strided.vec0 = shufflevector <12 x i32> %wide.vec, <12 x i32> poison, <2 x i32> <i32 0, i32 6>
+ %strided.vec1 = shufflevector <12 x i32> %wide.vec, <12 x i32> poison, <2 x i32> <i32 1, i32 7>
+ %strided.vec2 = shufflevector <12 x i32> %wide.vec, <12 x i32> poison, <2 x i32> <i32 2, i32 8>
+ %strided.vec3 = shufflevector <12 x i32> %wide.vec, <12 x i32> poison, <2 x i32> <i32 3, i32 9>
+ %strided.vec4 = shufflevector <12 x i32> %wide.vec, <12 x i32> poison, <2 x i32> <i32 4, i32 10>
+ %strided.vec5 = shufflevector <12 x i32> %wide.vec, <12 x i32> poison, <2 x i32> <i32 5, i32 11>
+
+ store <2 x i32> %strided.vec0, <2 x i32>* %out.vec0, align 32
+ store <2 x i32> %strided.vec1, <2 x i32>* %out.vec1, align 32
+ store <2 x i32> %strided.vec2, <2 x i32>* %out.vec2, align 32
+ store <2 x i32> %strided.vec3, <2 x i32>* %out.vec3, align 32
+ store <2 x i32> %strided.vec4, <2 x i32>* %out.vec4, align 32
+ store <2 x i32> %strided.vec5, <2 x i32>* %out.vec5, align 32
+
+ ret void
+}
+
+define void @load_i32_stride6_vf4(<24 x i32>* %in.vec, <4 x i32>* %out.vec0, <4 x i32>* %out.vec1, <4 x i32>* %out.vec2, <4 x i32>* %out.vec3, <4 x i32>* %out.vec4, <4 x i32>* %out.vec5) nounwind {
+; SSE-LABEL: load_i32_stride6_vf4:
+; SSE: # %bb.0:
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: movdqa 80(%rdi), %xmm9
+; SSE-NEXT: movdqa 64(%rdi), %xmm0
+; SSE-NEXT: movdqa (%rdi), %xmm4
+; SSE-NEXT: movdqa 16(%rdi), %xmm12
+; SSE-NEXT: movdqa 48(%rdi), %xmm3
+; SSE-NEXT: pshufd {{.*#+}} xmm8 = xmm12[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm10 = xmm4[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm11 = xmm4[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm5 = xmm4[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm8[0],xmm4[1],xmm8[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm8 = xmm0[2,2,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm3[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1]
+; SSE-NEXT: movsd {{.*#+}} xmm1 = xmm4[0],xmm1[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm12[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm10 = xmm10[0],xmm4[0],xmm10[1],xmm4[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm8 = xmm0[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm3[2,2,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm3[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1]
+; SSE-NEXT: movdqa 32(%rdi), %xmm7
+; SSE-NEXT: movsd {{.*#+}} xmm3 = xmm10[0],xmm3[1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm11 = xmm11[0],xmm7[0],xmm11[1],xmm7[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm9[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm4 = xmm11[0],xmm4[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm7[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm5 = xmm5[0],xmm2[0],xmm5[1],xmm2[1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm9[0],xmm6[1],xmm9[1]
+; SSE-NEXT: movsd {{.*#+}} xmm6 = xmm5[0],xmm6[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm7[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm5 = xmm12[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm12 = xmm12[0],xmm2[0],xmm12[1],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm8 = xmm9[2,2,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1]
+; SSE-NEXT: movsd {{.*#+}} xmm2 = xmm12[0],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm7 = xmm7[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm5 = xmm5[0],xmm7[0],xmm5[1],xmm7[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm7 = xmm9[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm5[0],xmm0[1]
+; SSE-NEXT: movapd %xmm1, (%rsi)
+; SSE-NEXT: movapd %xmm3, (%rdx)
+; SSE-NEXT: movapd %xmm4, (%rcx)
+; SSE-NEXT: movapd %xmm6, (%r8)
+; SSE-NEXT: movapd %xmm2, (%r9)
+; SSE-NEXT: movapd %xmm0, (%rax)
+; SSE-NEXT: retq
+;
+; AVX1-LABEL: load_i32_stride6_vf4:
+; AVX1: # %bb.0:
+; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX1-NEXT: vmovaps 32(%rdi), %ymm0
+; AVX1-NEXT: vmovaps (%rdi), %ymm1
+; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm3
+; AVX1-NEXT: vblendps {{.*#+}} xmm4 = xmm2[0,1],xmm3[2,3]
+; AVX1-NEXT: vshufps {{.*#+}} xmm4 = xmm4[0,2],xmm3[0,3]
+; AVX1-NEXT: vmovaps 64(%rdi), %xmm5
+; AVX1-NEXT: vinsertps {{.*#+}} xmm8 = xmm4[0,1,2],xmm5[2]
+; AVX1-NEXT: vshufps {{.*#+}} xmm2 = xmm2[1,0],xmm3[3,0]
+; AVX1-NEXT: vshufps {{.*#+}} xmm2 = xmm2[0,2],xmm3[1,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm9 = xmm2[0,1,2],xmm5[3]
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT: vshufps {{.*#+}} xmm3 = xmm0[2,0],xmm1[2,3]
+; AVX1-NEXT: vmovaps 80(%rdi), %xmm6
+; AVX1-NEXT: vinsertps {{.*#+}} xmm3 = xmm3[0,1,2],xmm6[0]
+; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[3,3]
+; AVX1-NEXT: vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm6[1]
+; AVX1-NEXT: vmovaps 32(%rdi), %xmm1
+; AVX1-NEXT: vpermilps {{.*#+}} xmm7 = xmm1[2,2,3,3]
+; AVX1-NEXT: vmovaps 16(%rdi), %xmm4
+; AVX1-NEXT: vblendps {{.*#+}} xmm7 = xmm4[0],xmm7[1],xmm4[2,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm2 = xmm5[0,1],xmm6[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm2 = xmm2[0,1,0,2]
+; AVX1-NEXT: vblendps {{.*#+}} xmm2 = xmm7[0,1],xmm2[2,3]
+; AVX1-NEXT: vpmovzxdq {{.*#+}} xmm5 = xmm5[0],zero,xmm5[1],zero
+; AVX1-NEXT: vpblendw {{.*#+}} xmm5 = xmm5[0,1,2,3,4,5],xmm6[6,7]
+; AVX1-NEXT: vblendps {{.*#+}} xmm1 = xmm4[0,1],xmm1[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm1[1,3,2,3]
+; AVX1-NEXT: vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm5[2,3]
+; AVX1-NEXT: vmovaps %xmm8, (%rsi)
+; AVX1-NEXT: vmovaps %xmm9, (%rdx)
+; AVX1-NEXT: vmovaps %xmm3, (%rcx)
+; AVX1-NEXT: vmovaps %xmm0, (%r8)
+; AVX1-NEXT: vmovaps %xmm2, (%r9)
+; AVX1-NEXT: vmovaps %xmm1, (%rax)
+; AVX1-NEXT: vzeroupper
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: load_i32_stride6_vf4:
+; AVX2: # %bb.0:
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: vmovdqa (%rdi), %ymm1
+; AVX2-NEXT: vmovdqa 32(%rdi), %ymm2
+; AVX2-NEXT: vmovdqa {{.*#+}} xmm0 = <0,6,4,u>
+; AVX2-NEXT: vpblendd {{.*#+}} ymm3 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6,7]
+; AVX2-NEXT: vpermd %ymm3, %ymm0, %ymm0
+; AVX2-NEXT: vmovdqa 64(%rdi), %xmm4
+; AVX2-NEXT: vpshufd {{.*#+}} xmm5 = xmm4[2,2,2,2]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm8 = xmm0[0,1,2],xmm5[3]
+; AVX2-NEXT: vmovdqa {{.*#+}} xmm5 = <1,7,5,u>
+; AVX2-NEXT: vpermd %ymm3, %ymm5, %ymm3
+; AVX2-NEXT: vpblendd {{.*#+}} xmm3 = xmm3[0,1,2],xmm4[3]
+; AVX2-NEXT: vmovdqa (%rdi), %xmm5
+; AVX2-NEXT: vpshufd {{.*#+}} xmm6 = xmm5[2,3,2,3]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm7 = ymm2[0,0,2,3,4,4,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0],ymm7[1,2,3],ymm6[4],ymm7[5,6,7]
+; AVX2-NEXT: vpermq {{.*#+}} ymm6 = ymm6[0,3,2,3]
+; AVX2-NEXT: vmovdqa 80(%rdi), %xmm7
+; AVX2-NEXT: vpbroadcastd %xmm7, %xmm0
+; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm6[0,1,2],xmm0[3]
+; AVX2-NEXT: vpshufd {{.*#+}} xmm5 = xmm5[3,3,3,3]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm6 = ymm2[0,1,3,3,4,5,7,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm5 = ymm5[0],ymm6[1,2,3],ymm5[4],ymm6[5,6,7]
+; AVX2-NEXT: vpermq {{.*#+}} ymm5 = ymm5[0,3,2,3]
+; AVX2-NEXT: vpbroadcastd 84(%rdi), %xmm6
+; AVX2-NEXT: vpblendd {{.*#+}} xmm5 = xmm5[0,1,2],xmm6[3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm6 = xmm4[0,1],xmm7[2,3]
+; AVX2-NEXT: vpshufd {{.*#+}} xmm6 = xmm6[0,1,0,2]
+; AVX2-NEXT: vmovdqa {{.*#+}} xmm9 = <4,2,u,u>
+; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm2[0,1,2,3],ymm1[4,5],ymm2[6,7]
+; AVX2-NEXT: vpermd %ymm1, %ymm9, %ymm2
+; AVX2-NEXT: vpblendd {{.*#+}} xmm2 = xmm2[0,1],xmm6[2,3]
+; AVX2-NEXT: vpmovzxdq {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero
+; AVX2-NEXT: vpblendd {{.*#+}} xmm4 = xmm4[0,1,2],xmm7[3]
+; AVX2-NEXT: vmovdqa {{.*#+}} xmm6 = <5,3,u,u>
+; AVX2-NEXT: vpermd %ymm1, %ymm6, %ymm1
+; AVX2-NEXT: vpblendd {{.*#+}} xmm1 = xmm1[0,1],xmm4[2,3]
+; AVX2-NEXT: vmovdqa %xmm8, (%rsi)
+; AVX2-NEXT: vmovdqa %xmm3, (%rdx)
+; AVX2-NEXT: vmovdqa %xmm0, (%rcx)
+; AVX2-NEXT: vmovdqa %xmm5, (%r8)
+; AVX2-NEXT: vmovdqa %xmm2, (%r9)
+; AVX2-NEXT: vmovdqa %xmm1, (%rax)
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: load_i32_stride6_vf4:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r10
+; AVX512-NEXT: vmovdqa (%rdi), %xmm2
+; AVX512-NEXT: vmovdqa 16(%rdi), %xmm0
+; AVX512-NEXT: vmovdqa 32(%rdi), %xmm1
+; AVX512-NEXT: vmovdqa 48(%rdi), %xmm3
+; AVX512-NEXT: vpextrd $2, %xmm0, %eax
+; AVX512-NEXT: vpinsrd $1, %eax, %xmm2, %xmm4
+; AVX512-NEXT: vmovd %xmm3, %eax
+; AVX512-NEXT: vpinsrd $2, %eax, %xmm4, %xmm4
+; AVX512-NEXT: vmovdqa 64(%rdi), %xmm5
+; AVX512-NEXT: vpextrd $2, %xmm5, %eax
+; AVX512-NEXT: vpinsrd $3, %eax, %xmm4, %xmm8
+; AVX512-NEXT: vpextrd $3, %xmm0, %eax
+; AVX512-NEXT: vpshufd {{.*#+}} xmm6 = xmm2[1,1,1,1]
+; AVX512-NEXT: vpinsrd $1, %eax, %xmm6, %xmm6
+; AVX512-NEXT: vpextrd $1, %xmm3, %eax
+; AVX512-NEXT: vpinsrd $2, %eax, %xmm6, %xmm6
+; AVX512-NEXT: vpblendd {{.*#+}} xmm6 = xmm6[0,1,2],xmm5[3]
+; AVX512-NEXT: vpbroadcastd 8(%rdi), %xmm7
+; AVX512-NEXT: vmovd %xmm1, %eax
+; AVX512-NEXT: vpinsrd $1, %eax, %xmm7, %xmm7
+; AVX512-NEXT: vpblendd {{.*#+}} xmm7 = xmm7[0,1],xmm3[2],xmm7[3]
+; AVX512-NEXT: vmovdqa 80(%rdi), %xmm4
+; AVX512-NEXT: vmovd %xmm4, %eax
+; AVX512-NEXT: vpinsrd $3, %eax, %xmm7, %xmm7
+; AVX512-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[3,3,3,3]
+; AVX512-NEXT: vpblendd {{.*#+}} xmm2 = xmm2[0],xmm1[1],xmm2[2,3]
+; AVX512-NEXT: vpextrd $3, %xmm3, %eax
+; AVX512-NEXT: vpinsrd $2, %eax, %xmm2, %xmm2
+; AVX512-NEXT: vpextrd $1, %xmm4, %eax
+; AVX512-NEXT: vpinsrd $3, %eax, %xmm2, %xmm2
+; AVX512-NEXT: vpextrd $2, %xmm1, %eax
+; AVX512-NEXT: vpinsrd $1, %eax, %xmm0, %xmm3
+; AVX512-NEXT: vmovd %xmm5, %eax
+; AVX512-NEXT: vpinsrd $2, %eax, %xmm3, %xmm3
+; AVX512-NEXT: vpextrd $2, %xmm4, %eax
+; AVX512-NEXT: vpinsrd $3, %eax, %xmm3, %xmm3
+; AVX512-NEXT: vpextrd $3, %xmm1, %eax
+; AVX512-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
+; AVX512-NEXT: vpinsrd $1, %eax, %xmm0, %xmm0
+; AVX512-NEXT: vpextrd $1, %xmm5, %eax
+; AVX512-NEXT: vpinsrd $2, %eax, %xmm0, %xmm0
+; AVX512-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1,2],xmm4[3]
+; AVX512-NEXT: vmovdqa %xmm8, (%rsi)
+; AVX512-NEXT: vmovdqa %xmm6, (%rdx)
+; AVX512-NEXT: vmovdqa %xmm7, (%rcx)
+; AVX512-NEXT: vmovdqa %xmm2, (%r8)
+; AVX512-NEXT: vmovdqa %xmm3, (%r9)
+; AVX512-NEXT: vmovdqa %xmm0, (%r10)
+; AVX512-NEXT: retq
+ %wide.vec = load <24 x i32>, <24 x i32>* %in.vec, align 32
+
+ %strided.vec0 = shufflevector <24 x i32> %wide.vec, <24 x i32> poison, <4 x i32> <i32 0, i32 6, i32 12, i32 18>
+ %strided.vec1 = shufflevector <24 x i32> %wide.vec, <24 x i32> poison, <4 x i32> <i32 1, i32 7, i32 13, i32 19>
+ %strided.vec2 = shufflevector <24 x i32> %wide.vec, <24 x i32> poison, <4 x i32> <i32 2, i32 8, i32 14, i32 20>
+ %strided.vec3 = shufflevector <24 x i32> %wide.vec, <24 x i32> poison, <4 x i32> <i32 3, i32 9, i32 15, i32 21>
+ %strided.vec4 = shufflevector <24 x i32> %wide.vec, <24 x i32> poison, <4 x i32> <i32 4, i32 10, i32 16, i32 22>
+ %strided.vec5 = shufflevector <24 x i32> %wide.vec, <24 x i32> poison, <4 x i32> <i32 5, i32 11, i32 17, i32 23>
+
+ store <4 x i32> %strided.vec0, <4 x i32>* %out.vec0, align 32
+ store <4 x i32> %strided.vec1, <4 x i32>* %out.vec1, align 32
+ store <4 x i32> %strided.vec2, <4 x i32>* %out.vec2, align 32
+ store <4 x i32> %strided.vec3, <4 x i32>* %out.vec3, align 32
+ store <4 x i32> %strided.vec4, <4 x i32>* %out.vec4, align 32
+ store <4 x i32> %strided.vec5, <4 x i32>* %out.vec5, align 32
+
+ ret void
+}
+
+define void @load_i32_stride6_vf8(<48 x i32>* %in.vec, <8 x i32>* %out.vec0, <8 x i32>* %out.vec1, <8 x i32>* %out.vec2, <8 x i32>* %out.vec3, <8 x i32>* %out.vec4, <8 x i32>* %out.vec5) nounwind {
+; SSE-LABEL: load_i32_stride6_vf8:
+; SSE: # %bb.0:
+; SSE-NEXT: movdqa 144(%rdi), %xmm10
+; SSE-NEXT: movdqa 160(%rdi), %xmm14
+; SSE-NEXT: movdqa 96(%rdi), %xmm11
+; SSE-NEXT: movdqa 112(%rdi), %xmm5
+; SSE-NEXT: movdqa 64(%rdi), %xmm7
+; SSE-NEXT: movdqa (%rdi), %xmm4
+; SSE-NEXT: movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa 16(%rdi), %xmm2
+; SSE-NEXT: movdqa 48(%rdi), %xmm12
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm2[2,3,2,3]
+; SSE-NEXT: movdqa %xmm2, %xmm8
+; SSE-NEXT: movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm4[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm4[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
+; SSE-NEXT: movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm7[2,2,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm12[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm1[0],xmm6[1],xmm1[1]
+; SSE-NEXT: movsd {{.*#+}} xmm6 = xmm4[0],xmm6[1]
+; SSE-NEXT: movapd %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa %xmm5, %xmm4
+; SSE-NEXT: pshufd {{.*#+}} xmm13 = xmm5[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm5 = xmm11[1,1,1,1]
+; SSE-NEXT: movdqa %xmm11, %xmm1
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm13[0],xmm1[1],xmm13[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm14[2,2,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm15 = xmm10[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm15 = xmm15[0],xmm6[0],xmm15[1],xmm6[1]
+; SSE-NEXT: movsd {{.*#+}} xmm15 = xmm1[0],xmm15[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm8[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm7[2,3,2,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm12[2,2,3,3]
+; SSE-NEXT: movdqa %xmm12, %xmm13
+; SSE-NEXT: punpckldq {{.*#+}} xmm13 = xmm13[0],xmm1[0],xmm13[1],xmm1[1]
+; SSE-NEXT: movsd {{.*#+}} xmm13 = xmm2[0],xmm13[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm4[3,3,3,3]
+; SSE-NEXT: movdqa %xmm4, %xmm7
+; SSE-NEXT: movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: punpckldq {{.*#+}} xmm5 = xmm5[0],xmm1[0],xmm5[1],xmm1[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm14[2,3,2,3]
+; SSE-NEXT: movdqa %xmm10, %xmm0
+; SSE-NEXT: movdqa %xmm10, %xmm9
+; SSE-NEXT: punpckldq {{.*#+}} xmm9 = xmm9[0],xmm1[0],xmm9[1],xmm1[1]
+; SSE-NEXT: movdqa 80(%rdi), %xmm10
+; SSE-NEXT: movsd {{.*#+}} xmm9 = xmm5[0],xmm9[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm10[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm1[0],xmm6[1],xmm1[1]
+; SSE-NEXT: movdqa 32(%rdi), %xmm4
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE-NEXT: movsd {{.*#+}} xmm6 = xmm3[0],xmm6[1]
+; SSE-NEXT: movapd %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm5 = xmm0[2,2,3,3]
+; SSE-NEXT: movdqa %xmm0, %xmm1
+; SSE-NEXT: movdqa 176(%rdi), %xmm8
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm8[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm5 = xmm5[0],xmm2[0],xmm5[1],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm11[2,3,2,3]
+; SSE-NEXT: movdqa 128(%rdi), %xmm3
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE-NEXT: movsd {{.*#+}} xmm5 = xmm0[0],xmm5[1]
+; SSE-NEXT: pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; SSE-NEXT: # xmm0 = mem[3,3,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm4[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm12 = xmm12[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm12 = xmm12[0],xmm10[0],xmm12[1],xmm10[1]
+; SSE-NEXT: movsd {{.*#+}} xmm12 = xmm0[0],xmm12[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm11[3,3,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm3[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm11 = xmm1[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm11 = xmm11[0],xmm8[0],xmm11[1],xmm8[1]
+; SSE-NEXT: movsd {{.*#+}} xmm11 = xmm0[0],xmm11[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm4[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
+; SSE-NEXT: movdqa %xmm2, %xmm6
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm10[2,2,3,3]
+; SSE-NEXT: pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; SSE-NEXT: # xmm1 = mem[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; SSE-NEXT: movsd {{.*#+}} xmm1 = xmm6[0],xmm1[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm3[2,3,2,3]
+; SSE-NEXT: movdqa %xmm7, %xmm6
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm7 = xmm8[2,2,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm14[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm6[0],xmm0[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm2[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm4[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm4[0],xmm6[1],xmm4[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm10[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
+; SSE-NEXT: punpckldq {{.*#+}} xmm7 = xmm7[0],xmm4[0],xmm7[1],xmm4[1]
+; SSE-NEXT: movsd {{.*#+}} xmm7 = xmm6[0],xmm7[1]
+; SSE-NEXT: pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
+; SSE-NEXT: # xmm4 = mem[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm3[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm8[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm14 = xmm14[0],xmm2[0],xmm14[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm14 = xmm4[0],xmm14[1]
+; SSE-NEXT: movapd %xmm15, 16(%rsi)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
+; SSE-NEXT: movaps %xmm2, (%rsi)
+; SSE-NEXT: movapd %xmm9, 16(%rdx)
+; SSE-NEXT: movapd %xmm13, (%rdx)
+; SSE-NEXT: movapd %xmm5, 16(%rcx)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
+; SSE-NEXT: movaps %xmm2, (%rcx)
+; SSE-NEXT: movapd %xmm11, 16(%r8)
+; SSE-NEXT: movapd %xmm12, (%r8)
+; SSE-NEXT: movapd %xmm0, 16(%r9)
+; SSE-NEXT: movapd %xmm1, (%r9)
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: movapd %xmm14, 16(%rax)
+; SSE-NEXT: movapd %xmm7, (%rax)
+; SSE-NEXT: retq
+;
+; AVX1-LABEL: load_i32_stride6_vf8:
+; AVX1: # %bb.0:
+; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX1-NEXT: vmovaps 96(%rdi), %ymm0
+; AVX1-NEXT: vmovaps 64(%rdi), %ymm1
+; AVX1-NEXT: vmovaps 32(%rdi), %ymm9
+; AVX1-NEXT: vmovaps (%rdi), %ymm11
+; AVX1-NEXT: vmovaps 128(%rdi), %ymm6
+; AVX1-NEXT: vmovaps 160(%rdi), %ymm7
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm7[0,1,2,3],ymm6[4,5],ymm7[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm2
+; AVX1-NEXT: vblendps {{.*#+}} xmm4 = xmm2[0,1],xmm3[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm4 = xmm4[0,1,0,2]
+; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm0, %ymm8
+; AVX1-NEXT: vblendps {{.*#+}} ymm10 = ymm11[0,1,2,3],ymm9[4,5],ymm11[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm10, %xmm5
+; AVX1-NEXT: vblendps {{.*#+}} xmm4 = xmm10[0,1],xmm5[2,3]
+; AVX1-NEXT: vshufps {{.*#+}} xmm4 = xmm4[0,2],xmm5[0,3]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm12 = ymm1[0,1],ymm0[0,1]
+; AVX1-NEXT: vshufps {{.*#+}} ymm13 = ymm0[2,0],ymm12[0,0],ymm0[6,4],ymm12[4,4]
+; AVX1-NEXT: vshufps {{.*#+}} ymm13 = ymm13[2,0],ymm1[2,2],ymm13[6,4],ymm1[6,6]
+; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2],ymm13[3,4,5],ymm4[6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm8 = ymm4[0,1,2,3,4,5],ymm8[6,7]
+; AVX1-NEXT: vinsertps {{.*#+}} xmm2 = zero,zero,xmm2[1],xmm3[3]
+; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm2
+; AVX1-NEXT: vshufps {{.*#+}} xmm3 = xmm10[1,0],xmm5[3,0]
+; AVX1-NEXT: vshufps {{.*#+}} xmm3 = xmm3[0,2],xmm5[1,3]
+; AVX1-NEXT: vshufps {{.*#+}} ymm4 = ymm0[3,0],ymm12[1,0],ymm0[7,4],ymm12[5,4]
+; AVX1-NEXT: vshufps {{.*#+}} ymm4 = ymm4[2,0],ymm1[2,3],ymm4[6,4],ymm1[6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1,2],ymm4[3,4,5],ymm3[6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm12 = ymm3[0,1,2,3,4,5],ymm2[6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm6[0,1,2,3],ymm7[4,5],ymm6[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm5
