diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i16-stride-5.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i16-stride-5.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -116,83 +116,44 @@ define void @load_i16_stride5_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX2-FAST-PERLANE-NEXT:    vmovd %xmm1, (%r9)
 ; AVX2-FAST-PERLANE-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i16_stride5_vf2:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,2,2,3]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,3,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[0,3,2,3]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[1,2,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm4 = xmm0[3,1,2,3]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm4 = xmm4[2,1,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512F-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512F-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm5
-; AVX512F-SLOW-NEXT:    vpblendw {{.*#+}} xmm1 = xmm5[0],xmm1[1],xmm5[2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vmovd %xmm2, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovd %xmm3, (%rdx)
-; AVX512F-SLOW-NEXT:    vmovd %xmm4, (%rcx)
-; AVX512F-SLOW-NEXT:    vmovd %xmm0, (%r8)
-; AVX512F-SLOW-NEXT:    vmovd %xmm1, (%r9)
-; AVX512F-SLOW-NEXT:    retq
-;
-; AVX512F-FAST-LABEL: load_i16_stride5_vf2:
-; AVX512F-FAST:       # %bb.0:
-; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-FAST-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[0,1,10,11,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512F-FAST-NEXT:    vpshufb {{.*#+}} xmm3 = xmm0[2,3,12,13,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512F-FAST-NEXT:    vpshufb {{.*#+}} xmm4 = xmm0[4,5,14,15,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512F-FAST-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512F-FAST-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512F-FAST-NEXT:    vpbroadcastw 8(%rdi), %xmm5
-; AVX512F-FAST-NEXT:    vpblendw {{.*#+}} xmm1 = xmm5[0],xmm1[1],xmm5[2,3,4,5,6,7]
-; AVX512F-FAST-NEXT:    vmovd %xmm2, (%rsi)
-; AVX512F-FAST-NEXT:    vmovd %xmm3, (%rdx)
-; AVX512F-FAST-NEXT:    vmovd %xmm4, (%rcx)
-; AVX512F-FAST-NEXT:    vmovd %xmm0, (%r8)
-; AVX512F-FAST-NEXT:    vmovd %xmm1, (%r9)
-; AVX512F-FAST-NEXT:    retq
-;
-; AVX512BW-SLOW-LABEL: load_i16_stride5_vf2:
-; AVX512BW-SLOW:       # %bb.0:
-; AVX512BW-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,2,2,3]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,3,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[0,3,2,3]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[1,2,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm4 = xmm0[3,1,2,3]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm4 = xmm4[2,1,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512BW-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512BW-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm5
-; AVX512BW-SLOW-NEXT:    vpblendw {{.*#+}} xmm1 = xmm5[0],xmm1[1],xmm5[2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vmovd %xmm2, (%rsi)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm3, (%rdx)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm4, (%rcx)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm0, (%r8)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm1, (%r9)
-; AVX512BW-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i16_stride5_vf2:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,2,2,3]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,3,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[0,3,2,3]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[1,2,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm4 = xmm0[3,1,2,3]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm4 = xmm4[2,1,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vpsrlq $48, %xmm0, %xmm0
+; AVX512-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm5
+; AVX512-SLOW-NEXT:    vpblendw {{.*#+}} xmm1 = xmm5[0],xmm1[1],xmm5[2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vmovd %xmm2, (%rsi)
+; AVX512-SLOW-NEXT:    vmovd %xmm3, (%rdx)
+; AVX512-SLOW-NEXT:    vmovd %xmm4, (%rcx)
+; AVX512-SLOW-NEXT:    vmovd %xmm0, (%r8)
+; AVX512-SLOW-NEXT:    vmovd %xmm1, (%r9)
+; AVX512-SLOW-NEXT:    retq
 ;
-; AVX512BW-FAST-LABEL: load_i16_stride5_vf2:
-; AVX512BW-FAST:       # %bb.0:
-; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-FAST-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[0,1,10,11,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512BW-FAST-NEXT:    vpshufb {{.*#+}} xmm3 = xmm0[2,3,12,13,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512BW-FAST-NEXT:    vpshufb {{.*#+}} xmm4 = xmm0[4,5,14,15,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512BW-FAST-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512BW-FAST-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512BW-FAST-NEXT:    vpbroadcastw 8(%rdi), %xmm5