+; AVX1-NEXT: vshufps {{.*#+}} xmm4 = xmm2[0,0],xmm5[2,0]
+; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm0, %ymm4
+; AVX1-NEXT: vblendps {{.*#+}} ymm6 = ymm9[0,1],ymm11[2,3],ymm9[4,5,6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm6, %xmm7
+; AVX1-NEXT: vshufps {{.*#+}} xmm9 = xmm6[2,0],xmm7[2,3]
+; AVX1-NEXT: vshufps {{.*#+}} ymm10 = ymm0[2,1],ymm1[2,0],ymm0[6,5],ymm1[6,4]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm10 = ymm10[2,3,0,1]
+; AVX1-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1,2],ymm10[3,4],ymm9[5,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1,2,3,4],ymm4[5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} xmm2 = xmm2[0,1],xmm5[3,1]
+; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm2
+; AVX1-NEXT: vshufps {{.*#+}} xmm5 = xmm6[3,1],xmm7[3,3]
+; AVX1-NEXT: vshufps {{.*#+}} ymm6 = ymm0[3,1],ymm1[2,1],ymm0[7,5],ymm1[6,5]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm6 = ymm6[2,3,0,1]
+; AVX1-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0,1,2],ymm6[3,4],ymm5[5,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm10 = ymm5[0,1,2,3,4],ymm2[5,6,7]
+; AVX1-NEXT: vmovdqa 160(%rdi), %xmm5
+; AVX1-NEXT: vpblendw {{.*#+}} xmm5 = xmm5[0,1,2,3],mem[4,5,6,7]
+; AVX1-NEXT: vmovdqa 176(%rdi), %xmm6
+; AVX1-NEXT: vpalignr {{.*#+}} xmm7 = xmm6[8,9,10,11,12,13,14,15],xmm5[0,1,2,3,4,5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} xmm7 = xmm5[0,2],xmm7[2,0]
+; AVX1-NEXT: vinsertf128 $1, %xmm7, %ymm0, %ymm7
+; AVX1-NEXT: vmovaps 32(%rdi), %xmm3
+; AVX1-NEXT: vpermilps {{.*#+}} xmm4 = xmm3[2,2,3,3]
+; AVX1-NEXT: vmovaps 16(%rdi), %xmm2
+; AVX1-NEXT: vblendps {{.*#+}} xmm4 = xmm2[0],xmm4[1],xmm2[2,3]
+; AVX1-NEXT: vmovapd 80(%rdi), %xmm11
+; AVX1-NEXT: vshufpd {{.*#+}} ymm13 = ymm11[1],ymm1[0],ymm11[2],ymm1[2]
+; AVX1-NEXT: vshufps {{.*#+}} ymm13 = ymm0[0,1],ymm13[2,0],ymm0[4,5],ymm13[6,4]
+; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1],ymm13[2,3,4,5,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3,4],ymm7[5,6,7]
+; AVX1-NEXT: vpalignr {{.*#+}} xmm6 = xmm6[12,13,14,15],xmm5[0,1,2,3,4,5,6,7,8,9,10,11]
+; AVX1-NEXT: vshufps {{.*#+}} xmm5 = xmm5[0,3],xmm6[2,0]
+; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm0, %ymm5
+; AVX1-NEXT: vblendps {{.*#+}} xmm2 = xmm2[0,1],xmm3[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm2 = xmm2[1,3,2,3]
+; AVX1-NEXT: vshufps {{.*#+}} ymm1 = ymm11[3,1],ymm1[1,3],ymm11[7,5],ymm1[5,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm0 = ymm0[1,1],ymm1[2,0],ymm0[5,5],ymm1[6,4]
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm2[0,1],ymm0[2,3,4,5,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm5[5,6,7]
+; AVX1-NEXT: vmovaps %ymm8, (%rsi)
+; AVX1-NEXT: vmovaps %ymm12, (%rdx)
+; AVX1-NEXT: vmovaps %ymm9, (%rcx)
+; AVX1-NEXT: vmovaps %ymm10, (%r8)
+; AVX1-NEXT: vmovaps %ymm4, (%r9)
+; AVX1-NEXT: vmovaps %ymm0, (%rax)
+; AVX1-NEXT: vzeroupper
+; AVX1-NEXT: retq
+;
+; AVX2-SLOW-LABEL: load_i32_stride6_vf8:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-SLOW-NEXT: vmovaps 128(%rdi), %ymm0
+; AVX2-SLOW-NEXT: vmovaps 160(%rdi), %ymm1
+; AVX2-SLOW-NEXT: vmovaps 96(%rdi), %ymm5
+; AVX2-SLOW-NEXT: vmovaps (%rdi), %ymm3
+; AVX2-SLOW-NEXT: vmovaps 32(%rdi), %ymm4
+; AVX2-SLOW-NEXT: vmovaps 64(%rdi), %ymm6
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm2 = <0,6,4,u>
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm7 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
+; AVX2-SLOW-NEXT: vpermps %ymm7, %ymm2, %ymm2
+; AVX2-SLOW-NEXT: vperm2f128 {{.*#+}} ymm8 = ymm6[0,1],ymm5[0,1]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm9 = ymm8[0,1,2,2,4,5,6,6]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm10 = ymm5[2,2,2,2,6,6,6,6]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm10[1],ymm9[2,3,4],ymm10[5],ymm9[6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2],ymm9[3,4,5],ymm2[6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm9 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
+; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm10 = [8589934596,8589934596,8589934596,8589934596]
+; AVX2-SLOW-NEXT: vpermps %ymm9, %ymm10, %ymm10
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5],ymm10[6,7]
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm10 = <1,7,5,u>
+; AVX2-SLOW-NEXT: vpermps %ymm7, %ymm10, %ymm7
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[1,1,2,3,5,5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm10 = ymm5[2,3,2,3,6,7,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0],ymm10[1],ymm8[2,3,4],ymm10[5],ymm8[6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2],ymm8[3,4,5],ymm7[6,7]
+; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm8 = [12884901893,12884901893,12884901893,12884901893]
+; AVX2-SLOW-NEXT: vpermps %ymm9, %ymm8, %ymm8
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4,5],ymm8[6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm8 = ymm6[0,1],ymm5[2,3],ymm6[4,5],ymm5[6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[2,0,2,3,6,4,6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm8 = ymm8[0,2,0,3]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm9 = ymm3[2,3,2,3,6,7,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm10 = ymm4[0,0,2,3,4,4,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm10[1,2,3],ymm9[4],ymm10[5,6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,3,2,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm8 = ymm9[0,1,2],ymm8[3,4],ymm9[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm9 = ymm1[0,0,0,0,4,4,4,4]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm10 = ymm0[0,0,2,3,4,4,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm9 = ymm10[0,1,2],ymm9[3],ymm10[4,5,6],ymm9[7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,1,0,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1,2,3,4],ymm9[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm9 = ymm5[3,3,3,3,7,7,7,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm6[1],ymm9[2,3,4],ymm6[5],ymm9[6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,2,0,3]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm10 = ymm3[3,3,3,3,7,7,7,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm11 = ymm4[0,1,3,3,4,5,7,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm10 = ymm10[0],ymm11[1,2,3],ymm10[4],ymm11[5,6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm10 = ymm10[0,3,2,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm9 = ymm10[0,1,2],ymm9[3,4],ymm10[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm10 = ymm1[0,1,0,1,4,5,4,5]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm11 = ymm0[0,1,3,3,4,5,7,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm10 = ymm11[0,1,2],ymm10[3],ymm11[4,5,6],ymm10[7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm10 = ymm10[0,1,0,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1,2,3,4],ymm10[5,6,7]
+; AVX2-SLOW-NEXT: vmovaps 80(%rdi), %xmm10
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm5 = ymm6[0,1,2,3],ymm5[4,5,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm6 = ymm5[0,1],ymm10[2,3],ymm5[4,5],ymm10[6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm6 = ymm6[0,1,0,2,4,5,4,6]
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm11 = <4,2,u,u>
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm3[4,5],ymm4[6,7]
+; AVX2-SLOW-NEXT: vpermps %ymm3, %ymm11, %ymm4
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1],ymm6[2,3,4,5,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5,6,7]
+; AVX2-SLOW-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [0,2,0,6,0,2,0,6]
+; AVX2-SLOW-NEXT: # ymm1 = mem[0,1,0,1]
+; AVX2-SLOW-NEXT: vpermps %ymm0, %ymm1, %ymm1
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm4[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm4 = ymm5[1,1,1,1,5,5,5,5]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2],ymm10[3],ymm4[4,5,6],ymm10[7]
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm5 = <5,3,u,u>
+; AVX2-SLOW-NEXT: vpermps %ymm3, %ymm5, %ymm3
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3,4,5,6,7]
+; AVX2-SLOW-NEXT: vbroadcastf128 {{.*#+}} ymm4 = [0,3,1,7,0,3,1,7]
+; AVX2-SLOW-NEXT: # ymm4 = mem[0,1,0,1]
+; AVX2-SLOW-NEXT: vpermps %ymm0, %ymm4, %ymm0
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4],ymm0[5,6,7]
+; AVX2-SLOW-NEXT: vmovaps %ymm2, (%rsi)
+; AVX2-SLOW-NEXT: vmovaps %ymm7, (%rdx)
+; AVX2-SLOW-NEXT: vmovaps %ymm8, (%rcx)
+; AVX2-SLOW-NEXT: vmovaps %ymm9, (%r8)
+; AVX2-SLOW-NEXT: vmovaps %ymm1, (%r9)
+; AVX2-SLOW-NEXT: vmovaps %ymm0, (%rax)
+; AVX2-SLOW-NEXT: vzeroupper
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-ALL-LABEL: load_i32_stride6_vf8:
+; AVX2-FAST-ALL: # %bb.0:
+; AVX2-FAST-ALL-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-FAST-ALL-NEXT: vmovaps 128(%rdi), %ymm0
+; AVX2-FAST-ALL-NEXT: vmovaps 160(%rdi), %ymm1
+; AVX2-FAST-ALL-NEXT: vmovaps 96(%rdi), %ymm5
+; AVX2-FAST-ALL-NEXT: vmovaps (%rdi), %ymm3
+; AVX2-FAST-ALL-NEXT: vmovaps 32(%rdi), %ymm4
+; AVX2-FAST-ALL-NEXT: vmovaps 64(%rdi), %ymm6
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm2 = <0,6,4,u>
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm7 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm7, %ymm2, %ymm2
+; AVX2-FAST-ALL-NEXT: vperm2f128 {{.*#+}} ymm8 = ymm6[0,1],ymm5[0,1]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm9 = ymm8[0,1,2,2,4,5,6,6]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm10 = ymm5[2,2,2,2,6,6,6,6]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm10[1],ymm9[2,3,4],ymm10[5],ymm9[6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2],ymm9[3,4,5],ymm2[6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm9 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
+; AVX2-FAST-ALL-NEXT: vbroadcastsd {{.*#+}} ymm10 = [8589934596,8589934596,8589934596,8589934596]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm9, %ymm10, %ymm10
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5],ymm10[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm10 = <1,7,5,u>
+; AVX2-FAST-ALL-NEXT: vpermps %ymm7, %ymm10, %ymm7
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[1,1,2,3,5,5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm10 = ymm5[2,3,2,3,6,7,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0],ymm10[1],ymm8[2,3,4],ymm10[5],ymm8[6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2],ymm8[3,4,5],ymm7[6,7]
+; AVX2-FAST-ALL-NEXT: vbroadcastsd {{.*#+}} ymm8 = [12884901893,12884901893,12884901893,12884901893]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm9, %ymm8, %ymm8
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4,5],ymm8[6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm8 = ymm6[0,1],ymm5[2,3],ymm6[4,5],ymm5[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} ymm9 = <u,u,u,4,2,u,u,u>
+; AVX2-FAST-ALL-NEXT: vpermps %ymm8, %ymm9, %ymm8
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm9 = ymm3[2,3,2,3,6,7,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm10 = ymm4[0,0,2,3,4,4,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm10[1,2,3],ymm9[4],ymm10[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,3,2,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm8 = ymm9[0,1,2],ymm8[3,4],ymm9[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm9 = ymm1[0,0,0,0,4,4,4,4]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm10 = ymm0[0,0,2,3,4,4,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm9 = ymm10[0,1,2],ymm9[3],ymm10[4,5,6],ymm9[7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,1,0,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1,2,3,4],ymm9[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm9 = ymm5[3,3,3,3,7,7,7,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm6[1],ymm9[2,3,4],ymm6[5],ymm9[6,7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,2,0,3]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm10 = ymm3[3,3,3,3,7,7,7,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm11 = ymm4[0,1,3,3,4,5,7,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm10 = ymm10[0],ymm11[1,2,3],ymm10[4],ymm11[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm10 = ymm10[0,3,2,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm9 = ymm10[0,1,2],ymm9[3,4],ymm10[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm10 = ymm1[0,1,0,1,4,5,4,5]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm11 = ymm0[0,1,3,3,4,5,7,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm10 = ymm11[0,1,2],ymm10[3],ymm11[4,5,6],ymm10[7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm10 = ymm10[0,1,0,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1,2,3,4],ymm10[5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps 80(%rdi), %xmm10
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm5 = ymm6[0,1,2,3],ymm5[4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm6 = ymm5[0,1],ymm10[2,3],ymm5[4,5],ymm10[6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm6 = ymm6[0,1,0,2,4,5,4,6]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm11 = <4,2,u,u>
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm3[4,5],ymm4[6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm3, %ymm11, %ymm4
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1],ymm6[2,3,4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [0,2,0,6,0,2,0,6]
+; AVX2-FAST-ALL-NEXT: # ymm1 = mem[0,1,0,1]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm0, %ymm1, %ymm1
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm4[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm4 = ymm5[1,1,1,1,5,5,5,5]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2],ymm10[3],ymm4[4,5,6],ymm10[7]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm5 = <5,3,u,u>
+; AVX2-FAST-ALL-NEXT: vpermps %ymm3, %ymm5, %ymm3
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3,4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vbroadcastf128 {{.*#+}} ymm4 = [0,3,1,7,0,3,1,7]
+; AVX2-FAST-ALL-NEXT: # ymm4 = mem[0,1,0,1]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm0, %ymm4, %ymm0
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4],ymm0[5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm2, (%rsi)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm7, (%rdx)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm8, (%rcx)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm9, (%r8)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm1, (%r9)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, (%rax)
+; AVX2-FAST-ALL-NEXT: vzeroupper
+; AVX2-FAST-ALL-NEXT: retq
+;
+; AVX2-FAST-PERLANE-LABEL: load_i32_stride6_vf8:
+; AVX2-FAST-PERLANE: # %bb.0:
+; AVX2-FAST-PERLANE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-FAST-PERLANE-NEXT: vmovaps 128(%rdi), %ymm0
+; AVX2-FAST-PERLANE-NEXT: vmovaps 160(%rdi), %ymm1
+; AVX2-FAST-PERLANE-NEXT: vmovaps 96(%rdi), %ymm5
+; AVX2-FAST-PERLANE-NEXT: vmovaps (%rdi), %ymm3
+; AVX2-FAST-PERLANE-NEXT: vmovaps 32(%rdi), %ymm4
+; AVX2-FAST-PERLANE-NEXT: vmovaps 64(%rdi), %ymm6
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm2 = <0,6,4,u>
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm7 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm7, %ymm2, %ymm2
+; AVX2-FAST-PERLANE-NEXT: vperm2f128 {{.*#+}} ymm8 = ymm6[0,1],ymm5[0,1]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm9 = ymm8[0,1,2,2,4,5,6,6]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm10 = ymm5[2,2,2,2,6,6,6,6]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm10[1],ymm9[2,3,4],ymm10[5],ymm9[6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2],ymm9[3,4,5],ymm2[6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm9 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
+; AVX2-FAST-PERLANE-NEXT: vbroadcastsd {{.*#+}} ymm10 = [8589934596,8589934596,8589934596,8589934596]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm9, %ymm10, %ymm10
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5],ymm10[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm10 = <1,7,5,u>
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm7, %ymm10, %ymm7
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[1,1,2,3,5,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm10 = ymm5[2,3,2,3,6,7,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0],ymm10[1],ymm8[2,3,4],ymm10[5],ymm8[6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2],ymm8[3,4,5],ymm7[6,7]
+; AVX2-FAST-PERLANE-NEXT: vbroadcastsd {{.*#+}} ymm8 = [12884901893,12884901893,12884901893,12884901893]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm9, %ymm8, %ymm8
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4,5],ymm8[6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm8 = ymm6[0,1],ymm5[2,3],ymm6[4,5],ymm5[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[2,0,2,3,6,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm8 = ymm8[0,2,0,3]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm9 = ymm3[2,3,2,3,6,7,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm10 = ymm4[0,0,2,3,4,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm10[1,2,3],ymm9[4],ymm10[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,3,2,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm8 = ymm9[0,1,2],ymm8[3,4],ymm9[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm9 = ymm1[0,0,0,0,4,4,4,4]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm10 = ymm0[0,0,2,3,4,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm9 = ymm10[0,1,2],ymm9[3],ymm10[4,5,6],ymm9[7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,1,0,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1,2,3,4],ymm9[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm9 = ymm5[3,3,3,3,7,7,7,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0],ymm6[1],ymm9[2,3,4],ymm6[5],ymm9[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm9 = ymm9[0,2,0,3]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm10 = ymm3[3,3,3,3,7,7,7,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm11 = ymm4[0,1,3,3,4,5,7,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm10 = ymm10[0],ymm11[1,2,3],ymm10[4],ymm11[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm10 = ymm10[0,3,2,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm9 = ymm10[0,1,2],ymm9[3,4],ymm10[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm10 = ymm1[0,1,0,1,4,5,4,5]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm11 = ymm0[0,1,3,3,4,5,7,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm10 = ymm11[0,1,2],ymm10[3],ymm11[4,5,6],ymm10[7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm10 = ymm10[0,1,0,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm9 = ymm9[0,1,2,3,4],ymm10[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps 80(%rdi), %xmm10
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm5 = ymm6[0,1,2,3],ymm5[4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm6 = ymm5[0,1],ymm10[2,3],ymm5[4,5],ymm10[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm6 = ymm6[0,1,0,2,4,5,4,6]
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm11 = <4,2,u,u>
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm3 = ymm4[0,1,2,3],ymm3[4,5],ymm4[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm3, %ymm11, %ymm4
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1],ymm6[2,3,4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [0,2,0,6,0,2,0,6]
+; AVX2-FAST-PERLANE-NEXT: # ymm1 = mem[0,1,0,1]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm0, %ymm1, %ymm1