-; AVX512BW-FAST-NEXT:    vpblendw {{.*#+}} xmm1 = xmm5[0],xmm1[1],xmm5[2,3,4,5,6,7]
-; AVX512BW-FAST-NEXT:    vmovd %xmm2, (%rsi)
-; AVX512BW-FAST-NEXT:    vmovd %xmm3, (%rdx)
-; AVX512BW-FAST-NEXT:    vmovd %xmm4, (%rcx)
-; AVX512BW-FAST-NEXT:    vmovd %xmm0, (%r8)
-; AVX512BW-FAST-NEXT:    vmovd %xmm1, (%r9)
-; AVX512BW-FAST-NEXT:    retq
+; AVX512-FAST-LABEL: load_i16_stride5_vf2:
+; AVX512-FAST:       # %bb.0:
+; AVX512-FAST-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-FAST-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[0,1,10,11,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512-FAST-NEXT:    vpshufb {{.*#+}} xmm3 = xmm0[2,3,12,13,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512-FAST-NEXT:    vpshufb {{.*#+}} xmm4 = xmm0[4,5,14,15,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512-FAST-NEXT:    vpsrlq $48, %xmm0, %xmm0
+; AVX512-FAST-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-FAST-NEXT:    vpbroadcastw 8(%rdi), %xmm5
+; AVX512-FAST-NEXT:    vpblendw {{.*#+}} xmm1 = xmm5[0],xmm1[1],xmm5[2,3,4,5,6,7]
+; AVX512-FAST-NEXT:    vmovd %xmm2, (%rsi)
+; AVX512-FAST-NEXT:    vmovd %xmm3, (%rdx)
+; AVX512-FAST-NEXT:    vmovd %xmm4, (%rcx)
+; AVX512-FAST-NEXT:    vmovd %xmm0, (%r8)
+; AVX512-FAST-NEXT:    vmovd %xmm1, (%r9)
+; AVX512-FAST-NEXT:    retq
   %wide.vec = load <10 x i16>, ptr %in.vec, align 64
   %strided.vec0 = shufflevector <10 x i16> %wide.vec, <10 x i16> poison, <2 x i32> <i32 0, i32 5>
   %strided.vec1 = shufflevector <10 x i16> %wide.vec, <10 x i16> poison, <2 x i32> <i32 1, i32 6>
@@ -7766,8 +7727,10 @@ define void @load_i16_stride5_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX2: {{.*}}
 ; AVX2-ONLY: {{.*}}
 ; AVX512: {{.*}}
+; AVX512BW-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
+; AVX512BW-SLOW: {{.*}}
 ; AVX512DQ-FAST: {{.*}}
 ; AVX512DQ-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i16-stride-6.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i16-stride-6.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -150,32 +150,32 @@ define void @load_i16_stride6_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX2-FAST-PERLANE-NEXT:    vmovd %xmm0, (%rax)
 ; AVX2-FAST-PERLANE-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i16_stride6_vf2:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm2[0,2,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[1,3,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpbroadcastw 4(%rdi), %xmm4
-; AVX512F-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3]
-; AVX512F-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm5 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm5 = xmm5[3,1,2,3]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm5 = xmm5[0,3,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpbroadcastw 20(%rdi), %xmm6
-; AVX512F-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm7
-; AVX512F-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm6 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
-; AVX512F-SLOW-NEXT:    vpsrlq $48, %xmm1, %xmm1
-; AVX512F-SLOW-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512F-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512F-SLOW-NEXT:    vmovd %xmm3, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovd %xmm2, (%rdx)
-; AVX512F-SLOW-NEXT:    vmovd %xmm4, (%rcx)
-; AVX512F-SLOW-NEXT:    vmovd %xmm5, (%r8)
-; AVX512F-SLOW-NEXT:    vmovd %xmm6, (%r9)
-; AVX512F-SLOW-NEXT:    vmovd %xmm0, (%rax)
-; AVX512F-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i16_stride6_vf2:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
+; AVX512-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm2[0,2,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[1,3,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vpbroadcastw 4(%rdi), %xmm4
+; AVX512-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3]
+; AVX512-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm5 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm5 = xmm5[3,1,2,3]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm5 = xmm5[0,3,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vpbroadcastw 20(%rdi), %xmm6
+; AVX512-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm7
+; AVX512-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm6 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
+; AVX512-SLOW-NEXT:    vpsrlq $48, %xmm1, %xmm1
+; AVX512-SLOW-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; AVX512-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-SLOW-NEXT:    vmovd %xmm3, (%rsi)
+; AVX512-SLOW-NEXT:    vmovd %xmm2, (%rdx)
+; AVX512-SLOW-NEXT:    vmovd %xmm4, (%rcx)
+; AVX512-SLOW-NEXT:    vmovd %xmm5, (%r8)
+; AVX512-SLOW-NEXT:    vmovd %xmm6, (%r9)
+; AVX512-SLOW-NEXT:    vmovd %xmm0, (%rax)