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm4[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm4 = ymm5[1,1,1,1,5,5,5,5]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2],ymm10[3],ymm4[4,5,6],ymm10[7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm5 = <5,3,u,u>
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm3, %ymm5, %ymm3
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3,4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vbroadcastf128 {{.*#+}} ymm4 = [0,3,1,7,0,3,1,7]
+; AVX2-FAST-PERLANE-NEXT: # ymm4 = mem[0,1,0,1]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm0, %ymm4, %ymm0
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4],ymm0[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm2, (%rsi)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm7, (%rdx)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm8, (%rcx)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm9, (%r8)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm1, (%r9)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, (%rax)
+; AVX2-FAST-PERLANE-NEXT: vzeroupper
+; AVX2-FAST-PERLANE-NEXT: retq
+;
+; AVX512-LABEL: load_i32_stride6_vf8:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: vmovdqu64 (%rdi), %zmm2
+; AVX512-NEXT: vmovdqu64 64(%rdi), %zmm3
+; AVX512-NEXT: vmovdqa 128(%rdi), %ymm0
+; AVX512-NEXT: vmovdqa 160(%rdi), %ymm1
+; AVX512-NEXT: vpblendd {{.*#+}} ymm4 = ymm1[0,1,2,3],ymm0[4,5],ymm1[6,7]
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm5 = <0,6,12,18,24,30,u,u>
+; AVX512-NEXT: vpermi2d %zmm3, %zmm2, %zmm5
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm6 = [0,1,2,3,4,5,12,10]
+; AVX512-NEXT: vpermi2d %ymm4, %ymm5, %ymm6
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm5 = <1,7,13,19,25,31,u,u>
+; AVX512-NEXT: vpermi2d %zmm3, %zmm2, %zmm5
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm7 = [0,1,2,3,4,5,13,11]
+; AVX512-NEXT: vpermi2d %ymm4, %ymm5, %ymm7
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm4 = <u,u,u,u,u,0,6,12>
+; AVX512-NEXT: vpermi2d %ymm1, %ymm0, %ymm4
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm5 = <2,8,14,20,26,u,u,u>
+; AVX512-NEXT: vpermi2d %zmm3, %zmm2, %zmm5
+; AVX512-NEXT: vpblendd {{.*#+}} ymm4 = ymm5[0,1,2,3,4],ymm4[5,6,7]
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm5 = <u,u,u,u,u,1,7,13>
+; AVX512-NEXT: vpermi2d %ymm1, %ymm0, %ymm5
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm8 = <3,9,15,21,27,u,u,u>
+; AVX512-NEXT: vpermi2d %zmm3, %zmm2, %zmm8
+; AVX512-NEXT: vpblendd {{.*#+}} ymm5 = ymm8[0,1,2,3,4],ymm5[5,6,7]
+; AVX512-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5,6,7]
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm1 = <20,26,0,6,12,u,u,u>
+; AVX512-NEXT: vpermi2d %zmm2, %zmm3, %zmm1
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm8 = [0,1,2,3,4,10,8,14]
+; AVX512-NEXT: vpermi2d %ymm0, %ymm1, %ymm8
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm1 = <21,27,1,7,13,u,u,u>
+; AVX512-NEXT: vpermi2d %zmm2, %zmm3, %zmm1
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm2 = [0,1,2,3,4,11,9,15]
+; AVX512-NEXT: vpermi2d %ymm0, %ymm1, %ymm2
+; AVX512-NEXT: vmovdqa %ymm6, (%rsi)
+; AVX512-NEXT: vmovdqa %ymm7, (%rdx)
+; AVX512-NEXT: vmovdqa %ymm4, (%rcx)
+; AVX512-NEXT: vmovdqa %ymm5, (%r8)
+; AVX512-NEXT: vmovdqa %ymm8, (%r9)
+; AVX512-NEXT: vmovdqa %ymm2, (%rax)
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %wide.vec = load <48 x i32>, <48 x i32>* %in.vec, align 32
+
+ %strided.vec0 = shufflevector <48 x i32> %wide.vec, <48 x i32> poison, <8 x i32> <i32 0, i32 6, i32 12, i32 18, i32 24, i32 30, i32 36, i32 42>
+ %strided.vec1 = shufflevector <48 x i32> %wide.vec, <48 x i32> poison, <8 x i32> <i32 1, i32 7, i32 13, i32 19, i32 25, i32 31, i32 37, i32 43>
+ %strided.vec2 = shufflevector <48 x i32> %wide.vec, <48 x i32> poison, <8 x i32> <i32 2, i32 8, i32 14, i32 20, i32 26, i32 32, i32 38, i32 44>
+ %strided.vec3 = shufflevector <48 x i32> %wide.vec, <48 x i32> poison, <8 x i32> <i32 3, i32 9, i32 15, i32 21, i32 27, i32 33, i32 39, i32 45>
+ %strided.vec4 = shufflevector <48 x i32> %wide.vec, <48 x i32> poison, <8 x i32> <i32 4, i32 10, i32 16, i32 22, i32 28, i32 34, i32 40, i32 46>
+ %strided.vec5 = shufflevector <48 x i32> %wide.vec, <48 x i32> poison, <8 x i32> <i32 5, i32 11, i32 17, i32 23, i32 29, i32 35, i32 41, i32 47>
+
+ store <8 x i32> %strided.vec0, <8 x i32>* %out.vec0, align 32
+ store <8 x i32> %strided.vec1, <8 x i32>* %out.vec1, align 32
+ store <8 x i32> %strided.vec2, <8 x i32>* %out.vec2, align 32
+ store <8 x i32> %strided.vec3, <8 x i32>* %out.vec3, align 32
+ store <8 x i32> %strided.vec4, <8 x i32>* %out.vec4, align 32
+ store <8 x i32> %strided.vec5, <8 x i32>* %out.vec5, align 32
+
+ ret void
+}
+
+define void @load_i32_stride6_vf16(<96 x i32>* %in.vec, <16 x i32>* %out.vec0, <16 x i32>* %out.vec1, <16 x i32>* %out.vec2, <16 x i32>* %out.vec3, <16 x i32>* %out.vec4, <16 x i32>* %out.vec5) nounwind {
+; SSE-LABEL: load_i32_stride6_vf16:
+; SSE: # %bb.0:
+; SSE-NEXT: subq $376, %rsp # imm = 0x178
+; SSE-NEXT: movdqa 144(%rdi), %xmm9
+; SSE-NEXT: movdqa %xmm9, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa 160(%rdi), %xmm8
+; SSE-NEXT: movdqa 96(%rdi), %xmm12
+; SSE-NEXT: movdqa %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa 112(%rdi), %xmm10
+; SSE-NEXT: movdqa 240(%rdi), %xmm11
+; SSE-NEXT: movdqa %xmm11, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa 256(%rdi), %xmm7
+; SSE-NEXT: movdqa 192(%rdi), %xmm0
+; SSE-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa 208(%rdi), %xmm1
+; SSE-NEXT: movdqa 64(%rdi), %xmm6
+; SSE-NEXT: movdqa (%rdi), %xmm4
+; SSE-NEXT: movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa 16(%rdi), %xmm14
+; SSE-NEXT: movdqa 48(%rdi), %xmm5
+; SSE-NEXT: movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm14[2,3,2,3]
+; SSE-NEXT: movdqa %xmm14, (%rsp) # 16-byte Spill
+; SSE-NEXT: movdqa %xmm4, %xmm3
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm6[2,2,3,3]
+; SSE-NEXT: movdqa %xmm6, %xmm15
+; SSE-NEXT: movdqa %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm6 = xmm3[0],xmm6[1]
+; SSE-NEXT: movapd %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,2,3]
+; SSE-NEXT: movdqa %xmm1, %xmm13
+; SSE-NEXT: movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa %xmm0, %xmm3
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm7[2,2,3,3]
+; SSE-NEXT: movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm11[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm6 = xmm3[0],xmm6[1]
+; SSE-NEXT: movapd %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm10[2,3,2,3]
+; SSE-NEXT: movdqa %xmm10, %xmm11
+; SSE-NEXT: movdqa %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa %xmm12, %xmm3
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm8[2,2,3,3]
+; SSE-NEXT: movdqa %xmm8, %xmm6
+; SSE-NEXT: movdqa %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm9[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm1 = xmm3[0],xmm1[1]
+; SSE-NEXT: movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa 288(%rdi), %xmm9
+; SSE-NEXT: movdqa 304(%rdi), %xmm10
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm10[2,3,2,3]
+; SSE-NEXT: movdqa %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa %xmm9, %xmm4
+; SSE-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
+; SSE-NEXT: movdqa 336(%rdi), %xmm12
+; SSE-NEXT: movdqa 352(%rdi), %xmm5
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm5[2,2,3,3]
+; SSE-NEXT: movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm12[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
+; SSE-NEXT: movsd {{.*#+}} xmm1 = xmm4[0],xmm1[1]
+; SSE-NEXT: movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Reload
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm8[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm14[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm15[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
+; SSE-NEXT: movdqa %xmm2, %xmm0
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm3[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm1[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm13[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm7[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
+; SSE-NEXT: movdqa %xmm7, %xmm0
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm3[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Reload
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm13[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm11[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm6[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
+; SSE-NEXT: movdqa %xmm6, %xmm0
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm3[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm9[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm10[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm4 = xmm5[2,3,2,3]
+; SSE-NEXT: movdqa %xmm12, %xmm0
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm3[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,2,3,3]
+; SSE-NEXT: movdqa 80(%rdi), %xmm2
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE-NEXT: movdqa %xmm8, %xmm14
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm8[2,3,2,3]
+; SSE-NEXT: movdqa 32(%rdi), %xmm4
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE-NEXT: movdqa %xmm4, %xmm5
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm3[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm7[2,2,3,3]
+; SSE-NEXT: movdqa 272(%rdi), %xmm3
+; SSE-NEXT: movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm1[2,3,2,3]
+; SSE-NEXT: movdqa %xmm1, %xmm10
+; SSE-NEXT: movdqa 224(%rdi), %xmm4
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm3[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm12[2,2,3,3]
+; SSE-NEXT: movdqa 368(%rdi), %xmm15
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm15[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm9[2,3,2,3]
+; SSE-NEXT: movdqa 320(%rdi), %xmm11
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm11[0],xmm3[1],xmm11[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm3[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm6[2,2,3,3]
+; SSE-NEXT: movdqa 176(%rdi), %xmm1
+; SSE-NEXT: movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm1[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE-NEXT: movdqa %xmm13, %xmm6
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm13[2,3,2,3]
+; SSE-NEXT: movdqa 128(%rdi), %xmm8
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm3[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm14[3,3,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm5[1,1,1,1]
+; SSE-NEXT: movdqa %xmm5, %xmm7
+; SSE-NEXT: movdqa %xmm5, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE-NEXT: pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; SSE-NEXT: # xmm1 = mem[2,3,2,3]
+; SSE-NEXT: movdqa %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
+; SSE-NEXT: movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm10[3,3,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm4[1,1,1,1]
+; SSE-NEXT: movdqa %xmm4, %xmm14
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE-NEXT: pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 16-byte Folded Reload
+; SSE-NEXT: # xmm13 = mem[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
+; SSE-NEXT: punpckldq {{.*#+}} xmm13 = xmm13[0],xmm3[0],xmm13[1],xmm3[1]
+; SSE-NEXT: movsd {{.*#+}} xmm13 = xmm0[0],xmm13[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm9[3,3,3,3]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm11[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm12 = xmm12[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm12 = xmm12[0],xmm15[0],xmm12[1],xmm15[1]
+; SSE-NEXT: movsd {{.*#+}} xmm12 = xmm0[0],xmm12[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm6[3,3,3,3]
+; SSE-NEXT: movdqa %xmm8, %xmm5
+; SSE-NEXT: movdqa %xmm8, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm8[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE-NEXT: pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 16-byte Folded Reload
+; SSE-NEXT: # xmm10 = mem[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
+; SSE-NEXT: punpckldq {{.*#+}} xmm10 = xmm10[0],xmm4[0],xmm10[1],xmm4[1]
+; SSE-NEXT: movsd {{.*#+}} xmm10 = xmm0[0],xmm10[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm7[2,3,2,3]
+; SSE-NEXT: movdqa (%rsp), %xmm1 # 16-byte Reload
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm2[2,2,3,3]
+; SSE-NEXT: pshufd $80, {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
+; SSE-NEXT: # xmm8 = mem[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm8 = xmm8[0],xmm0[0],xmm8[1],xmm0[1]
+; SSE-NEXT: movsd {{.*#+}} xmm8 = xmm1[0],xmm8[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm14[2,3,2,3]
+; SSE-NEXT: movdqa %xmm14, %xmm7
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm3[2,2,3,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 16-byte Reload
+; SSE-NEXT: pshufd {{.*#+}} xmm6 = xmm9[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1]
+; SSE-NEXT: movsd {{.*#+}} xmm6 = xmm1[0],xmm6[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm5[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm4[2,2,3,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Reload
+; SSE-NEXT: pshufd {{.*#+}} xmm5 = xmm14[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm5 = xmm5[0],xmm1[0],xmm5[1],xmm1[1]
+; SSE-NEXT: movsd {{.*#+}} xmm5 = xmm0[0],xmm5[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm11[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
+; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm4[1,1,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm15[2,2,3,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm0[0,0,1,1]
+; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm3 = xmm4[0],xmm3[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm11[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm15[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
+; SSE-NEXT: movapd %xmm0, %xmm11
+; SSE-NEXT: pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; SSE-NEXT: # xmm1 = mem[1,1,1,1]
+; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm7[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; SSE-NEXT: pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
+; SSE-NEXT: # xmm2 = mem[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm9 = xmm9[0],xmm2[0],xmm9[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm9 = xmm1[0],xmm9[1]
+; SSE-NEXT: pshufd $85, (%rsp), %xmm1 # 16-byte Folded Reload
+; SSE-NEXT: # xmm1 = mem[1,1,1,1]
+; SSE-NEXT: pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
+; SSE-NEXT: # xmm2 = mem[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; SSE-NEXT: pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
+; SSE-NEXT: # xmm2 = mem[2,3,2,3]
+; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 16-byte Reload
+; SSE-NEXT: punpckldq {{.*#+}} xmm7 = xmm7[0],xmm2[0],xmm7[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm7 = xmm1[0],xmm7[1]
+; SSE-NEXT: pshufd $85, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; SSE-NEXT: # xmm1 = mem[1,1,1,1]
+; SSE-NEXT: pshufd $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
+; SSE-NEXT: # xmm2 = mem[3,3,3,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; SSE-NEXT: pshufd $238, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Folded Reload
+; SSE-NEXT: # xmm2 = mem[2,3,2,3]
+; SSE-NEXT: punpckldq {{.*#+}} xmm14 = xmm14[0],xmm2[0],xmm14[1],xmm2[1]
+; SSE-NEXT: movsd {{.*#+}} xmm14 = xmm1[0],xmm14[1]
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 48(%rsi)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 16(%rsi)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 32(%rsi)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; SSE-NEXT: movaps %xmm1, (%rsi)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 48(%rdx)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 16(%rdx)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 32(%rdx)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, (%rdx)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 16(%rcx)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 48(%rcx)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 32(%rcx)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, (%rcx)
+; SSE-NEXT: movapd %xmm10, 16(%r8)
+; SSE-NEXT: movapd %xmm12, 48(%r8)
+; SSE-NEXT: movapd %xmm13, 32(%r8)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, (%r8)
+; SSE-NEXT: movapd %xmm3, 48(%r9)
+; SSE-NEXT: movapd %xmm5, 16(%r9)
+; SSE-NEXT: movapd %xmm6, 32(%r9)
+; SSE-NEXT: movapd %xmm8, (%r9)
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: movapd %xmm14, 16(%rax)
+; SSE-NEXT: movapd %xmm7, (%rax)
+; SSE-NEXT: movapd %xmm9, 32(%rax)
+; SSE-NEXT: movapd %xmm11, 48(%rax)
+; SSE-NEXT: addq $376, %rsp # imm = 0x178
+; SSE-NEXT: retq
+;
+; AVX1-LABEL: load_i32_stride6_vf16:
+; AVX1: # %bb.0:
+; AVX1-NEXT: subq $264, %rsp # imm = 0x108
+; AVX1-NEXT: vmovaps 32(%rdi), %ymm7
+; AVX1-NEXT: vmovups %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps (%rdi), %ymm8
+; AVX1-NEXT: vmovups %ymm8, (%rsp) # 32-byte Spill
+; AVX1-NEXT: vmovaps 288(%rdi), %ymm2
+; AVX1-NEXT: vmovaps 256(%rdi), %ymm3
+; AVX1-NEXT: vmovaps 224(%rdi), %ymm1
+; AVX1-NEXT: vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 192(%rdi), %ymm4
+; AVX1-NEXT: vmovups %ymm4, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 320(%rdi), %ymm0
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 352(%rdi), %ymm5
+; AVX1-NEXT: vmovups %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vblendps {{.*#+}} ymm10 = ymm5[0,1,2,3],ymm0[4,5],ymm5[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm10, %xmm0
+; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; AVX1-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0,1],xmm10[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,1,0,2]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm5
+; AVX1-NEXT: vblendps {{.*#+}} ymm15 = ymm4[0,1,2,3],ymm1[4,5],ymm4[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm15, %xmm4
+; AVX1-NEXT: vblendps {{.*#+}} xmm1 = xmm15[0,1],xmm4[2,3]
+; AVX1-NEXT: vshufps {{.*#+}} xmm6 = xmm1[0,2],xmm4[0,3]
+; AVX1-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm13 = ymm3[0,1],ymm2[0,1]
+; AVX1-NEXT: vshufps {{.*#+}} ymm14 = ymm2[2,0],ymm13[0,0],ymm2[6,4],ymm13[4,4]
+; AVX1-NEXT: vshufps {{.*#+}} ymm14 = ymm14[2,0],ymm3[2,2],ymm14[6,4],ymm3[6,6]
+; AVX1-NEXT: vblendps {{.*#+}} ymm6 = ymm6[0,1,2],ymm14[3,4,5],ymm6[6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm6[0,1,2,3,4,5],ymm5[6,7]
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vblendps {{.*#+}} ymm14 = ymm8[0,1,2,3],ymm7[4,5],ymm8[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm14, %xmm1
+; AVX1-NEXT: vblendps {{.*#+}} xmm5 = xmm14[0,1],xmm1[2,3]
+; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm5[0,2],xmm1[0,3]
+; AVX1-NEXT: vmovaps 96(%rdi), %ymm9
+; AVX1-NEXT: vmovaps 64(%rdi), %ymm8
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm3 = ymm8[0,1],ymm9[0,1]
+; AVX1-NEXT: vshufps {{.*#+}} ymm12 = ymm9[2,0],ymm3[0,0],ymm9[6,4],ymm3[4,4]
+; AVX1-NEXT: vshufps {{.*#+}} ymm12 = ymm12[2,0],ymm8[2,2],ymm12[6,4],ymm8[6,6]
+; AVX1-NEXT: vblendps {{.*#+}} ymm5 = ymm0[0,1,2],ymm12[3,4,5],ymm0[6,7]
+; AVX1-NEXT: vmovaps 128(%rdi), %ymm12
+; AVX1-NEXT: vmovaps 160(%rdi), %ymm6
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm6[0,1,2,3],ymm12[4,5],ymm6[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT: vblendps {{.*#+}} xmm11 = xmm2[0,1],xmm0[2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm7 = xmm11[0,1,0,2]
+; AVX1-NEXT: vinsertf128 $1, %xmm7, %ymm0, %ymm7
+; AVX1-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0,1,2,3,4,5],ymm7[6,7]
+; AVX1-NEXT: vmovups %ymm5, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vinsertps {{.*#+}} xmm0 = zero,zero,xmm2[1],xmm0[3]
+; AVX1-NEXT: vshufps {{.*#+}} xmm2 = xmm14[1,0],xmm1[3,0]
+; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm2[0,2],xmm1[1,3]
+; AVX1-NEXT: vshufps {{.*#+}} ymm2 = ymm9[3,0],ymm3[1,0],ymm9[7,4],ymm3[5,4]
+; AVX1-NEXT: vshufps {{.*#+}} ymm2 = ymm2[2,0],ymm8[2,3],ymm2[6,4],ymm8[6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3,4,5],ymm1[6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3,4,5],ymm0[6,7]
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vinsertps $35, {{[-0-9]+}}(%r{{[sb]}}p), %xmm10, %xmm0 # 16-byte Folded Reload
+; AVX1-NEXT: # xmm0 = zero,zero,mem[0],xmm10[3]
+; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm15[1,0],xmm4[3,0]
+; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm1[0,2],xmm4[1,3]
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
+; AVX1-NEXT: vshufps {{.*#+}} ymm2 = ymm4[3,0],ymm13[1,0],ymm4[7,4],ymm13[5,4]
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm13 # 32-byte Reload
+; AVX1-NEXT: vshufps {{.*#+}} ymm2 = ymm2[2,0],ymm13[2,3],ymm2[6,4],ymm13[6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3,4,5],ymm1[6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1,2,3,4,5],ymm0[6,7]
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm12[0,1,2,3],ymm6[4,5],ymm12[6,7]
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX1-NEXT: vblendps $12, (%rsp), %ymm0, %ymm3 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm3 = ymm0[0,1],mem[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm15
+; AVX1-NEXT: vshufps {{.*#+}} ymm0 = ymm9[2,1],ymm8[2,0],ymm9[6,5],ymm8[6,4]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm3[2,0],xmm15[2,3]
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4],ymm1[5,6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm6
+; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm2[0,0],xmm6[2,0]
+; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX1-NEXT: vmovups %ymm0, (%rsp) # 32-byte Spill
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX1-NEXT: vblendps $48, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm7 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm7 = ymm0[0,1,2,3],mem[4,5],ymm0[6,7]
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX1-NEXT: vblendps $12, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm10 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm10 = ymm0[0,1],mem[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm1 = ymm4[2,1],ymm13[2,0],ymm4[6,5],ymm13[6,4]
+; AVX1-NEXT: vmovaps %ymm4, %ymm11
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm1[2,3,0,1]
+; AVX1-NEXT: vextractf128 $1, %ymm10, %xmm4
+; AVX1-NEXT: vshufps {{.*#+}} xmm12 = xmm10[2,0],xmm4[2,3]
+; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm12[0,1,2],ymm1[3,4],ymm12[5,6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm7, %xmm0
+; AVX1-NEXT: vshufps {{.*#+}} xmm5 = xmm7[0,0],xmm0[2,0]
+; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm0, %ymm5
+; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm5[5,6,7]
+; AVX1-NEXT: vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vshufps {{.*#+}} xmm2 = xmm2[0,1],xmm6[3,1]
+; AVX1-NEXT: vshufps {{.*#+}} xmm3 = xmm3[3,1],xmm15[3,3]
+; AVX1-NEXT: vshufps {{.*#+}} ymm5 = ymm9[3,1],ymm8[2,1],ymm9[7,5],ymm8[6,5]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm5 = ymm5[2,3,0,1]
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1,2],ymm5[3,4],ymm3[5,6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm2
+; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm3[0,1,2,3,4],ymm2[5,6,7]
+; AVX1-NEXT: vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm7[0,1],xmm0[3,1]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT: vshufps {{.*#+}} xmm3 = xmm10[3,1],xmm4[3,3]
+; AVX1-NEXT: vshufps {{.*#+}} ymm4 = ymm11[3,1],ymm13[2,1],ymm11[7,5],ymm13[6,5]
+; AVX1-NEXT: vmovaps %ymm11, %ymm5
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm4 = ymm4[2,3,0,1]
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1,2],ymm4[3,4],ymm3[5,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm3[0,1,2,3,4],ymm0[5,6,7]
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 32(%rdi), %xmm12
+; AVX1-NEXT: vpermilps {{.*#+}} xmm4 = xmm12[2,2,3,3]
+; AVX1-NEXT: vmovaps 16(%rdi), %xmm10
+; AVX1-NEXT: vblendps {{.*#+}} xmm4 = xmm10[0],xmm4[1],xmm10[2,3]
+; AVX1-NEXT: vmovapd 80(%rdi), %xmm11
+; AVX1-NEXT: vshufpd {{.*#+}} ymm7 = ymm11[1],ymm8[0],ymm11[2],ymm8[2]
+; AVX1-NEXT: vshufps {{.*#+}} ymm7 = ymm9[0,1],ymm7[2,0],ymm9[4,5],ymm7[6,4]
+; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1],ymm7[2,3,4,5,6,7]
+; AVX1-NEXT: vmovdqa 160(%rdi), %xmm7
+; AVX1-NEXT: vpblendw {{.*#+}} xmm7 = xmm7[0,1,2,3],mem[4,5,6,7]
+; AVX1-NEXT: vmovdqa 176(%rdi), %xmm2
+; AVX1-NEXT: vpalignr {{.*#+}} xmm1 = xmm2[8,9,10,11,12,13,14,15],xmm7[0,1,2,3,4,5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm7[0,2],xmm1[2,0]
+; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm4[0,1,2,3,4],ymm1[5,6,7]
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 224(%rdi), %xmm1
+; AVX1-NEXT: vpermilps {{.*#+}} xmm4 = xmm1[2,2,3,3]
+; AVX1-NEXT: vmovaps 208(%rdi), %xmm3
+; AVX1-NEXT: vblendps {{.*#+}} xmm4 = xmm3[0],xmm4[1],xmm3[2,3]
+; AVX1-NEXT: vmovapd 272(%rdi), %xmm0
+; AVX1-NEXT: vshufpd {{.*#+}} ymm14 = ymm0[1],ymm13[0],ymm0[2],ymm13[2]
+; AVX1-NEXT: vshufps {{.*#+}} ymm14 = ymm5[0,1],ymm14[2,0],ymm5[4,5],ymm14[6,4]
+; AVX1-NEXT: vmovaps %ymm5, %ymm13
+; AVX1-NEXT: vblendps {{.*#+}} ymm14 = ymm4[0,1],ymm14[2,3,4,5,6,7]
+; AVX1-NEXT: vmovdqa 352(%rdi), %xmm5
+; AVX1-NEXT: vpblendw {{.*#+}} xmm5 = xmm5[0,1,2,3],mem[4,5,6,7]
+; AVX1-NEXT: vmovdqa 368(%rdi), %xmm4
+; AVX1-NEXT: vpalignr {{.*#+}} xmm15 = xmm4[8,9,10,11,12,13,14,15],xmm5[0,1,2,3,4,5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} xmm6 = xmm5[0,2],xmm15[2,0]
+; AVX1-NEXT: vinsertf128 $1, %xmm6, %ymm0, %ymm6
+; AVX1-NEXT: vblendps {{.*#+}} ymm6 = ymm14[0,1,2,3,4],ymm6[5,6,7]
+; AVX1-NEXT: vpalignr {{.*#+}} xmm2 = xmm2[12,13,14,15],xmm7[0,1,2,3,4,5,6,7,8,9,10,11]
+; AVX1-NEXT: vshufps {{.*#+}} xmm2 = xmm7[0,3],xmm2[2,0]
+; AVX1-NEXT: vblendps {{.*#+}} xmm7 = xmm10[0,1],xmm12[2,3]
+; AVX1-NEXT: vshufps {{.*#+}} ymm8 = ymm11[3,1],ymm8[1,3],ymm11[7,5],ymm8[5,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm8 = ymm9[1,1],ymm8[2,0],ymm9[5,5],ymm8[6,4]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm7 = xmm7[1,3,2,3]
+; AVX1-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1],ymm8[2,3,4,5,6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm2
+; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm7[0,1,2,3,4],ymm2[5,6,7]
+; AVX1-NEXT: vpalignr {{.*#+}} xmm4 = xmm4[12,13,14,15],xmm5[0,1,2,3,4,5,6,7,8,9,10,11]
+; AVX1-NEXT: vshufps {{.*#+}} xmm4 = xmm5[0,3],xmm4[2,0]
+; AVX1-NEXT: vblendps {{.*#+}} xmm1 = xmm3[0,1],xmm1[2,3]
+; AVX1-NEXT: vshufps $215, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm0 = ymm0[3,1],mem[1,3],ymm0[7,5],mem[5,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm0 = ymm13[1,1],ymm0[2,0],ymm13[5,5],ymm0[6,4]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm1[1,3,2,3]
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3,4,5,6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm0, %ymm1
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, (%rsi)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, 32(%rsi)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, 32(%rdx)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, (%rdx)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, 32(%rcx)
+; AVX1-NEXT: vmovups (%rsp), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, (%rcx)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, 32(%r8)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, (%r8)
+; AVX1-NEXT: vmovaps %ymm6, 32(%r9)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm1, (%r9)
+; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX1-NEXT: vmovaps %ymm0, 32(%rax)
+; AVX1-NEXT: vmovaps %ymm2, (%rax)
+; AVX1-NEXT: addq $264, %rsp # imm = 0x108
+; AVX1-NEXT: vzeroupper
+; AVX1-NEXT: retq
+;
+; AVX2-SLOW-LABEL: load_i32_stride6_vf16:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: subq $232, %rsp
+; AVX2-SLOW-NEXT: vmovaps 96(%rdi), %ymm8
+; AVX2-SLOW-NEXT: vmovaps 64(%rdi), %ymm3
+; AVX2-SLOW-NEXT: vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vmovaps 320(%rdi), %ymm2
+; AVX2-SLOW-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vmovaps 352(%rdi), %ymm0
+; AVX2-SLOW-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vmovaps 288(%rdi), %ymm7
+; AVX2-SLOW-NEXT: vmovaps 256(%rdi), %ymm1
+; AVX2-SLOW-NEXT: vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vmovaps 224(%rdi), %ymm9
+; AVX2-SLOW-NEXT: vmovaps 192(%rdi), %ymm10
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm4 = <0,6,4,u>
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm5 = ymm10[0,1,2,3],ymm9[4,5],ymm10[6,7]
+; AVX2-SLOW-NEXT: vpermps %ymm5, %ymm4, %ymm6
+; AVX2-SLOW-NEXT: vperm2f128 {{.*#+}} ymm11 = ymm1[0,1],ymm7[0,1]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm12 = ymm11[0,1,2,2,4,5,6,6]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm13 = ymm7[2,2,2,2,6,6,6,6]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm12 = ymm12[0],ymm13[1],ymm12[2,3,4],ymm13[5],ymm12[6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm6 = ymm6[0,1,2],ymm12[3,4,5],ymm6[6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm4 = ymm0[0,1,2,3],ymm2[4,5],ymm0[6,7]
+; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm2 = [8589934596,8589934596,8589934596,8589934596]
+; AVX2-SLOW-NEXT: vpermps %ymm4, %ymm2, %ymm12
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm6[0,1,2,3,4,5],ymm12[6,7]
+; AVX2-SLOW-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vperm2f128 {{.*#+}} ymm6 = ymm3[0,1],ymm8[0,1]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm12 = ymm6[0,1,2,2,4,5,6,6]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm13 = ymm8[2,2,2,2,6,6,6,6]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm12 = ymm12[0],ymm13[1],ymm12[2,3,4],ymm13[5],ymm12[6,7]
+; AVX2-SLOW-NEXT: vmovaps (%rdi), %ymm14
+; AVX2-SLOW-NEXT: vmovaps 32(%rdi), %ymm15
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm14[0,1,2,3],ymm15[4,5],ymm14[6,7]
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm1 = <0,6,4,u>
+; AVX2-SLOW-NEXT: vpermps %ymm0, %ymm1, %ymm13
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm3 = ymm13[0,1,2],ymm12[3,4,5],ymm13[6,7]
+; AVX2-SLOW-NEXT: vmovaps 128(%rdi), %ymm12
+; AVX2-SLOW-NEXT: vmovaps 160(%rdi), %ymm13
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm13[0,1,2,3],ymm12[4,5],ymm13[6,7]
+; AVX2-SLOW-NEXT: vpermps %ymm1, %ymm2, %ymm2
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm2 = ymm3[0,1,2,3,4,5],ymm2[6,7]
+; AVX2-SLOW-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm2 = ymm6[1,1,2,3,5,5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm3 = ymm8[2,3,2,3,6,7,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0],ymm3[1],ymm2[2,3,4],ymm3[5],ymm2[6,7]
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm3 = <1,7,5,u>
+; AVX2-SLOW-NEXT: vpermps %ymm0, %ymm3, %ymm0
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm2[3,4,5],ymm0[6,7]
+; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm2 = [12884901893,12884901893,12884901893,12884901893]
+; AVX2-SLOW-NEXT: vpermps %ymm1, %ymm2, %ymm1
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
+; AVX2-SLOW-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vpermps %ymm5, %ymm3, %ymm0
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm11[1,1,2,3,5,5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm3 = ymm7[2,3,2,3,6,7,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm3[1],ymm1[2,3,4],ymm3[5],ymm1[6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4,5],ymm0[6,7]
+; AVX2-SLOW-NEXT: vpermps %ymm4, %ymm2, %ymm1
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
+; AVX2-SLOW-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm14[2,3,2,3,6,7,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm15[0,0,2,3,4,4,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm6[0,1],ymm8[2,3],ymm6[4,5],ymm8[6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[2,0,2,3,6,4,6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm13[0,0,0,0,4,4,4,4]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm2 = ymm12[0,0,2,3,4,4,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm10[2,3,2,3,6,7,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm9[0,0,2,3,4,4,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm5[0,1],ymm7[2,3],ymm5[4,5],ymm7[6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[2,0,2,3,6,4,6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm3[0,0,0,0,4,4,4,4]
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm2 = ymm4[0,0,2,3,4,4,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm14[3,3,3,3,7,7,7,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm15[0,1,3,3,4,5,7,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm8[3,3,3,3,7,7,7,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm6[1],ymm1[2,3,4],ymm6[5],ymm1[6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm13[0,1,0,1,4,5,4,5]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm2 = ymm12[0,1,3,3,4,5,7,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vmovups %ymm0, (%rsp) # 32-byte Spill
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm10[3,3,3,3,7,7,7,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm9[0,1,3,3,4,5,7,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm7[3,3,3,3,7,7,7,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm5[1],ymm1[2,3,4],ymm5[5],ymm1[6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm3[0,1,0,1,4,5,4,5]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm2 = ymm4[0,1,3,3,4,5,7,7]
+; AVX2-SLOW-NEXT: vmovaps %ymm4, %ymm11
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm2 = ymm9[0,1,2,3],ymm10[4,5],ymm9[6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3],ymm7[4,5,6,7]
+; AVX2-SLOW-NEXT: vmovaps 272(%rdi), %xmm5
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm4[0,1],ymm5[2,3],ymm4[4,5],ymm5[6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,1,0,2,4,5,4,6]
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm7 = <4,2,u,u>
+; AVX2-SLOW-NEXT: vpermps %ymm2, %ymm7, %ymm9
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm9[0,1],ymm1[2,3,4,5,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm9 = ymm3[0,1],ymm11[2,3],ymm3[4,5,6,7]
+; AVX2-SLOW-NEXT: vbroadcastf128 {{.*#+}} ymm10 = [0,2,0,6,0,2,0,6]
+; AVX2-SLOW-NEXT: # ymm10 = mem[0,1,0,1]
+; AVX2-SLOW-NEXT: vpermps %ymm9, %ymm10, %ymm11
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm11[5,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm8 = ymm6[0,1,2,3],ymm8[4,5,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm11 = ymm15[0,1,2,3],ymm14[4,5],ymm15[6,7]
+; AVX2-SLOW-NEXT: vmovaps 80(%rdi), %xmm14
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm15 = ymm8[0,1],ymm14[2,3],ymm8[4,5],ymm14[6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm15 = ymm15[0,1,0,2,4,5,4,6]
+; AVX2-SLOW-NEXT: vpermps %ymm11, %ymm7, %ymm7
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1],ymm15[2,3,4,5,6,7]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm12 = ymm13[0,1],ymm12[2,3],ymm13[4,5,6,7]
+; AVX2-SLOW-NEXT: vpermps %ymm12, %ymm10, %ymm10
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4],ymm10[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm4 = ymm4[1,1,1,1,5,5,5,5]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2],ymm5[3],ymm4[4,5,6],ymm5[7]
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} xmm5 = <5,3,u,u>
+; AVX2-SLOW-NEXT: vpermps %ymm2, %ymm5, %ymm2
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm4[2,3,4,5,6,7]
+; AVX2-SLOW-NEXT: vbroadcastf128 {{.*#+}} ymm4 = [0,3,1,7,0,3,1,7]
+; AVX2-SLOW-NEXT: # ymm4 = mem[0,1,0,1]
+; AVX2-SLOW-NEXT: vpermps %ymm9, %ymm4, %ymm9
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3,4],ymm9[5,6,7]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[1,1,1,1,5,5,5,5]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1,2],ymm14[3],ymm8[4,5,6],ymm14[7]
+; AVX2-SLOW-NEXT: vpermps %ymm11, %ymm5, %ymm5
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0,1],ymm8[2,3,4,5,6,7]
+; AVX2-SLOW-NEXT: vpermps %ymm12, %ymm4, %ymm4
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3,4],ymm4[5,6,7]
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-SLOW-NEXT: vmovaps %ymm3, (%rsi)
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-SLOW-NEXT: vmovaps %ymm5, 32(%rsi)
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-SLOW-NEXT: vmovaps %ymm3, 32(%rdx)
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-SLOW-NEXT: vmovaps %ymm3, (%rdx)
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-SLOW-NEXT: vmovaps %ymm3, 32(%rcx)
+; AVX2-SLOW-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-SLOW-NEXT: vmovaps %ymm3, (%rcx)
+; AVX2-SLOW-NEXT: vmovaps %ymm0, 32(%r8)
+; AVX2-SLOW-NEXT: vmovups (%rsp), %ymm0 # 32-byte Reload
+; AVX2-SLOW-NEXT: vmovaps %ymm0, (%r8)
+; AVX2-SLOW-NEXT: vmovaps %ymm1, 32(%r9)
+; AVX2-SLOW-NEXT: vmovaps %ymm7, (%r9)
+; AVX2-SLOW-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-SLOW-NEXT: vmovaps %ymm2, 32(%rax)
+; AVX2-SLOW-NEXT: vmovaps %ymm4, (%rax)
+; AVX2-SLOW-NEXT: addq $232, %rsp
+; AVX2-SLOW-NEXT: vzeroupper
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-ALL-LABEL: load_i32_stride6_vf16:
+; AVX2-FAST-ALL: # %bb.0:
+; AVX2-FAST-ALL-NEXT: subq $232, %rsp
+; AVX2-FAST-ALL-NEXT: vmovaps 96(%rdi), %ymm8
+; AVX2-FAST-ALL-NEXT: vmovaps 64(%rdi), %ymm1
+; AVX2-FAST-ALL-NEXT: vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vmovaps 320(%rdi), %ymm3
+; AVX2-FAST-ALL-NEXT: vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vmovaps 352(%rdi), %ymm0
+; AVX2-FAST-ALL-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vmovaps 288(%rdi), %ymm7
+; AVX2-FAST-ALL-NEXT: vmovaps 256(%rdi), %ymm2
+; AVX2-FAST-ALL-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vmovaps 224(%rdi), %ymm9
+; AVX2-FAST-ALL-NEXT: vmovaps 192(%rdi), %ymm10
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm4 = <0,6,4,u>
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm5 = ymm10[0,1,2,3],ymm9[4,5],ymm10[6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm5, %ymm4, %ymm6
+; AVX2-FAST-ALL-NEXT: vperm2f128 {{.*#+}} ymm11 = ymm2[0,1],ymm7[0,1]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm12 = ymm11[0,1,2,2,4,5,6,6]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm13 = ymm7[2,2,2,2,6,6,6,6]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm12 = ymm12[0],ymm13[1],ymm12[2,3,4],ymm13[5],ymm12[6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm6 = ymm6[0,1,2],ymm12[3,4,5],ymm6[6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm4 = ymm0[0,1,2,3],ymm3[4,5],ymm0[6,7]
+; AVX2-FAST-ALL-NEXT: vbroadcastsd {{.*#+}} ymm2 = [8589934596,8589934596,8589934596,8589934596]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm4, %ymm2, %ymm12
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm6[0,1,2,3,4,5],ymm12[6,7]
+; AVX2-FAST-ALL-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vperm2f128 {{.*#+}} ymm6 = ymm1[0,1],ymm8[0,1]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm12 = ymm6[0,1,2,2,4,5,6,6]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm13 = ymm8[2,2,2,2,6,6,6,6]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm12 = ymm12[0],ymm13[1],ymm12[2,3,4],ymm13[5],ymm12[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps (%rdi), %ymm14
+; AVX2-FAST-ALL-NEXT: vmovaps 32(%rdi), %ymm15
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm14[0,1,2,3],ymm15[4,5],ymm14[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm1 = <0,6,4,u>
+; AVX2-FAST-ALL-NEXT: vpermps %ymm0, %ymm1, %ymm13
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm3 = ymm13[0,1,2],ymm12[3,4,5],ymm13[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps 128(%rdi), %ymm12
+; AVX2-FAST-ALL-NEXT: vmovaps 160(%rdi), %ymm13
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm13[0,1,2,3],ymm12[4,5],ymm13[6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm1, %ymm2, %ymm2
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm2 = ymm3[0,1,2,3,4,5],ymm2[6,7]
+; AVX2-FAST-ALL-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm2 = ymm6[1,1,2,3,5,5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm3 = ymm8[2,3,2,3,6,7,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0],ymm3[1],ymm2[2,3,4],ymm3[5],ymm2[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm3 = <1,7,5,u>
+; AVX2-FAST-ALL-NEXT: vpermps %ymm0, %ymm3, %ymm0
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm2[3,4,5],ymm0[6,7]
+; AVX2-FAST-ALL-NEXT: vbroadcastsd {{.*#+}} ymm2 = [12884901893,12884901893,12884901893,12884901893]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm1, %ymm2, %ymm1
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
+; AVX2-FAST-ALL-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vpermps %ymm5, %ymm3, %ymm0
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm11[1,1,2,3,5,5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm3 = ymm7[2,3,2,3,6,7,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm3[1],ymm1[2,3,4],ymm3[5],ymm1[6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4,5],ymm0[6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm4, %ymm2, %ymm1
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
+; AVX2-FAST-ALL-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm0 = ymm14[2,3,2,3,6,7,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm15[0,0,2,3,4,4,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm6[0,1],ymm8[2,3],ymm6[4,5],ymm8[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} ymm2 = [2,0,6,4,2,0,6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm1, %ymm2, %ymm1
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm13[0,0,0,0,4,4,4,4]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm3 = ymm12[0,0,2,3,4,4,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm3[0,1,2],ymm1[3],ymm3[4,5,6],ymm1[7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm5[0,1],ymm7[2,3],ymm5[4,5],ymm7[6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm0, %ymm2, %ymm0
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm10[2,3,2,3,6,7,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm2 = ymm9[0,0,2,3,4,4,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm2[1,2,3],ymm1[4],ymm2[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,3,2,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm3[0,0,0,0,4,4,4,4]
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm2 = ymm4[0,0,2,3,4,4,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm0 = ymm14[3,3,3,3,7,7,7,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm15[0,1,3,3,4,5,7,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm8[3,3,3,3,7,7,7,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm6[1],ymm1[2,3,4],ymm6[5],ymm1[6,7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm13[0,1,0,1,4,5,4,5]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm2 = ymm12[0,1,3,3,4,5,7,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovups %ymm0, (%rsp) # 32-byte Spill
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm0 = ymm10[3,3,3,3,7,7,7,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm9[0,1,3,3,4,5,7,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm7[3,3,3,3,7,7,7,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm5[1],ymm1[2,3,4],ymm5[5],ymm1[6,7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm3[0,1,0,1,4,5,4,5]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm2 = ymm4[0,1,3,3,4,5,7,7]
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm4, %ymm11
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm2 = ymm9[0,1,2,3],ymm10[4,5],ymm9[6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3],ymm7[4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps 272(%rdi), %xmm5
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm4[0,1],ymm5[2,3],ymm4[4,5],ymm5[6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,1,0,2,4,5,4,6]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm7 = <4,2,u,u>
+; AVX2-FAST-ALL-NEXT: vpermps %ymm2, %ymm7, %ymm9
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm9[0,1],ymm1[2,3,4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm9 = ymm3[0,1],ymm11[2,3],ymm3[4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vbroadcastf128 {{.*#+}} ymm10 = [0,2,0,6,0,2,0,6]
+; AVX2-FAST-ALL-NEXT: # ymm10 = mem[0,1,0,1]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm9, %ymm10, %ymm11
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm11[5,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm8 = ymm6[0,1,2,3],ymm8[4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm11 = ymm15[0,1,2,3],ymm14[4,5],ymm15[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps 80(%rdi), %xmm14
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm15 = ymm8[0,1],ymm14[2,3],ymm8[4,5],ymm14[6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm15 = ymm15[0,1,0,2,4,5,4,6]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm11, %ymm7, %ymm7
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1],ymm15[2,3,4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm12 = ymm13[0,1],ymm12[2,3],ymm13[4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm12, %ymm10, %ymm10
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4],ymm10[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm4 = ymm4[1,1,1,1,5,5,5,5]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2],ymm5[3],ymm4[4,5,6],ymm5[7]
+; AVX2-FAST-ALL-NEXT: vmovaps {{.*#+}} xmm5 = <5,3,u,u>
+; AVX2-FAST-ALL-NEXT: vpermps %ymm2, %ymm5, %ymm2
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm4[2,3,4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vbroadcastf128 {{.*#+}} ymm4 = [0,3,1,7,0,3,1,7]
+; AVX2-FAST-ALL-NEXT: # ymm4 = mem[0,1,0,1]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm9, %ymm4, %ymm9
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3,4],ymm9[5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[1,1,1,1,5,5,5,5]
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1,2],ymm14[3],ymm8[4,5,6],ymm14[7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm11, %ymm5, %ymm5
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0,1],ymm8[2,3,4,5,6,7]
+; AVX2-FAST-ALL-NEXT: vpermps %ymm12, %ymm4, %ymm4
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3,4],ymm4[5,6,7]
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm5, (%rsi)
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm5, 32(%rsi)
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm5, 32(%rdx)
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm5, (%rdx)
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm3, 32(%rcx)
+; AVX2-FAST-ALL-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm3, (%rcx)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, 32(%r8)
+; AVX2-FAST-ALL-NEXT: vmovups (%rsp), %ymm0 # 32-byte Reload
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, (%r8)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm1, 32(%r9)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm7, (%r9)
+; AVX2-FAST-ALL-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm2, 32(%rax)
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm4, (%rax)
+; AVX2-FAST-ALL-NEXT: addq $232, %rsp
+; AVX2-FAST-ALL-NEXT: vzeroupper
+; AVX2-FAST-ALL-NEXT: retq
+;
+; AVX2-FAST-PERLANE-LABEL: load_i32_stride6_vf16:
+; AVX2-FAST-PERLANE: # %bb.0:
+; AVX2-FAST-PERLANE-NEXT: subq $232, %rsp
+; AVX2-FAST-PERLANE-NEXT: vmovaps 96(%rdi), %ymm8
+; AVX2-FAST-PERLANE-NEXT: vmovaps 64(%rdi), %ymm3
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vmovaps 320(%rdi), %ymm2
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vmovaps 352(%rdi), %ymm0
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vmovaps 288(%rdi), %ymm7
+; AVX2-FAST-PERLANE-NEXT: vmovaps 256(%rdi), %ymm1
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vmovaps 224(%rdi), %ymm9
+; AVX2-FAST-PERLANE-NEXT: vmovaps 192(%rdi), %ymm10
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm4 = <0,6,4,u>
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm5 = ymm10[0,1,2,3],ymm9[4,5],ymm10[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm5, %ymm4, %ymm6
+; AVX2-FAST-PERLANE-NEXT: vperm2f128 {{.*#+}} ymm11 = ymm1[0,1],ymm7[0,1]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm12 = ymm11[0,1,2,2,4,5,6,6]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm13 = ymm7[2,2,2,2,6,6,6,6]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm12 = ymm12[0],ymm13[1],ymm12[2,3,4],ymm13[5],ymm12[6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm6 = ymm6[0,1,2],ymm12[3,4,5],ymm6[6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm4 = ymm0[0,1,2,3],ymm2[4,5],ymm0[6,7]
+; AVX2-FAST-PERLANE-NEXT: vbroadcastsd {{.*#+}} ymm2 = [8589934596,8589934596,8589934596,8589934596]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm4, %ymm2, %ymm12
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm6[0,1,2,3,4,5],ymm12[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vperm2f128 {{.*#+}} ymm6 = ymm3[0,1],ymm8[0,1]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm12 = ymm6[0,1,2,2,4,5,6,6]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm13 = ymm8[2,2,2,2,6,6,6,6]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm12 = ymm12[0],ymm13[1],ymm12[2,3,4],ymm13[5],ymm12[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps (%rdi), %ymm14
+; AVX2-FAST-PERLANE-NEXT: vmovaps 32(%rdi), %ymm15
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm14[0,1,2,3],ymm15[4,5],ymm14[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm1 = <0,6,4,u>
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm0, %ymm1, %ymm13
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm3 = ymm13[0,1,2],ymm12[3,4,5],ymm13[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps 128(%rdi), %ymm12
+; AVX2-FAST-PERLANE-NEXT: vmovaps 160(%rdi), %ymm13
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm13[0,1,2,3],ymm12[4,5],ymm13[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm1, %ymm2, %ymm2
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm2 = ymm3[0,1,2,3,4,5],ymm2[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm2 = ymm6[1,1,2,3,5,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm3 = ymm8[2,3,2,3,6,7,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0],ymm3[1],ymm2[2,3,4],ymm3[5],ymm2[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm3 = <1,7,5,u>
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm0, %ymm3, %ymm0
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm2[3,4,5],ymm0[6,7]
+; AVX2-FAST-PERLANE-NEXT: vbroadcastsd {{.*#+}} ymm2 = [12884901893,12884901893,12884901893,12884901893]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm1, %ymm2, %ymm1
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm5, %ymm3, %ymm0
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm11[1,1,2,3,5,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm3 = ymm7[2,3,2,3,6,7,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm3[1],ymm1[2,3,4],ymm3[5],ymm1[6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4,5],ymm0[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm4, %ymm2, %ymm1
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm1[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm0 = ymm14[2,3,2,3,6,7,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm15[0,0,2,3,4,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm6 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm6[0,1],ymm8[2,3],ymm6[4,5],ymm8[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[2,0,2,3,6,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm13[0,0,0,0,4,4,4,4]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm2 = ymm12[0,0,2,3,4,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm0 = ymm10[2,3,2,3,6,7,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm9[0,0,2,3,4,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm5[0,1],ymm7[2,3],ymm5[4,5],ymm7[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[2,0,2,3,6,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm3[0,0,0,0,4,4,4,4]
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm4 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm2 = ymm4[0,0,2,3,4,4,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm0 = ymm14[3,3,3,3,7,7,7,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm15[0,1,3,3,4,5,7,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm8[3,3,3,3,7,7,7,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm6[1],ymm1[2,3,4],ymm6[5],ymm1[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm13[0,1,0,1,4,5,4,5]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm2 = ymm12[0,1,3,3,4,5,7,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups %ymm0, (%rsp) # 32-byte Spill
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm0 = ymm10[3,3,3,3,7,7,7,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm9[0,1,3,3,4,5,7,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm7[3,3,3,3,7,7,7,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0],ymm5[1],ymm1[2,3,4],ymm5[5],ymm1[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,0,3]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4],ymm0[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm3[0,1,0,1,4,5,4,5]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm2 = ymm4[0,1,3,3,4,5,7,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm4, %ymm11
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6],ymm1[7]
+; AVX2-FAST-PERLANE-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm2 = ymm9[0,1,2,3],ymm10[4,5],ymm9[6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3],ymm7[4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps 272(%rdi), %xmm5
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm4[0,1],ymm5[2,3],ymm4[4,5],ymm5[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,1,0,2,4,5,4,6]
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm7 = <4,2,u,u>
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm2, %ymm7, %ymm9
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm9[0,1],ymm1[2,3,4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm9 = ymm3[0,1],ymm11[2,3],ymm3[4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vbroadcastf128 {{.*#+}} ymm10 = [0,2,0,6,0,2,0,6]
+; AVX2-FAST-PERLANE-NEXT: # ymm10 = mem[0,1,0,1]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm9, %ymm10, %ymm11
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3,4],ymm11[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm8 = ymm6[0,1,2,3],ymm8[4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm11 = ymm15[0,1,2,3],ymm14[4,5],ymm15[6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps 80(%rdi), %xmm14
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm15 = ymm8[0,1],ymm14[2,3],ymm8[4,5],ymm14[6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm15 = ymm15[0,1,0,2,4,5,4,6]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm11, %ymm7, %ymm7
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1],ymm15[2,3,4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm12 = ymm13[0,1],ymm12[2,3],ymm13[4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm12, %ymm10, %ymm10
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1,2,3,4],ymm10[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm4 = ymm4[1,1,1,1,5,5,5,5]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2],ymm5[3],ymm4[4,5,6],ymm5[7]
+; AVX2-FAST-PERLANE-NEXT: vmovaps {{.*#+}} xmm5 = <5,3,u,u>
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm2, %ymm5, %ymm2
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm4[2,3,4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vbroadcastf128 {{.*#+}} ymm4 = [0,3,1,7,0,3,1,7]
+; AVX2-FAST-PERLANE-NEXT: # ymm4 = mem[0,1,0,1]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm9, %ymm4, %ymm9
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3,4],ymm9[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermilps {{.*#+}} ymm8 = ymm8[1,1,1,1,5,5,5,5]
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm8 = ymm8[0,1,2],ymm14[3],ymm8[4,5,6],ymm14[7]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm11, %ymm5, %ymm5
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0,1],ymm8[2,3,4,5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vpermps %ymm12, %ymm4, %ymm4
+; AVX2-FAST-PERLANE-NEXT: vblendps {{.*#+}} ymm4 = ymm5[0,1,2,3,4],ymm4[5,6,7]
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm3, (%rsi)
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm5 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm5, 32(%rsi)
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm3, 32(%rdx)
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm3, (%rdx)
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm3, 32(%rcx)
+; AVX2-FAST-PERLANE-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm3 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm3, (%rcx)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, 32(%r8)
+; AVX2-FAST-PERLANE-NEXT: vmovups (%rsp), %ymm0 # 32-byte Reload
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm0, (%r8)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm1, 32(%r9)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm7, (%r9)
+; AVX2-FAST-PERLANE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm2, 32(%rax)
+; AVX2-FAST-PERLANE-NEXT: vmovaps %ymm4, (%rax)
+; AVX2-FAST-PERLANE-NEXT: addq $232, %rsp
+; AVX2-FAST-PERLANE-NEXT: vzeroupper
+; AVX2-FAST-PERLANE-NEXT: retq
+;
+; AVX512-LABEL: load_i32_stride6_vf16:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX512-NEXT: vmovdqu64 320(%rdi), %zmm0
+; AVX512-NEXT: vmovdqu64 256(%rdi), %zmm1
+; AVX512-NEXT: vmovdqu64 (%rdi), %zmm3
+; AVX512-NEXT: vmovdqu64 64(%rdi), %zmm4
+; AVX512-NEXT: vmovdqu64 128(%rdi), %zmm5
+; AVX512-NEXT: vmovdqu64 192(%rdi), %zmm6
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm7 = [0,6,12,0,0,0,20,26,0,6,12,0,0,0,20,26]
+; AVX512-NEXT: # zmm7 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm5, %zmm6, %zmm7
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm2 = <0,6,12,18,24,30,u,u>
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm2
+; AVX512-NEXT: movb $56, %dil
+; AVX512-NEXT: kmovd %edi, %k2
+; AVX512-NEXT: vmovdqa64 %zmm7, %zmm2 {%k2}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm7 = [0,0,0,2,8,14,20,26,0,0,0,2,8,14,20,26]
+; AVX512-NEXT: # zmm7 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm0, %zmm1, %zmm7
+; AVX512-NEXT: movw $-2048, %di # imm = 0xF800
+; AVX512-NEXT: kmovd %edi, %k1
+; AVX512-NEXT: vmovdqa32 %zmm7, %zmm2 {%k1}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [1,7,13,0,0,0,21,27,1,7,13,0,0,0,21,27]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm5, %zmm6, %zmm8
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm7 = <1,7,13,19,25,31,u,u>
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm7
+; AVX512-NEXT: vmovdqa64 %zmm8, %zmm7 {%k2}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [0,0,0,3,9,15,21,27,0,0,0,3,9,15,21,27]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm0, %zmm1, %zmm8
+; AVX512-NEXT: vmovdqa32 %zmm8, %zmm7 {%k1}
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm8 = <2,8,14,20,26,u,u,u>
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm8
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm9 = [18,24,30,0,0,0,6,12,18,24,30,0,0,0,6,12]
+; AVX512-NEXT: # zmm9 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm6, %zmm5, %zmm9
+; AVX512-NEXT: movw $31, %di
+; AVX512-NEXT: kmovd %edi, %k2
+; AVX512-NEXT: vmovdqa32 %zmm8, %zmm9 {%k2}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [0,0,0,20,26,0,6,12,0,0,0,20,26,0,6,12]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm1, %zmm0, %zmm8
+; AVX512-NEXT: vmovdqa32 %zmm8, %zmm9 {%k1}
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm8 = <3,9,15,21,27,u,u,u>
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm8
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm10 = [19,25,31,0,0,1,7,13,19,25,31,0,0,1,7,13]
+; AVX512-NEXT: # zmm10 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm6, %zmm5, %zmm10
+; AVX512-NEXT: vmovdqa32 %zmm8, %zmm10 {%k2}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [0,0,0,21,27,1,7,13,0,0,0,21,27,1,7,13]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm1, %zmm0, %zmm8
+; AVX512-NEXT: vmovdqa32 %zmm8, %zmm10 {%k1}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [20,26,0,0,0,2,8,14,20,26,0,0,0,2,8,14]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm6, %zmm5, %zmm8
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm11 = <20,26,0,6,12,u,u,u>
+; AVX512-NEXT: vpermi2d %zmm3, %zmm4, %zmm11
+; AVX512-NEXT: movw $992, %di # imm = 0x3E0
+; AVX512-NEXT: kmovd %edi, %k1
+; AVX512-NEXT: vmovdqa32 %zmm8, %zmm11 {%k1}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [0,0,0,6,12,18,24,30,0,0,0,6,12,18,24,30]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm0, %zmm1, %zmm8
+; AVX512-NEXT: movb $-32, %dil
+; AVX512-NEXT: kmovd %edi, %k2
+; AVX512-NEXT: vmovdqa64 %zmm8, %zmm11 {%k2}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [21,27,0,0,0,3,9,15,21,27,0,0,0,3,9,15]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm6, %zmm5, %zmm8
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm5 = <21,27,1,7,13,u,u,u>
+; AVX512-NEXT: vpermi2d %zmm3, %zmm4, %zmm5
+; AVX512-NEXT: vmovdqa32 %zmm8, %zmm5 {%k1}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm3 = [0,0,1,7,13,19,25,31,0,0,1,7,13,19,25,31]
+; AVX512-NEXT: # zmm3 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm0, %zmm1, %zmm3
+; AVX512-NEXT: vmovdqa64 %zmm3, %zmm5 {%k2}
+; AVX512-NEXT: vmovdqu64 %zmm2, (%rsi)
+; AVX512-NEXT: vmovdqu64 %zmm7, (%rdx)
+; AVX512-NEXT: vmovdqu64 %zmm9, (%rcx)
+; AVX512-NEXT: vmovdqu64 %zmm10, (%r8)
+; AVX512-NEXT: vmovdqu64 %zmm11, (%r9)
+; AVX512-NEXT: vmovdqu64 %zmm5, (%rax)
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %wide.vec = load <96 x i32>, <96 x i32>* %in.vec, align 32
+
+ %strided.vec0 = shufflevector <96 x i32> %wide.vec, <96 x i32> poison, <16 x i32> <i32 0, i32 6, i32 12, i32 18, i32 24, i32 30, i32 36, i32 42, i32 48, i32 54, i32 60, i32 66, i32 72, i32 78, i32 84, i32 90>
+ %strided.vec1 = shufflevector <96 x i32> %wide.vec, <96 x i32> poison, <16 x i32> <i32 1, i32 7, i32 13, i32 19, i32 25, i32 31, i32 37, i32 43, i32 49, i32 55, i32 61, i32 67, i32 73, i32 79, i32 85, i32 91>
+ %strided.vec2 = shufflevector <96 x i32> %wide.vec, <96 x i32> poison, <16 x i32> <i32 2, i32 8, i32 14, i32 20, i32 26, i32 32, i32 38, i32 44, i32 50, i32 56, i32 62, i32 68, i32 74, i32 80, i32 86, i32 92>
+ %strided.vec3 = shufflevector <96 x i32> %wide.vec, <96 x i32> poison, <16 x i32> <i32 3, i32 9, i32 15, i32 21, i32 27, i32 33, i32 39, i32 45, i32 51, i32 57, i32 63, i32 69, i32 75, i32 81, i32 87, i32 93>
+ %strided.vec4 = shufflevector <96 x i32> %wide.vec, <96 x i32> poison, <16 x i32> <i32 4, i32 10, i32 16, i32 22, i32 28, i32 34, i32 40, i32 46, i32 52, i32 58, i32 64, i32 70, i32 76, i32 82, i32 88, i32 94>
+ %strided.vec5 = shufflevector <96 x i32> %wide.vec, <96 x i32> poison, <16 x i32> <i32 5, i32 11, i32 17, i32 23, i32 29, i32 35, i32 41, i32 47, i32 53, i32 59, i32 65, i32 71, i32 77, i32 83, i32 89, i32 95>
+
+ store <16 x i32> %strided.vec0, <16 x i32>* %out.vec0, align 32
+ store <16 x i32> %strided.vec1, <16 x i32>* %out.vec1, align 32
+ store <16 x i32> %strided.vec2, <16 x i32>* %out.vec2, align 32
+ store <16 x i32> %strided.vec3, <16 x i32>* %out.vec3, align 32
+ store <16 x i32> %strided.vec4, <16 x i32>* %out.vec4, align 32
+ store <16 x i32> %strided.vec5, <16 x i32>* %out.vec5, align 32
+
+ ret void
+}
--- /dev/null
+; 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(<2 x i32>* %in.vecptr0, <2 x i32>* %in.vecptr1, <2 x i32>* %in.vecptr2, <2 x i32>* %in.vecptr3, <2 x i32>* %in.vecptr4, <2 x i32>* %in.vecptr5, <12 x i32>* %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: movlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT: movsd {{.*#+}} xmm2 = mem[0],zero
+; SSE-NEXT: movlhps {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; SSE-NEXT: movsd {{.*#+}} xmm2 = mem[0],zero
+; SSE-NEXT: movsd {{.*#+}} xmm3 = mem[0],zero
+; SSE-NEXT: movlhps {{.*#+}} xmm2 = xmm2[0],xmm3[0]
+; SSE-NEXT: movaps %xmm2, %xmm3
+; SSE-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,2],xmm0[1,3]
+; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,3],xmm2[1,3]
+; SSE-NEXT: movaps %xmm1, 32(%rax)
+; SSE-NEXT: movaps %xmm3, 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: vmovsd {{.*#+}} xmm2 = mem[0],zero
+; AVX1-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
+; AVX1-NEXT: vmovlhps {{.*#+}} xmm2 = xmm2[0],xmm3[0]
+; AVX1-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
+; AVX1-NEXT: vmovsd {{.*#+}} xmm4 = mem[0],zero
+; AVX1-NEXT: vmovlhps {{.*#+}} xmm3 = xmm3[0],xmm4[0]
+; AVX1-NEXT: vmovlhps {{.*#+}} xmm4 = xmm0[0],xmm1[0]
+; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm2, %ymm4
+; AVX1-NEXT: vpermilps {{.*#+}} ymm4 = ymm4[u,u,0,2,u,u,5,7]
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm4[2,3,4,5,6,7]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = xmm3[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 = xmm2[1,3],xmm3[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 = xmm0[0],xmm1[0]
+; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
+; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
+; AVX2-SLOW-NEXT: vmovlhps {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
+; AVX2-SLOW-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
+; AVX2-SLOW-NEXT: vmovlhps {{.*#+}} xmm2 = xmm2[0],xmm3[0]
+; AVX2-SLOW-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; 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 = xmm0[0],xmm1[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 = xmm1[0],xmm2[0]
+; AVX2-FAST-ALL-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
+; AVX2-FAST-ALL-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
+; AVX2-FAST-ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-FAST-ALL-NEXT: vmovlhps {{.*#+}} xmm1 = xmm2[0],xmm3[0]
+; AVX2-FAST-ALL-NEXT: vshufps {{.*#+}} ymm1 = ymm0[1,3],ymm1[1,3],ymm0[5,7],ymm1[5,7]
+; AVX2-FAST-ALL-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[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 {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; AVX2-FAST-ALL-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm2
+; AVX2-FAST-ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm2[4,5],ymm0[6,7]
+; AVX2-FAST-ALL-NEXT: vmovaps %ymm0, (%rax)
+; AVX2-FAST-ALL-NEXT: vmovaps %xmm1, 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 = xmm0[0],xmm1[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 = xmm1[0],xmm2[0]
+; AVX2-FAST-PERLANE-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
+; AVX2-FAST-PERLANE-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
+; AVX2-FAST-PERLANE-NEXT: vmovlhps {{.*#+}} xmm2 = xmm2[0],xmm3[0]
+; AVX2-FAST-PERLANE-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; 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 = xmm0[0],xmm1[0]
+; AVX512-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
+; AVX512-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
+; AVX512-NEXT: vmovlhps {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX512-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero
+; AVX512-NEXT: vmovsd {{.*#+}} xmm3 = mem[0],zero
+; AVX512-NEXT: vmovlhps {{.*#+}} xmm2 = xmm2[0],xmm3[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>, <2 x i32>* %in.vecptr0, align 32
+ %in.vec1 = load <2 x i32>, <2 x i32>* %in.vecptr1, align 32
+ %in.vec2 = load <2 x i32>, <2 x i32>* %in.vecptr2, align 32
+ %in.vec3 = load <2 x i32>, <2 x i32>* %in.vecptr3, align 32
+ %in.vec4 = load <2 x i32>, <2 x i32>* %in.vecptr4, align 32
+ %in.vec5 = load <2 x i32>, <2 x i32>* %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, <12 x i32>* %out.vec, align 32
+
+ ret void
+}
+
+define void @store_i32_stride6_vf4(<4 x i32>* %in.vecptr0, <4 x i32>* %in.vecptr1, <4 x i32>* %in.vecptr2, <4 x i32>* %in.vecptr3, <4 x i32>* %in.vecptr4, <4 x i32>* %in.vecptr5, <24 x i32>* %out.vec) nounwind {
+; SSE-LABEL: store_i32_stride6_vf4:
+; SSE: # %bb.0:
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: movapd (%rdi), %xmm0
+; SSE-NEXT: movapd (%rsi), %xmm9
+; SSE-NEXT: movapd (%rdx), %xmm2
+; SSE-NEXT: movapd (%rcx), %xmm8
+; SSE-NEXT: movapd (%r8), %xmm1
+; SSE-NEXT: movapd (%r9), %xmm5
+; SSE-NEXT: movapd %xmm2, %xmm6
+; SSE-NEXT: unpcklps {{.*#+}} xmm6 = xmm6[0],xmm8[0],xmm6[1],xmm8[1]
+; SSE-NEXT: movapd %xmm0, %xmm7
+; SSE-NEXT: unpcklps {{.*#+}} xmm7 = xmm7[0],xmm9[0],xmm7[1],xmm9[1]
+; SSE-NEXT: movapd %xmm7, %xmm3
+; SSE-NEXT: unpcklpd {{.*#+}} xmm3 = xmm3[0],xmm6[0]
+; SSE-NEXT: movapd %xmm1, %xmm4
+; SSE-NEXT: unpcklps {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1]
+; SSE-NEXT: unpckhpd {{.*#+}} xmm6 = xmm6[1],xmm4[1]
+; SSE-NEXT: unpckhps {{.*#+}} xmm1 = xmm1[2],xmm5[2],xmm1[3],xmm5[3]
+; SSE-NEXT: unpckhps {{.*#+}} xmm0 = xmm0[2],xmm9[2],xmm0[3],xmm9[3]
+; SSE-NEXT: movapd %xmm0, %xmm5
+; SSE-NEXT: movsd {{.*#+}} xmm5 = xmm1[0],xmm5[1]
+; SSE-NEXT: unpckhps {{.*#+}} xmm2 = xmm2[2],xmm8[2],xmm2[3],xmm8[3]
+; SSE-NEXT: movhlps {{.*#+}} xmm1 = xmm2[1],xmm1[1]
+; SSE-NEXT: unpcklpd {{.*#+}} xmm0 = xmm0[0],xmm2[0]
+; SSE-NEXT: shufpd {{.*#+}} xmm4 = xmm4[0],xmm7[1]
+; SSE-NEXT: movapd %xmm4, 16(%rax)
+; SSE-NEXT: movapd %xmm0, 48(%rax)
+; SSE-NEXT: movaps %xmm1, 80(%rax)
+; SSE-NEXT: movapd %xmm5, 64(%rax)
+; SSE-NEXT: movapd %xmm6, 32(%rax)
+; SSE-NEXT: movapd %xmm3, (%rax)
+; SSE-NEXT: retq
+;
+; AVX1-LABEL: store_i32_stride6_vf4:
+; AVX1: # %bb.0:
+; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX1-NEXT: vmovaps (%rdi), %xmm2
+; AVX1-NEXT: vmovaps (%rsi), %xmm3
+; AVX1-NEXT: vmovaps (%rdx), %xmm0
+; AVX1-NEXT: vmovaps (%rcx), %xmm1
+; AVX1-NEXT: vmovaps (%r8), %xmm4
+; AVX1-NEXT: vmovaps (%r9), %xmm5
+; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm3, %ymm6
+; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm2, %ymm7
+; AVX1-NEXT: vunpcklpd {{.*#+}} ymm6 = ymm7[0],ymm6[0],ymm7[2],ymm6[2]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm6 = ymm6[0,2,3,1,4,6,7,5]
+; AVX1-NEXT: vshufps {{.*#+}} xmm7 = xmm1[0,0],xmm0[0,0]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm7 = xmm7[0,1,2,0]
+; AVX1-NEXT: vblendps {{.*#+}} ymm6 = ymm6[0,1],ymm7[2,3],ymm6[4,5,6,7]
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm7 = xmm4[0],xmm5[0],xmm4[1],xmm5[1]
+; AVX1-NEXT: vinsertf128 $1, %xmm7, %ymm0, %ymm7
+; AVX1-NEXT: vblendps {{.*#+}} ymm6 = ymm6[0,1,2,3],ymm7[4,5],ymm6[6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm7
+; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm8
+; AVX1-NEXT: vshufps {{.*#+}} ymm7 = ymm8[1,2],ymm7[1,2],ymm8[5,6],ymm7[5,6]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm7 = ymm7[0,2,3,1,4,6,7,5]
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm8 = xmm4[0],xmm5[0],xmm4[1],xmm5[1]
+; AVX1-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1],ymm8[2,3],ymm7[4,5,6,7]
+; AVX1-NEXT: vunpckhps {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
+; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm3
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm7[0,1,2,3],ymm3[4,5],ymm7[6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm5, %ymm7
+; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm4, %ymm4
+; AVX1-NEXT: vunpckhpd {{.*#+}} ymm4 = ymm4[1],ymm7[1],ymm4[3],ymm7[3]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm4 = ymm4[0,2,3,1,4,6,7,5]
+; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm4[0,1],ymm2[2,3],ymm4[4,5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm1[3,0],xmm0[3,0]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[2,0,2,3]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5],ymm2[6,7]
+; AVX1-NEXT: vmovaps %ymm0, 64(%rax)
+; AVX1-NEXT: vmovaps %ymm3, 32(%rax)
+; AVX1-NEXT: vmovaps %ymm6, (%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 (%r8), %xmm3
+; AVX2-NEXT: vmovaps (%r9), %xmm4
+; AVX2-NEXT: vmovaps {{.*#+}} xmm5 = <u,u,0,4>
+; AVX2-NEXT: vinsertf128 $1, (%rcx), %ymm2, %ymm2
+; AVX2-NEXT: vpermps %ymm2, %ymm5, %ymm5
+; AVX2-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm6
+; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm7 = [0,4,1,5,0,4,1,5]
+; AVX2-NEXT: # ymm7 = mem[0,1,0,1]
+; AVX2-NEXT: vpermps %ymm6, %ymm7, %ymm7
+; AVX2-NEXT: vblendps {{.*#+}} ymm5 = ymm7[0,1],ymm5[2,3],ymm7[4,5,6,7]
+; AVX2-NEXT: vinsertf128 $1, %xmm4, %ymm3, %ymm7
+; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm8 = [17179869184,17179869184,17179869184,17179869184]
+; AVX2-NEXT: vpermps %ymm7, %ymm8, %ymm8
+; AVX2-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0,1,2,3],ymm8[4,5],ymm5[6,7]
+; AVX2-NEXT: vunpcklps {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
+; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm4 = [1,5,2,6,1,5,2,6]
+; AVX2-NEXT: # ymm4 = mem[0,1,0,1]
+; AVX2-NEXT: vpermps %ymm2, %ymm4, %ymm4
+; AVX2-NEXT: vblendps {{.*#+}} ymm3 = ymm4[0,1],ymm3[2,3],ymm4[4,5,6,7]
+; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm4 = [25769803778,25769803778,25769803778,25769803778]
+; AVX2-NEXT: vpermps %ymm6, %ymm4, %ymm4
+; AVX2-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
+; AVX2-NEXT: vunpckhps {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2,6,3,7,2,6,3,7]
+; AVX2-NEXT: # ymm1 = mem[0,1,0,1]
+; AVX2-NEXT: vpermps %ymm7, %ymm1, %ymm1
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5,6,7]
+; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [30064771075,30064771075,30064771075,30064771075]
+; AVX2-NEXT: vpermps %ymm2, %ymm1, %ymm1
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
+; AVX2-NEXT: vmovaps %ymm0, 64(%rax)
+; AVX2-NEXT: vmovaps %ymm3, 32(%rax)
+; AVX2-NEXT: vmovaps %ymm5, (%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 = [2,6,19,23,27,31,3,7]
+; AVX512-NEXT: vpermi2d %zmm0, %zmm1, %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>, <4 x i32>* %in.vecptr0, align 32
+ %in.vec1 = load <4 x i32>, <4 x i32>* %in.vecptr1, align 32
+ %in.vec2 = load <4 x i32>, <4 x i32>* %in.vecptr2, align 32
+ %in.vec3 = load <4 x i32>, <4 x i32>* %in.vecptr3, align 32
+ %in.vec4 = load <4 x i32>, <4 x i32>* %in.vecptr4, align 32
+ %in.vec5 = load <4 x i32>, <4 x i32>* %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, <24 x i32>* %out.vec, align 32
+
+ ret void
+}
+
+define void @store_i32_stride6_vf8(<8 x i32>* %in.vecptr0, <8 x i32>* %in.vecptr1, <8 x i32>* %in.vecptr2, <8 x i32>* %in.vecptr3, <8 x i32>* %in.vecptr4, <8 x i32>* %in.vecptr5, <48 x i32>* %out.vec) nounwind {
+; SSE-LABEL: store_i32_stride6_vf8:
+; SSE: # %bb.0:
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: movaps (%rdi), %xmm1
+; SSE-NEXT: movaps 16(%rdi), %xmm6
+; SSE-NEXT: movaps (%rsi), %xmm8
+; SSE-NEXT: movaps 16(%rsi), %xmm12
+; SSE-NEXT: movaps (%rdx), %xmm14
+; SSE-NEXT: movaps 16(%rdx), %xmm5
+; SSE-NEXT: movaps (%rcx), %xmm9
+; SSE-NEXT: movaps 16(%rcx), %xmm13
+; SSE-NEXT: movaps (%r8), %xmm7
+; SSE-NEXT: movaps 16(%r8), %xmm3
+; SSE-NEXT: movaps (%r9), %xmm11
+; SSE-NEXT: movaps 16(%r9), %xmm15
+; SSE-NEXT: movaps %xmm3, %xmm2
+; SSE-NEXT: unpckhps {{.*#+}} xmm2 = xmm2[2],xmm15[2],xmm2[3],xmm15[3]
+; SSE-NEXT: movaps %xmm5, %xmm4
+; SSE-NEXT: unpckhps {{.*#+}} xmm4 = xmm4[2],xmm13[2],xmm4[3],xmm13[3]
+; SSE-NEXT: movaps %xmm4, %xmm10
+; SSE-NEXT: unpckhpd {{.*#+}} xmm10 = xmm10[1],xmm2[1]
+; SSE-NEXT: movaps %xmm6, %xmm0
+; SSE-NEXT: unpckhps {{.*#+}} xmm0 = xmm0[2],xmm12[2],xmm0[3],xmm12[3]
+; SSE-NEXT: shufps {{.*#+}} xmm2 = xmm2[0,1],xmm0[2,3]
+; SSE-NEXT: movlhps {{.*#+}} xmm0 = xmm0[0],xmm4[0]
+; SSE-NEXT: unpcklps {{.*#+}} xmm3 = xmm3[0],xmm15[0],xmm3[1],xmm15[1]
+; SSE-NEXT: unpcklps {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1]
+; SSE-NEXT: movaps %xmm5, %xmm13
+; SSE-NEXT: unpckhpd {{.*#+}} xmm13 = xmm13[1],xmm3[1]
+; SSE-NEXT: unpcklps {{.*#+}} xmm6 = xmm6[0],xmm12[0],xmm6[1],xmm12[1]
+; SSE-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,1],xmm6[2,3]
+; SSE-NEXT: movlhps {{.*#+}} xmm6 = xmm6[0],xmm5[0]
+; SSE-NEXT: movaps %xmm7, %xmm15
+; SSE-NEXT: unpckhps {{.*#+}} xmm15 = xmm15[2],xmm11[2],xmm15[3],xmm11[3]
+; SSE-NEXT: movaps %xmm14, %xmm5
+; SSE-NEXT: unpckhps {{.*#+}} xmm5 = xmm5[2],xmm9[2],xmm5[3],xmm9[3]
+; SSE-NEXT: movaps %xmm5, %xmm12
+; SSE-NEXT: unpckhpd {{.*#+}} xmm12 = xmm12[1],xmm15[1]
+; SSE-NEXT: movaps %xmm1, %xmm4
+; SSE-NEXT: unpckhps {{.*#+}} xmm4 = xmm4[2],xmm8[2],xmm4[3],xmm8[3]
+; SSE-NEXT: shufps {{.*#+}} xmm15 = xmm15[0,1],xmm4[2,3]
+; SSE-NEXT: movlhps {{.*#+}} xmm4 = xmm4[0],xmm5[0]
+; SSE-NEXT: unpcklps {{.*#+}} xmm7 = xmm7[0],xmm11[0],xmm7[1],xmm11[1]
+; SSE-NEXT: unpcklps {{.*#+}} xmm14 = xmm14[0],xmm9[0],xmm14[1],xmm9[1]
+; SSE-NEXT: movaps %xmm14, %xmm5
+; SSE-NEXT: unpckhpd {{.*#+}} xmm5 = xmm5[1],xmm7[1]
+; SSE-NEXT: unpcklps {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1]
+; SSE-NEXT: shufps {{.*#+}} xmm7 = xmm7[0,1],xmm1[2,3]
+; SSE-NEXT: movlhps {{.*#+}} xmm1 = xmm1[0],xmm14[0]
+; SSE-NEXT: movaps %xmm1, (%rax)
+; SSE-NEXT: movaps %xmm7, 16(%rax)
+; SSE-NEXT: movaps %xmm5, 32(%rax)
+; SSE-NEXT: movaps %xmm4, 48(%rax)
+; SSE-NEXT: movaps %xmm15, 64(%rax)
+; SSE-NEXT: movaps %xmm12, 80(%rax)
+; SSE-NEXT: movaps %xmm6, 96(%rax)
+; SSE-NEXT: movaps %xmm3, 112(%rax)
+; SSE-NEXT: movaps %xmm13, 128(%rax)
+; SSE-NEXT: movaps %xmm0, 144(%rax)
+; SSE-NEXT: movaps %xmm2, 160(%rax)
+; SSE-NEXT: movaps %xmm10, 176(%rax)
+; SSE-NEXT: retq
+;
+; AVX1-LABEL: store_i32_stride6_vf8:
+; AVX1: # %bb.0:
+; AVX1-NEXT: vmovaps (%rdi), %ymm9
+; AVX1-NEXT: vmovaps (%rsi), %ymm11
+; AVX1-NEXT: vmovaps (%rdx), %ymm12
+; AVX1-NEXT: vmovaps (%rcx), %ymm1
+; AVX1-NEXT: vunpcklps {{.*#+}} ymm4 = ymm9[0],ymm11[0],ymm9[1],ymm11[1],ymm9[4],ymm11[4],ymm9[5],ymm11[5]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm4 = ymm4[2,3,2,3]
+; AVX1-NEXT: vunpcklpd {{.*#+}} ymm5 = ymm1[0],ymm12[0],ymm1[2],ymm12[2]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm5 = ymm5[0,1,2,0,4,5,6,4]
+; AVX1-NEXT: vextractf128 $1, %ymm5, %xmm5
+; AVX1-NEXT: vblendps {{.*#+}} ymm8 = ymm4[0,1],ymm5[2,3],ymm4[4,5,6,7]
+; AVX1-NEXT: vmovaps (%rcx), %xmm6
+; AVX1-NEXT: vmovaps (%rdx), %xmm7
+; AVX1-NEXT: vshufps {{.*#+}} xmm5 = xmm7[1,2],xmm6[1,2]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm5 = xmm5[0,2,1,3]
+; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm5, %ymm5
+; AVX1-NEXT: vmovaps (%r9), %xmm4
+; AVX1-NEXT: vmovaps (%r8), %xmm2
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm10 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
+; AVX1-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0,1],ymm10[2,3],ymm5[4,5,6,7]
+; AVX1-NEXT: vmovaps (%rsi), %xmm3
+; AVX1-NEXT: vmovaps (%rdi), %xmm0
+; AVX1-NEXT: vunpckhps {{.*#+}} xmm13 = xmm0[2],xmm3[2],xmm0[3],xmm3[3]
+; AVX1-NEXT: vinsertf128 $1, %xmm13, %ymm0, %ymm14
+; AVX1-NEXT: vblendps {{.*#+}} ymm14 = ymm5[0,1,2,3],ymm14[4,5],ymm5[6,7]
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; AVX1-NEXT: vbroadcastss (%rcx), %xmm3
+; AVX1-NEXT: vbroadcastss (%rdx), %xmm5
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm3 = xmm5[0],xmm3[0],xmm5[1],xmm3[1]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm3[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm10, %ymm0, %ymm3
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm3[4,5],ymm0[6,7]
+; AVX1-NEXT: vunpckhps {{.*#+}} xmm2 = xmm2[2],xmm4[2],xmm2[3],xmm4[3]
+; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm2, %ymm2
+; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm13[2,3],ymm2[4,5,6,7]
+; AVX1-NEXT: vmovaps (%r8), %ymm3
+; AVX1-NEXT: vunpckhps {{.*#+}} xmm4 = xmm7[2],xmm6[2],xmm7[3],xmm6[3]
+; AVX1-NEXT: vmovaps (%r9), %ymm5
+; AVX1-NEXT: vpermilps {{.*#+}} xmm4 = xmm4[2,3,2,3]
+; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm0, %ymm4
+; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm4[4,5],ymm2[6,7]
+; AVX1-NEXT: vunpcklps {{.*#+}} ymm4 = ymm3[0],ymm5[0],ymm3[1],ymm5[1],ymm3[4],ymm5[4],ymm3[5],ymm5[5]
+; AVX1-NEXT: vunpckhps {{.*#+}} ymm3 = ymm3[2],ymm5[2],ymm3[3],ymm5[3],ymm3[6],ymm5[6],ymm3[7],ymm5[7]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm3 = ymm3[2,3,2,3]
+; AVX1-NEXT: vunpckhps {{.*#+}} ymm5 = ymm9[2],ymm11[2],ymm9[3],ymm11[3],ymm9[6],ymm11[6],ymm9[7],ymm11[7]
+; AVX1-NEXT: vextractf128 $1, %ymm5, %xmm6
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1],ymm6[2,3],ymm3[4,5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm6 = ymm1[3,0],ymm12[3,0],ymm1[7,4],ymm12[7,4]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm6 = ymm6[2,0,2,3,6,4,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm6[4,5],ymm3[6,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm1 = ymm12[1,2],ymm1[1,2],ymm12[5,6],ymm1[5,6]
+; AVX1-NEXT: vblendps {{.*#+}} ymm6 = ymm8[0,1,2,3],ymm4[4,5],ymm8[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm4, %xmm4
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm1 = ymm1[2,3,2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,2,1,3,4,6,5,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1],ymm4[2,3],ymm1[4,5,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm5[4,5],ymm1[6,7]
+; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX1-NEXT: vmovaps %ymm1, 128(%rax)
+; AVX1-NEXT: vmovaps %ymm3, 160(%rax)
+; AVX1-NEXT: vmovaps %ymm2, 64(%rax)
+; AVX1-NEXT: vmovaps %ymm0, (%rax)
+; AVX1-NEXT: vmovaps %ymm14, 32(%rax)
+; AVX1-NEXT: vmovaps %ymm6, 96(%rax)
+; AVX1-NEXT: vzeroupper
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: store_i32_stride6_vf8:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vmovdqa (%rdi), %ymm10
+; AVX2-NEXT: vmovdqa (%rsi), %ymm11
+; AVX2-NEXT: vmovdqa (%rdx), %ymm12
+; AVX2-NEXT: vmovaps (%r9), %xmm3
+; AVX2-NEXT: vmovaps (%r8), %xmm4
+; AVX2-NEXT: vunpcklps {{.*#+}} xmm9 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
+; AVX2-NEXT: vmovaps (%rcx), %xmm0
+; AVX2-NEXT: vpermilps {{.*#+}} xmm5 = xmm0[0,1,2,2]
+; AVX2-NEXT: vmovaps (%rdx), %xmm1
+; AVX2-NEXT: vpermilps {{.*#+}} xmm6 = xmm1[1,1,2,3]
+; AVX2-NEXT: vblendps {{.*#+}} xmm5 = xmm6[0],xmm5[1],xmm6[2],xmm5[3]
+; AVX2-NEXT: vpermpd {{.*#+}} ymm5 = ymm5[0,1,2,1]
+; AVX2-NEXT: vblendps {{.*#+}} ymm5 = ymm5[0,1],ymm9[2,3],ymm5[4,5,6,7]
+; AVX2-NEXT: vmovaps (%rsi), %xmm7
+; AVX2-NEXT: vmovaps (%rdi), %xmm2
+; AVX2-NEXT: vunpckhps {{.*#+}} xmm13 = xmm2[2],xmm7[2],xmm2[3],xmm7[3]
+; AVX2-NEXT: vinsertf128 $1, %xmm13, %ymm0, %ymm6
+; AVX2-NEXT: vblendps {{.*#+}} ymm8 = ymm5[0,1,2,3],ymm6[4,5],ymm5[6,7]
+; AVX2-NEXT: vbroadcastss (%rcx), %xmm5
+; AVX2-NEXT: vbroadcastss (%rdx), %xmm6
+; AVX2-NEXT: vunpcklps {{.*#+}} xmm14 = xmm6[0],xmm5[0],xmm6[1],xmm5[1]
+; AVX2-NEXT: vmovdqa (%rcx), %ymm6
+; AVX2-NEXT: vunpcklps {{.*#+}} xmm2 = xmm2[0],xmm7[0],xmm2[1],xmm7[1]
+; AVX2-NEXT: vmovdqa (%r8), %ymm5
+; AVX2-NEXT: vpermpd {{.*#+}} ymm2 = ymm2[0,1,2,1]
+; AVX2-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm14[2,3],ymm2[4,5,6,7]
+; AVX2-NEXT: vmovdqa (%r9), %ymm7
+; AVX2-NEXT: vinsertf128 $1, %xmm9, %ymm0, %ymm9
+; AVX2-NEXT: vblendps {{.*#+}} ymm9 = ymm2[0,1,2,3],ymm9[4,5],ymm2[6,7]
+; AVX2-NEXT: vunpckhps {{.*#+}} xmm2 = xmm4[2],xmm3[2],xmm4[3],xmm3[3]
+; AVX2-NEXT: vpermpd {{.*#+}} ymm2 = ymm2[0,1,2,1]
+; AVX2-NEXT: vblendps {{.*#+}} ymm2 = ymm2[0,1],ymm13[2,3],ymm2[4,5,6,7]
+; AVX2-NEXT: vunpckhps {{.*#+}} xmm0 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; AVX2-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; AVX2-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5],ymm2[6,7]
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm1 = ymm10[2],ymm11[2],ymm10[3],ymm11[3],ymm10[6],ymm11[6],ymm10[7],ymm11[7]
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm3 = ymm5[2],ymm7[2],ymm5[3],ymm7[3],ymm5[6],ymm7[6],ymm5[7],ymm7[7]
+; AVX2-NEXT: vpermq {{.*#+}} ymm3 = ymm3[2,1,2,3]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5,6,7]
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm3 = ymm12[2],ymm6[2],ymm12[3],ymm6[3],ymm12[6],ymm6[6],ymm12[7],ymm6[7]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm3 = ymm3[2,3,2,3,6,7,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5],ymm2[6,7]
+; AVX2-NEXT: vpmovzxdq {{.*#+}} xmm3 = mem[0],zero,mem[1],zero
+; AVX2-NEXT: vpbroadcastd 20(%r9), %ymm4
+; AVX2-NEXT: vpblendd {{.*#+}} xmm3 = xmm3[0,1,2],xmm4[3]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm4 = ymm6[0,1,2,2,4,5,6,6]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm13 = ymm12[1,1,2,3,5,5,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm4 = ymm13[0],ymm4[1],ymm13[2],ymm4[3],ymm13[4],ymm4[5],ymm13[6],ymm4[7]
+; AVX2-NEXT: vpermq {{.*#+}} ymm4 = ymm4[2,1,2,3]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm3 = ymm4[0,1],ymm3[2,3],ymm4[4,5,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm3[0,1,2,3],ymm1[4,5],ymm3[6,7]
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm3 = ymm12[0],ymm6[0],ymm12[1],ymm6[1],ymm12[4],ymm6[4],ymm12[5],ymm6[5]
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm4 = ymm10[0],ymm11[0],ymm10[1],ymm11[1],ymm10[4],ymm11[4],ymm10[5],ymm11[5]
+; AVX2-NEXT: vpermq {{.*#+}} ymm3 = ymm3[2,2,2,2]
+; AVX2-NEXT: vpermq {{.*#+}} ymm4 = ymm4[2,1,2,3]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm3 = ymm4[0,1],ymm3[2,3],ymm4[4,5,6,7]
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm4 = ymm5[0],ymm7[0],ymm5[1],ymm7[1],ymm5[4],ymm7[4],ymm5[5],ymm7[5]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: vmovdqa %ymm3, 96(%rax)
+; AVX2-NEXT: vmovdqa %ymm1, 128(%rax)
+; AVX2-NEXT: vmovdqa %ymm2, 160(%rax)
+; AVX2-NEXT: vmovaps %ymm0, 64(%rax)
+; AVX2-NEXT: vmovaps %ymm9, (%rax)
+; AVX2-NEXT: vmovaps %ymm8, 32(%rax)
+; AVX2-NEXT: vzeroupper
+; AVX2-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, (%rcx), %zmm1, %zmm1
+; AVX512-NEXT: vinserti64x4 $1, (%rsi), %zmm0, %zmm0
+; 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: vinserti64x4 $1, (%r9), %zmm2, %zmm2
+; 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 = <2,10,19,27,u,u,3,11,20,28,u,u,4,12,21,29>
+; AVX512-NEXT: vpermi2d %zmm0, %zmm2, %zmm3
+; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm5 = [0,1,2,3,19,27,6,7,8,9,20,28,12,13,14,15]
+; AVX512-NEXT: vpermi2d %zmm1, %zmm3, %zmm5
+; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm3 = <5,13,21,29,u,u,6,14,22,30,u,u,7,15,23,31>
+; AVX512-NEXT: vpermi2d %zmm2, %zmm1, %zmm3
+; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm1 = [0,1,2,3,22,30,6,7,8,9,23,31,12,13,14,15]
+; AVX512-NEXT: vpermi2d %zmm0, %zmm3, %zmm1
+; AVX512-NEXT: vmovdqu64 %zmm1, 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>, <8 x i32>* %in.vecptr0, align 32
+ %in.vec1 = load <8 x i32>, <8 x i32>* %in.vecptr1, align 32
+ %in.vec2 = load <8 x i32>, <8 x i32>* %in.vecptr2, align 32
+ %in.vec3 = load <8 x i32>, <8 x i32>* %in.vecptr3, align 32
+ %in.vec4 = load <8 x i32>, <8 x i32>* %in.vecptr4, align 32
+ %in.vec5 = load <8 x i32>, <8 x i32>* %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, <48 x i32>* %out.vec, align 32
+
+ ret void
+}
+
+define void @store_i32_stride6_vf16(<16 x i32>* %in.vecptr0, <16 x i32>* %in.vecptr1, <16 x i32>* %in.vecptr2, <16 x i32>* %in.vecptr3, <16 x i32>* %in.vecptr4, <16 x i32>* %in.vecptr5, <96 x i32>* %out.vec) nounwind {
+; SSE-LABEL: store_i32_stride6_vf16:
+; SSE: # %bb.0:
+; SSE-NEXT: subq $72, %rsp
+; SSE-NEXT: movapd (%rdi), %xmm12
+; SSE-NEXT: movapd 16(%rdi), %xmm10
+; SSE-NEXT: movapd (%rsi), %xmm13
+; SSE-NEXT: movapd 16(%rsi), %xmm8
+; SSE-NEXT: movapd (%rdx), %xmm4
+; SSE-NEXT: movapd 16(%rdx), %xmm2
+; SSE-NEXT: movapd (%rcx), %xmm5
+; SSE-NEXT: movapd 16(%rcx), %xmm9
+; SSE-NEXT: movapd (%r8), %xmm7
+; SSE-NEXT: movapd 16(%r8), %xmm14
+; SSE-NEXT: movapd (%r9), %xmm0
+; SSE-NEXT: movapd 16(%r9), %xmm11
+; SSE-NEXT: movapd %xmm4, %xmm1
+; SSE-NEXT: unpcklps {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
+; SSE-NEXT: movapd %xmm12, %xmm3
+; SSE-NEXT: unpcklps {{.*#+}} xmm3 = xmm3[0],xmm13[0],xmm3[1],xmm13[1]
+; SSE-NEXT: movapd %xmm3, %xmm6
+; SSE-NEXT: unpcklpd {{.*#+}} xmm6 = xmm6[0],xmm1[0]
+; SSE-NEXT: movapd %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movapd %xmm7, %xmm6
+; SSE-NEXT: unpcklps {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1]
+; SSE-NEXT: movsd {{.*#+}} xmm3 = xmm6[0],xmm3[1]
+; SSE-NEXT: movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movhlps {{.*#+}} xmm6 = xmm1[1],xmm6[1]
+; SSE-NEXT: movaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: unpckhps {{.*#+}} xmm4 = xmm4[2],xmm5[2],xmm4[3],xmm5[3]
+; SSE-NEXT: unpckhps {{.*#+}} xmm12 = xmm12[2],xmm13[2],xmm12[3],xmm13[3]
+; SSE-NEXT: movapd %xmm12, %xmm1
+; SSE-NEXT: unpcklpd {{.*#+}} xmm1 = xmm1[0],xmm4[0]
+; SSE-NEXT: movapd %xmm1, (%rsp) # 16-byte Spill
+; SSE-NEXT: unpckhps {{.*#+}} xmm7 = xmm7[2],xmm0[2],xmm7[3],xmm0[3]
+; SSE-NEXT: movsd {{.*#+}} xmm12 = xmm7[0],xmm12[1]
+; SSE-NEXT: movapd %xmm12, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movhlps {{.*#+}} xmm7 = xmm4[1],xmm7[1]
+; SSE-NEXT: movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movapd %xmm2, %xmm0
+; SSE-NEXT: unpcklps {{.*#+}} xmm0 = xmm0[0],xmm9[0],xmm0[1],xmm9[1]
+; SSE-NEXT: movapd %xmm10, %xmm3
+; SSE-NEXT: unpcklps {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1]
+; SSE-NEXT: movapd %xmm3, %xmm1
+; SSE-NEXT: unpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
+; SSE-NEXT: movapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movapd %xmm14, %xmm1
+; SSE-NEXT: movapd %xmm14, %xmm13
+; SSE-NEXT: unpcklps {{.*#+}} xmm13 = xmm13[0],xmm11[0],xmm13[1],xmm11[1]
+; SSE-NEXT: movsd {{.*#+}} xmm3 = xmm13[0],xmm3[1]
+; SSE-NEXT: movapd %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movhlps {{.*#+}} xmm13 = xmm0[1],xmm13[1]
+; SSE-NEXT: movapd 32(%rdi), %xmm14
+; SSE-NEXT: unpckhps {{.*#+}} xmm2 = xmm2[2],xmm9[2],xmm2[3],xmm9[3]
+; SSE-NEXT: movapd 32(%rdx), %xmm0
+; SSE-NEXT: unpckhps {{.*#+}} xmm10 = xmm10[2],xmm8[2],xmm10[3],xmm8[3]
+; SSE-NEXT: movapd 32(%rcx), %xmm3
+; SSE-NEXT: unpckhps {{.*#+}} xmm1 = xmm1[2],xmm11[2],xmm1[3],xmm11[3]
+; SSE-NEXT: movapd %xmm10, %xmm4
+; SSE-NEXT: unpcklpd {{.*#+}} xmm4 = xmm4[0],xmm2[0]
+; SSE-NEXT: movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movsd {{.*#+}} xmm10 = xmm1[0],xmm10[1]
+; SSE-NEXT: movapd %xmm10, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movhlps {{.*#+}} xmm1 = xmm2[1],xmm1[1]
+; SSE-NEXT: movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movapd %xmm0, %xmm2
+; SSE-NEXT: unpcklps {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; SSE-NEXT: movapd 32(%rsi), %xmm1
+; SSE-NEXT: movapd %xmm14, %xmm8
+; SSE-NEXT: unpcklps {{.*#+}} xmm8 = xmm8[0],xmm1[0],xmm8[1],xmm1[1]
+; SSE-NEXT: movapd %xmm8, %xmm4
+; SSE-NEXT: unpcklpd {{.*#+}} xmm4 = xmm4[0],xmm2[0]
+; SSE-NEXT: movapd %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE-NEXT: movapd 32(%r8), %xmm4
+; SSE-NEXT: movapd 32(%r9), %xmm6
+; SSE-NEXT: movapd %xmm4, %xmm5
+; SSE-NEXT: unpcklps {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1]
+; SSE-NEXT: movsd {{.*#+}} xmm8 = xmm5[0],xmm8[1]
+; SSE-NEXT: movhlps {{.*#+}} xmm5 = xmm2[1],xmm5[1]
+; SSE-NEXT: unpckhps {{.*#+}} xmm0 = xmm0[2],xmm3[2],xmm0[3],xmm3[3]
+; SSE-NEXT: unpckhps {{.*#+}} xmm14 = xmm14[2],xmm1[2],xmm14[3],xmm1[3]
+; SSE-NEXT: unpckhps {{.*#+}} xmm4 = xmm4[2],xmm6[2],xmm4[3],xmm6[3]
+; SSE-NEXT: movapd %xmm14, %xmm15
+; SSE-NEXT: unpcklpd {{.*#+}} xmm15 = xmm15[0],xmm0[0]
+; SSE-NEXT: movsd {{.*#+}} xmm14 = xmm4[0],xmm14[1]
+; SSE-NEXT: movhlps {{.*#+}} xmm4 = xmm0[1],xmm4[1]
+; SSE-NEXT: movapd 48(%rdx), %xmm3
+; SSE-NEXT: movapd 48(%rcx), %xmm10
+; SSE-NEXT: movapd %xmm3, %xmm6
+; SSE-NEXT: unpcklps {{.*#+}} xmm6 = xmm6[0],xmm10[0],xmm6[1],xmm10[1]
+; SSE-NEXT: movapd 48(%rdi), %xmm2
+; SSE-NEXT: movapd 48(%rsi), %xmm12
+; SSE-NEXT: movapd %xmm2, %xmm7
+; SSE-NEXT: unpcklps {{.*#+}} xmm7 = xmm7[0],xmm12[0],xmm7[1],xmm12[1]
+; SSE-NEXT: movapd %xmm7, %xmm9
+; SSE-NEXT: unpcklpd {{.*#+}} xmm9 = xmm9[0],xmm6[0]
+; SSE-NEXT: movapd 48(%r8), %xmm0
+; SSE-NEXT: movapd 48(%r9), %xmm11
+; SSE-NEXT: movapd %xmm0, %xmm1
+; SSE-NEXT: unpcklps {{.*#+}} xmm1 = xmm1[0],xmm11[0],xmm1[1],xmm11[1]
+; SSE-NEXT: movsd {{.*#+}} xmm7 = xmm1[0],xmm7[1]
+; SSE-NEXT: movhlps {{.*#+}} xmm1 = xmm6[1],xmm1[1]
+; SSE-NEXT: unpckhps {{.*#+}} xmm3 = xmm3[2],xmm10[2],xmm3[3],xmm10[3]
+; SSE-NEXT: unpckhps {{.*#+}} xmm2 = xmm2[2],xmm12[2],xmm2[3],xmm12[3]
+; SSE-NEXT: unpckhps {{.*#+}} xmm0 = xmm0[2],xmm11[2],xmm0[3],xmm11[3]
+; SSE-NEXT: movapd %xmm2, %xmm6
+; SSE-NEXT: unpcklpd {{.*#+}} xmm6 = xmm6[0],xmm3[0]
+; SSE-NEXT: movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
+; SSE-NEXT: movhlps {{.*#+}} xmm0 = xmm3[1],xmm0[1]
+; SSE-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE-NEXT: movaps %xmm0, 368(%rax)
+; SSE-NEXT: movapd %xmm2, 352(%rax)
+; SSE-NEXT: movapd %xmm6, 336(%rax)
+; SSE-NEXT: movaps %xmm1, 320(%rax)
+; SSE-NEXT: movapd %xmm7, 304(%rax)
+; SSE-NEXT: movapd %xmm9, 288(%rax)
+; SSE-NEXT: movaps %xmm4, 272(%rax)
+; SSE-NEXT: movapd %xmm14, 256(%rax)
+; SSE-NEXT: movapd %xmm15, 240(%rax)
+; SSE-NEXT: movaps %xmm5, 224(%rax)
+; SSE-NEXT: movapd %xmm8, 208(%rax)
+; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 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 %xmm13, 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 {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE-NEXT: movaps %xmm0, 64(%rax)
+; SSE-NEXT: movaps (%rsp), %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 $312, %rsp # imm = 0x138
+; AVX1-NEXT: vmovaps (%rdi), %ymm2
+; AVX1-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 32(%rdi), %ymm1
+; AVX1-NEXT: vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 32(%rsi), %ymm0
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 32(%rdx), %ymm12
+; AVX1-NEXT: vmovaps 32(%rcx), %ymm6
+; AVX1-NEXT: vunpcklps {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[4],ymm0[4],ymm1[5],ymm0[5]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX1-NEXT: vunpcklpd {{.*#+}} ymm1 = ymm6[0],ymm12[0],ymm6[2],ymm12[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: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 32(%rcx), %xmm1
+; AVX1-NEXT: vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; AVX1-NEXT: vmovaps 32(%rdx), %xmm0
+; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm0[1,2],xmm1[1,2]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT: vmovaps 32(%r9), %xmm15
+; AVX1-NEXT: vmovaps 32(%r8), %xmm13
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm1 = xmm13[0],xmm15[0],xmm13[1],xmm15[1]
+; AVX1-NEXT: vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vblendps {{.*#+}} ymm7 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vmovaps 32(%rsi), %xmm0
+; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; AVX1-NEXT: vmovaps 32(%rdi), %xmm14
+; AVX1-NEXT: vunpckhps {{.*#+}} xmm9 = xmm14[2],xmm0[2],xmm14[3],xmm0[3]
+; AVX1-NEXT: vinsertf128 $1, %xmm9, %ymm0, %ymm8
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm7[0,1,2,3],ymm8[4,5],ymm7[6,7]
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps (%rsi), %ymm0
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vunpcklps {{.*#+}} ymm7 = ymm2[0],ymm0[0],ymm2[1],ymm0[1],ymm2[4],ymm0[4],ymm2[5],ymm0[5]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm10 = ymm7[2,3,2,3]
+; AVX1-NEXT: vmovaps (%rdx), %ymm7
+; AVX1-NEXT: vmovaps (%rcx), %ymm8
+; AVX1-NEXT: vunpcklpd {{.*#+}} ymm11 = ymm8[0],ymm7[0],ymm8[2],ymm7[2]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm11 = ymm11[0,1,2,0,4,5,6,4]
+; AVX1-NEXT: vextractf128 $1, %ymm11, %xmm11
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm10[0,1],ymm11[2,3],ymm10[4,5,6,7]
+; 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 {{.*#+}} xmm4 = xmm1[1,2],xmm0[1,2]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm4 = xmm4[0,2,1,3]
+; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm4, %ymm4
+; AVX1-NEXT: vmovaps (%r9), %xmm11
+; AVX1-NEXT: vmovaps (%r8), %xmm10
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm5 = xmm10[0],xmm11[0],xmm10[1],xmm11[1]
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm4[0,1],ymm5[2,3],ymm4[4,5,6,7]
+; AVX1-NEXT: vmovaps (%rsi), %xmm2
+; AVX1-NEXT: vmovaps (%rdi), %xmm1
+; AVX1-NEXT: vunpckhps {{.*#+}} xmm0 = xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm4
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
+; AVX1-NEXT: vmovups %ymm3, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm0 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; AVX1-NEXT: vbroadcastss (%rcx), %xmm1
+; AVX1-NEXT: vbroadcastss (%rdx), %xmm3
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm1 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm0, %ymm1
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
+; AVX1-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vmovaps 32(%r8), %ymm0
+; AVX1-NEXT: vmovaps 32(%r9), %ymm1
+; AVX1-NEXT: vunpcklps {{.*#+}} ymm3 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5]
+; AVX1-NEXT: vunpckhps {{.*#+}} ymm0 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7]
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; AVX1-NEXT: vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %ymm1, %ymm4 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm4 = ymm1[2],mem[2],ymm1[3],mem[3],ymm1[6],mem[6],ymm1[7],mem[7]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX1-NEXT: vextractf128 $1, %ymm4, %xmm1
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm1 = ymm6[3,0],ymm12[3,0],ymm6[7,4],ymm12[7,4]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[2,0,2,3,6,4,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm1 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm0 = ymm12[1,2],ymm6[1,2],ymm12[5,6],ymm6[5,6]
+; AVX1-NEXT: vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm3, %ymm2 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm2 = mem[0,1,2,3],ymm3[4,5],mem[6,7]
+; AVX1-NEXT: vmovups %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm3
+; 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],ymm3[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm6 = ymm0[0,1,2,3],ymm4[4,5],ymm0[6,7]
+; AVX1-NEXT: vunpckhps {{.*#+}} xmm0 = xmm13[2],xmm15[2],xmm13[3],xmm15[3]
+; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm9[2,3],ymm0[4,5,6,7]
+; AVX1-NEXT: vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
+; AVX1-NEXT: vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm3 # 16-byte Folded Reload
+; AVX1-NEXT: # xmm3 = xmm2[2],mem[2],xmm2[3],mem[3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm3 = xmm3[2,3,2,3]
+; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm0, %ymm3
+; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm3[4,5],ymm0[6,7]
+; AVX1-NEXT: vunpcklps {{[-0-9]+}}(%r{{[sb]}}p), %xmm14, %xmm3 # 16-byte Folded Reload
+; AVX1-NEXT: # xmm3 = xmm14[0],mem[0],xmm14[1],mem[1]
+; AVX1-NEXT: vbroadcastss 32(%rcx), %xmm4
+; AVX1-NEXT: vbroadcastss 32(%rdx), %xmm5
+; AVX1-NEXT: vunpcklps {{.*#+}} xmm4 = xmm5[0],xmm4[0],xmm5[1],xmm4[1]
+; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm3, %ymm3
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1],ymm4[2,3],ymm3[4,5,6,7]
+; AVX1-NEXT: vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm4 # 16-byte Folded Reload
+; AVX1-NEXT: vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],ymm4[4,5],ymm3[6,7]
+; AVX1-NEXT: vmovaps (%r8), %ymm4
+; AVX1-NEXT: vmovaps (%r9), %ymm5
+; AVX1-NEXT: vunpcklps {{.*#+}} ymm9 = ymm4[0],ymm5[0],ymm4[1],ymm5[1],ymm4[4],ymm5[4],ymm4[5],ymm5[5]
+; AVX1-NEXT: vunpckhps {{.*#+}} ymm4 = ymm4[2],ymm5[2],ymm4[3],ymm5[3],ymm4[6],ymm5[6],ymm4[7],ymm5[7]
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm2 # 32-byte Reload
+; AVX1-NEXT: vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %ymm2, %ymm5 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm5 = ymm2[2],mem[2],ymm2[3],mem[3],ymm2[6],mem[6],ymm2[7],mem[7]
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm4 = ymm4[2,3,2,3]
+; AVX1-NEXT: vextractf128 $1, %ymm5, %xmm12
+; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1],ymm12[2,3],ymm4[4,5,6,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm12 = ymm8[3,0],ymm7[3,0],ymm8[7,4],ymm7[7,4]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm12 = ymm12[2,0,2,3,6,4,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm4 = ymm4[0,1,2,3],ymm12[4,5],ymm4[6,7]
+; AVX1-NEXT: vshufps {{.*#+}} ymm7 = ymm7[1,2],ymm8[1,2],ymm7[5,6],ymm8[5,6]
+; AVX1-NEXT: vblendps $207, {{[-0-9]+}}(%r{{[sb]}}p), %ymm9, %ymm8 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm8 = mem[0,1,2,3],ymm9[4,5],mem[6,7]
+; AVX1-NEXT: vextractf128 $1, %ymm9, %xmm9
+; AVX1-NEXT: vperm2f128 {{.*#+}} ymm7 = ymm7[2,3,2,3]
+; AVX1-NEXT: vpermilps {{.*#+}} ymm7 = ymm7[0,2,1,3,4,6,5,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm7 = ymm7[0,1],ymm9[2,3],ymm7[4,5,6,7]
+; AVX1-NEXT: vblendps {{.*#+}} ymm5 = ymm7[0,1,2,3],ymm5[4,5],ymm7[6,7]
+; AVX1-NEXT: vunpckhps {{.*#+}} xmm7 = xmm10[2],xmm11[2],xmm10[3],xmm11[3]
+; AVX1-NEXT: vinsertf128 $1, %xmm7, %ymm7, %ymm7
+; AVX1-NEXT: vblendps $12, {{[-0-9]+}}(%r{{[sb]}}p), %ymm7, %ymm7 # 32-byte Folded Reload
+; AVX1-NEXT: # ymm7 = ymm7[0,1],mem[2,3],ymm7[4,5,6,7]
+; AVX1-NEXT: vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload
+; AVX1-NEXT: vunpckhps {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 16-byte Folded Reload
+; AVX1-NEXT: # xmm2 = xmm2[2],mem[2],xmm2[3],mem[3]
+; AVX1-NEXT: vpermilps {{.*#+}} xmm2 = xmm2[2,3,2,3]
+; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm2
+; AVX1-NEXT: vblendps {{.*#+}} ymm2 = ymm7[0,1,2,3],ymm2[4,5],ymm7[6,7]
+; AVX1-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX1-NEXT: vmovaps %ymm2, 64(%rax)
+; AVX1-NEXT: vmovaps %ymm5, 128(%rax)
+; AVX1-NEXT: vmovaps %ymm4, 160(%rax)
+; AVX1-NEXT: vmovaps %ymm3, 192(%rax)
+; AVX1-NEXT: vmovaps %ymm0, 256(%rax)
+; AVX1-NEXT: vmovaps %ymm6, 320(%rax)
+; AVX1-NEXT: vmovaps %ymm1, 352(%rax)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm0, (%rax)
+; AVX1-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX1-NEXT: vmovaps %ymm0, 32(%rax)
+; AVX1-NEXT: vmovaps %ymm8, 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 $312, %rsp # imm = 0x138
+; AVX1-NEXT: vzeroupper
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: store_i32_stride6_vf16:
+; AVX2: # %bb.0:
+; AVX2-NEXT: subq $232, %rsp
+; AVX2-NEXT: vmovaps (%r9), %xmm13
+; AVX2-NEXT: vmovaps %xmm13, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; AVX2-NEXT: vmovdqa 32(%r9), %xmm12
+; AVX2-NEXT: vmovaps (%r8), %xmm14
+; AVX2-NEXT: vmovaps %xmm14, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; AVX2-NEXT: vmovdqa 32(%r8), %xmm10
+; AVX2-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm10[0],xmm12[0],xmm10[1],xmm12[1]
+; AVX2-NEXT: vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vmovaps (%rcx), %xmm6
+; AVX2-NEXT: vmovaps %xmm6, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; AVX2-NEXT: vmovdqa 32(%rcx), %xmm8
+; AVX2-NEXT: vpshufd {{.*#+}} xmm0 = xmm8[0,1,2,2]
+; AVX2-NEXT: vmovaps (%rdx), %xmm7
+; AVX2-NEXT: vmovaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; AVX2-NEXT: vmovdqa 32(%rdx), %xmm4
+; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm4[1,1,2,3]
+; AVX2-NEXT: vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2],xmm0[3]
+; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm2[2,3],ymm0[4,5,6,7]
+; AVX2-NEXT: vmovaps (%rsi), %xmm11
+; AVX2-NEXT: vmovdqa 32(%rsi), %xmm5
+; AVX2-NEXT: vmovdqa 32(%rdi), %xmm3
+; AVX2-NEXT: vpunpckhdq {{.*#+}} xmm2 = xmm3[2],xmm5[2],xmm3[3],xmm5[3]
+; AVX2-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm9
+; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm9[4,5],ymm0[6,7]
+; AVX2-NEXT: vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vpermilps {{.*#+}} xmm0 = xmm6[0,1,2,2]
+; AVX2-NEXT: vpermilps {{.*#+}} xmm6 = xmm7[1,1,2,3]
+; AVX2-NEXT: vblendps {{.*#+}} xmm0 = xmm6[0],xmm0[1],xmm6[2],xmm0[3]
+; AVX2-NEXT: vunpcklps {{.*#+}} xmm6 = xmm14[0],xmm13[0],xmm14[1],xmm13[1]
+; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,2,1]
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm6[2,3],ymm0[4,5,6,7]
+; AVX2-NEXT: vmovaps (%rdi), %xmm1
+; AVX2-NEXT: vunpckhps {{.*#+}} xmm7 = xmm1[2],xmm11[2],xmm1[3],xmm11[3]
+; AVX2-NEXT: vmovups %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vinsertf128 $1, %xmm7, %ymm0, %ymm9
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm9[4,5],ymm0[6,7]
+; AVX2-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vbroadcastss (%rcx), %xmm0
+; AVX2-NEXT: vbroadcastss (%rdx), %xmm7
+; AVX2-NEXT: vunpcklps {{.*#+}} xmm0 = xmm7[0],xmm0[0],xmm7[1],xmm0[1]
+; AVX2-NEXT: vunpcklps {{.*#+}} xmm1 = xmm1[0],xmm11[0],xmm1[1],xmm11[1]
+; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,1,2,1]
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5,6,7]
+; AVX2-NEXT: vinsertf128 $1, %xmm6, %ymm0, %ymm1
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
+; AVX2-NEXT: vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vmovdqa 32(%r8), %ymm0
+; AVX2-NEXT: vmovdqu %ymm0, (%rsp) # 32-byte Spill
+; AVX2-NEXT: vmovdqa 32(%r9), %ymm11
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm0 = ymm0[2],ymm11[2],ymm0[3],ymm11[3],ymm0[6],ymm11[6],ymm0[7],ymm11[7]
+; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,1,2,3]
+; AVX2-NEXT: vmovdqa 32(%rdi), %ymm13
+; AVX2-NEXT: vmovdqa 32(%rsi), %ymm14
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm6 = ymm13[2],ymm14[2],ymm13[3],ymm14[3],ymm13[6],ymm14[6],ymm13[7],ymm14[7]
+; AVX2-NEXT: vextracti128 $1, %ymm6, %xmm1
+; AVX2-NEXT: vpblendd {{.*#+}} ymm7 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
+; AVX2-NEXT: vmovdqa 32(%rdx), %ymm1
+; AVX2-NEXT: vmovdqa 32(%rcx), %ymm0
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm15 = ymm1[2],ymm0[2],ymm1[3],ymm0[3],ymm1[6],ymm0[6],ymm1[7],ymm0[7]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm15 = ymm15[2,3,2,3,6,7,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm7 = ymm7[0,1,2,3],ymm15[4,5],ymm7[6,7]
+; AVX2-NEXT: vmovdqu %ymm7, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vpmovzxdq {{.*#+}} xmm7 = mem[0],zero,mem[1],zero
+; AVX2-NEXT: vpbroadcastd 52(%r9), %ymm15
+; AVX2-NEXT: vpblendd {{.*#+}} xmm7 = xmm7[0,1,2],xmm15[3]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm15 = ymm0[0,1,2,2,4,5,6,6]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm9 = ymm1[1,1,2,3,5,5,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm9 = ymm9[0],ymm15[1],ymm9[2],ymm15[3],ymm9[4],ymm15[5],ymm9[6],ymm15[7]
+; AVX2-NEXT: vpermq {{.*#+}} ymm9 = ymm9[2,1,2,3]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm7 = ymm9[0,1],ymm7[2,3],ymm9[4,5,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm6 = ymm7[0,1,2,3],ymm6[4,5],ymm7[6,7]
+; AVX2-NEXT: vmovdqu %ymm6, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vpunpckhdq {{.*#+}} xmm6 = xmm10[2],xmm12[2],xmm10[3],xmm12[3]
+; AVX2-NEXT: vpermq {{.*#+}} ymm6 = ymm6[0,1,2,1]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm6[0,1],ymm2[2,3],ymm6[4,5,6,7]
+; AVX2-NEXT: vpunpckhdq {{.*#+}} xmm4 = xmm4[2],xmm8[2],xmm4[3],xmm8[3]
+; AVX2-NEXT: vpshufd {{.*#+}} xmm4 = xmm4[2,3,2,3]
+; AVX2-NEXT: vinserti128 $1, %xmm4, %ymm0, %ymm4
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm4[4,5],ymm2[6,7]
+; AVX2-NEXT: vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vpbroadcastd 32(%rcx), %xmm2
+; AVX2-NEXT: vpbroadcastd 32(%rdx), %xmm6
+; AVX2-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm6[0],xmm2[0],xmm6[1],xmm2[1]
+; AVX2-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1]
+; AVX2-NEXT: vpermq {{.*#+}} ymm3 = ymm3[0,1,2,1]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm3[0,1],ymm2[2,3],ymm3[4,5,6,7]
+; AVX2-NEXT: vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm3 # 16-byte Folded Reload
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5],ymm2[6,7]
+; AVX2-NEXT: vmovdqu %ymm2, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; AVX2-NEXT: vmovdqa (%r8), %ymm3
+; AVX2-NEXT: vmovdqa (%r9), %ymm5
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm6 = ymm3[2],ymm5[2],ymm3[3],ymm5[3],ymm3[6],ymm5[6],ymm3[7],ymm5[7]
+; AVX2-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
+; AVX2-NEXT: vmovdqa (%rdi), %ymm7
+; AVX2-NEXT: vmovdqa (%rsi), %ymm8
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm9 = ymm7[2],ymm8[2],ymm7[3],ymm8[3],ymm7[6],ymm8[6],ymm7[7],ymm8[7]
+; AVX2-NEXT: vextracti128 $1, %ymm9, %xmm10
+; AVX2-NEXT: vpblendd {{.*#+}} ymm6 = ymm6[0,1],ymm10[2,3],ymm6[4,5,6,7]
+; AVX2-NEXT: vmovdqa (%rdx), %ymm10
+; AVX2-NEXT: vmovdqa (%rcx), %ymm12
+; AVX2-NEXT: vpunpckhdq {{.*#+}} ymm15 = ymm10[2],ymm12[2],ymm10[3],ymm12[3],ymm10[6],ymm12[6],ymm10[7],ymm12[7]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm15 = ymm15[2,3,2,3,6,7,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm15 = ymm6[0,1,2,3],ymm15[4,5],ymm6[6,7]
+; AVX2-NEXT: vpmovzxdq {{.*#+}} xmm6 = mem[0],zero,mem[1],zero
+; AVX2-NEXT: vpbroadcastd 20(%r9), %ymm4
+; AVX2-NEXT: vpblendd {{.*#+}} xmm4 = xmm6[0,1,2],xmm4[3]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm6 = ymm12[0,1,2,2,4,5,6,6]
+; AVX2-NEXT: vpshufd {{.*#+}} ymm2 = ymm10[1,1,2,3,5,5,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0],ymm6[1],ymm2[2],ymm6[3],ymm2[4],ymm6[5],ymm2[6],ymm6[7]
+; AVX2-NEXT: vpermq {{.*#+}} ymm2 = ymm2[2,1,2,3]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1],ymm4[2,3],ymm2[4,5,6,7]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm9[4,5],ymm2[6,7]
+; AVX2-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload
+; AVX2-NEXT: vpunpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm4, %xmm4 # 16-byte Folded Reload
+; AVX2-NEXT: # xmm4 = xmm4[2],mem[2],xmm4[3],mem[3]
+; AVX2-NEXT: vpermq {{.*#+}} ymm4 = ymm4[0,1,2,1]
+; AVX2-NEXT: vpblendd $12, {{[-0-9]+}}(%r{{[sb]}}p), %ymm4, %ymm4 # 32-byte Folded Reload
+; AVX2-NEXT: # ymm4 = ymm4[0,1],mem[2,3],ymm4[4,5,6,7]
+; AVX2-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Reload
+; AVX2-NEXT: vpunpckhdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm6, %xmm6 # 16-byte Folded Reload
+; AVX2-NEXT: # xmm6 = xmm6[2],mem[2],xmm6[3],mem[3]
+; AVX2-NEXT: vpshufd {{.*#+}} xmm6 = xmm6[2,3,2,3]
+; AVX2-NEXT: vinserti128 $1, %xmm6, %ymm0, %ymm6
+; AVX2-NEXT: vpblendd {{.*#+}} ymm4 = ymm4[0,1,2,3],ymm6[4,5],ymm4[6,7]
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[4],ymm0[4],ymm1[5],ymm0[5]
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm1 = ymm13[0],ymm14[0],ymm13[1],ymm14[1],ymm13[4],ymm14[4],ymm13[5],ymm14[5]
+; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,2,2,2]
+; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[2,1,2,3]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5,6,7]
+; AVX2-NEXT: vmovdqu (%rsp), %ymm1 # 32-byte Reload
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm1 = ymm1[0],ymm11[0],ymm1[1],ymm11[1],ymm1[4],ymm11[4],ymm1[5],ymm11[5]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5],ymm0[6,7]
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm1 = ymm10[0],ymm12[0],ymm10[1],ymm12[1],ymm10[4],ymm12[4],ymm10[5],ymm12[5]
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm6 = ymm7[0],ymm8[0],ymm7[1],ymm8[1],ymm7[4],ymm8[4],ymm7[5],ymm8[5]
+; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[2,2,2,2]
+; AVX2-NEXT: vpermq {{.*#+}} ymm6 = ymm6[2,1,2,3]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm6[0,1],ymm1[2,3],ymm6[4,5,6,7]
+; AVX2-NEXT: vpunpckldq {{.*#+}} ymm3 = ymm3[0],ymm5[0],ymm3[1],ymm5[1],ymm3[4],ymm5[4],ymm3[5],ymm5[5]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5],ymm1[6,7]
+; AVX2-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; AVX2-NEXT: vmovdqa %ymm1, 96(%rax)
+; AVX2-NEXT: vmovdqa %ymm0, 288(%rax)
+; AVX2-NEXT: vmovdqa %ymm4, 64(%rax)
+; AVX2-NEXT: vmovdqa %ymm2, 128(%rax)
+; AVX2-NEXT: vmovdqa %ymm15, 160(%rax)
+; AVX2-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX2-NEXT: vmovaps %ymm0, 192(%rax)
+; AVX2-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX2-NEXT: vmovaps %ymm0, 256(%rax)
+; AVX2-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX2-NEXT: vmovaps %ymm0, 320(%rax)
+; AVX2-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX2-NEXT: vmovaps %ymm0, 352(%rax)
+; AVX2-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX2-NEXT: vmovaps %ymm0, (%rax)
+; AVX2-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX2-NEXT: vmovaps %ymm0, 32(%rax)
+; AVX2-NEXT: vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; AVX2-NEXT: vmovaps %ymm0, 224(%rax)
+; AVX2-NEXT: addq $232, %rsp
+; AVX2-NEXT: vzeroupper
+; AVX2-NEXT: retq
+;
+; AVX512-LABEL: store_i32_stride6_vf16:
+; AVX512: # %bb.0:
+; AVX512-NEXT: movq {{[0-9]+}}(%rsp), %r10
+; AVX512-NEXT: vmovdqu64 (%rdi), %zmm1
+; AVX512-NEXT: vmovdqu64 (%rsi), %zmm2
+; AVX512-NEXT: vmovdqu64 (%rdx), %zmm3
+; AVX512-NEXT: vmovdqu64 (%rcx), %zmm4
+; AVX512-NEXT: vmovdqu64 (%r8), %zmm6
+; AVX512-NEXT: vmovdqu64 (%r9), %zmm7
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm5 = [9,25,8,24,0,0,10,26,9,25,8,24,0,0,10,26]
+; AVX512-NEXT: # zmm5 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm5
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm0 = [8,24,0,0,10,26,9,25,8,24,0,0,10,26,9,25]
+; AVX512-NEXT: # zmm0 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm2, %zmm1, %zmm0
+; AVX512-NEXT: movb $-110, %al
+; AVX512-NEXT: kmovd %eax, %k1
+; AVX512-NEXT: vmovdqa64 %zmm5, %zmm0 {%k1}
+; AVX512-NEXT: vpunpckldq {{.*#+}} zmm5 = zmm6[0],zmm7[0],zmm6[1],zmm7[1],zmm6[4],zmm7[4],zmm6[5],zmm7[5],zmm6[8],zmm7[8],zmm6[9],zmm7[9],zmm6[12],zmm7[12],zmm6[13],zmm7[13]
+; AVX512-NEXT: movb $36, %al
+; AVX512-NEXT: kmovd %eax, %k2
+; AVX512-NEXT: vshufi64x2 {{.*#+}} zmm0 {%k2} = zmm5[0,1,4,5,4,5,0,1]
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [4,20,3,19,0,0,5,21,4,20,3,19,0,0,5,21]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm2, %zmm1, %zmm8
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm5 = [2,18,0,0,4,20,3,19,2,18,0,0,4,20,3,19]
+; AVX512-NEXT: # zmm5 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm7, %zmm6, %zmm5
+; AVX512-NEXT: vmovdqa64 %zmm8, %zmm5 {%k1}
+; AVX512-NEXT: vmovdqa (%rdx), %ymm8
+; AVX512-NEXT: vmovdqa {{.*#+}} ymm9 = [3,11,0,8,7,15,4,12]
+; AVX512-NEXT: vpermi2d (%rcx), %ymm8, %ymm9
+; AVX512-NEXT: vshufi64x2 {{.*#+}} zmm5 {%k2} = zmm9[0,1,0,1,2,3,0,1]
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [6,22,5,21,0,0,7,23,6,22,5,21,0,0,7,23]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm7, %zmm6, %zmm8
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm9 = [5,21,0,0,7,23,6,22,5,21,0,0,7,23,6,22]
+; AVX512-NEXT: # zmm9 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm9
+; AVX512-NEXT: vmovdqa64 %zmm8, %zmm9 {%k1}
+; AVX512-NEXT: vmovdqa (%rdi), %ymm8
+; AVX512-NEXT: vpunpckhdq {{.*#+}} ymm8 = ymm8[2],mem[2],ymm8[3],mem[3],ymm8[6],mem[6],ymm8[7],mem[7]
+; AVX512-NEXT: vshufi64x2 {{.*#+}} zmm9 {%k2} = zmm8[0,1,2,3,2,3,0,1]
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm8 = [14,30,13,29,0,0,15,31,14,30,13,29,0,0,15,31]
+; AVX512-NEXT: # zmm8 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm7, %zmm6, %zmm8
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm10 = [13,29,0,0,15,31,14,30,13,29,0,0,15,31,14,30]
+; AVX512-NEXT: # zmm10 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm10
+; AVX512-NEXT: vmovdqa64 %zmm8, %zmm10 {%k1}
+; AVX512-NEXT: vpunpckhdq {{.*#+}} zmm8 = zmm1[2],zmm2[2],zmm1[3],zmm2[3],zmm1[6],zmm2[6],zmm1[7],zmm2[7],zmm1[10],zmm2[10],zmm1[11],zmm2[11],zmm1[14],zmm2[14],zmm1[15],zmm2[15]
+; AVX512-NEXT: vshufi64x2 {{.*#+}} zmm10 {%k2} = zmm8[0,1,6,7,6,7,0,1]
+; 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 %zmm2, %zmm1, %zmm8
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm11 = [10,26,0,0,12,28,11,27,10,26,0,0,12,28,11,27]
+; AVX512-NEXT: # zmm11 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm7, %zmm6, %zmm11
+; AVX512-NEXT: vmovdqa64 %zmm8, %zmm11 {%k1}
+; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm6 = [3,19,0,16,11,27,8,24,15,31,12,28,3,19,0,16]
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm6
+; AVX512-NEXT: vmovdqa64 %zmm6, %zmm11 {%k2}
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm6 = [1,17,0,16,0,0,2,18,1,17,0,16,0,0,2,18]
+; AVX512-NEXT: # zmm6 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm4, %zmm3, %zmm6
+; AVX512-NEXT: vbroadcasti64x4 {{.*#+}} zmm3 = [0,16,0,0,2,18,1,17,0,16,0,0,2,18,1,17]
+; AVX512-NEXT: # zmm3 = mem[0,1,2,3,0,1,2,3]
+; AVX512-NEXT: vpermi2d %zmm2, %zmm1, %zmm3
+; AVX512-NEXT: vmovdqa64 %zmm6, %zmm3 {%k1}
+; AVX512-NEXT: vmovdqa (%r8), %xmm1
+; AVX512-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
+; AVX512-NEXT: vshufi64x2 {{.*#+}} zmm3 {%k2} = zmm1[0,1,0,1,0,1,0,1]
+; AVX512-NEXT: vmovdqu64 %zmm3, (%r10)
+; AVX512-NEXT: vmovdqu64 %zmm11, 256(%r10)
+; AVX512-NEXT: vmovdqu64 %zmm10, 320(%r10)
+; AVX512-NEXT: vmovdqu64 %zmm9, 128(%r10)
+; AVX512-NEXT: vmovdqu64 %zmm5, 64(%r10)
+; AVX512-NEXT: vmovdqu64 %zmm0, 192(%r10)
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
+ %in.vec0 = load <16 x i32>, <16 x i32>* %in.vecptr0, align 32
+ %in.vec1 = load <16 x i32>, <16 x i32>* %in.vecptr1, align 32
+ %in.vec2 = load <16 x i32>, <16 x i32>* %in.vecptr2, align 32
+ %in.vec3 = load <16 x i32>, <16 x i32>* %in.vecptr3, align 32
+ %in.vec4 = load <16 x i32>, <16 x i32>* %in.vecptr4, align 32
+ %in.vec5 = load <16 x i32>, <16 x i32>* %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, <96 x i32>* %out.vec, align 32
+
+ ret void
+}
projects / platform / upstream / llvm.git / commitdiff