+; AVX512-SLOW-NEXT:    retq
 ;
 ; AVX512F-FAST-LABEL: load_i16_stride6_vf2:
 ; AVX512F-FAST:       # %bb.0:
@@ -203,33 +203,6 @@ define void @load_i16_stride6_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX512F-FAST-NEXT:    vmovd %xmm0, (%rax)
 ; AVX512F-FAST-NEXT:    retq
 ;
-; AVX512BW-SLOW-LABEL: load_i16_stride6_vf2:
-; AVX512BW-SLOW:       # %bb.0:
-; AVX512BW-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512BW-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm2[0,2,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[1,3,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpbroadcastw 4(%rdi), %xmm4
-; AVX512BW-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3]
-; AVX512BW-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm5 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm5 = xmm5[3,1,2,3]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm5 = xmm5[0,3,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpbroadcastw 20(%rdi), %xmm6
-; AVX512BW-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm7
-; AVX512BW-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm6 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
-; AVX512BW-SLOW-NEXT:    vpsrlq $48, %xmm1, %xmm1
-; AVX512BW-SLOW-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512BW-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512BW-SLOW-NEXT:    vmovd %xmm3, (%rsi)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm2, (%rdx)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm4, (%rcx)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm5, (%r8)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm6, (%r9)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm0, (%rax)
-; AVX512BW-SLOW-NEXT:    retq
-;
 ; AVX512BW-FAST-LABEL: load_i16_stride6_vf2:
 ; AVX512BW-FAST:       # %bb.0:
 ; AVX512BW-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %rax
@@ -11906,8 +11879,10 @@ define void @load_i16_stride6_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX2: {{.*}}
 ; AVX2-ONLY: {{.*}}
 ; AVX512: {{.*}}
+; AVX512-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
+; AVX512BW-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}
 ; AVX512DQBW-SLOW: {{.*}}
 ; AVX512F: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i16-stride-7.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i16-stride-7.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -166,35 +166,35 @@ define void @load_i16_stride7_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX2-FAST-PERLANE-NEXT:    vmovd %xmm0, (%rax)
 ; AVX2-FAST-PERLANE-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i16_stride7_vf2:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512F-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %r10
-; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,3,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpsrld $16, %xmm0, %xmm3
-; AVX512F-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
-; AVX512F-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm5 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm6 = xmm5[2,1,2,3]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm6 = xmm6[0,3,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,6,5,6,7]
-; AVX512F-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm7
-; AVX512F-SLOW-NEXT:    vpsrlq $48, %xmm1, %xmm8
-; AVX512F-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm7 = xmm7[0],xmm8[0],xmm7[1],xmm8[1],xmm7[2],xmm8[2],xmm7[3],xmm8[3]
-; AVX512F-SLOW-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; AVX512F-SLOW-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
-; AVX512F-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
-; AVX512F-SLOW-NEXT:    vmovd %xmm2, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovd %xmm4, (%rdx)
-; AVX512F-SLOW-NEXT:    vmovd %xmm6, (%rcx)
-; AVX512F-SLOW-NEXT:    vpextrd $2, %xmm5, (%r8)
-; AVX512F-SLOW-NEXT:    vmovd %xmm7, (%r9)
-; AVX512F-SLOW-NEXT:    vmovd %xmm3, (%r10)
-; AVX512F-SLOW-NEXT:    vmovd %xmm0, (%rax)
-; AVX512F-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i16_stride7_vf2:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
+; AVX512-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %r10
+; AVX512-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,3,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vpsrld $16, %xmm0, %xmm3
+; AVX512-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
+; AVX512-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm5 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm6 = xmm5[2,1,2,3]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm6 = xmm6[0,3,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vpshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,6,5,6,7]
+; AVX512-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm7
+; AVX512-SLOW-NEXT:    vpsrlq $48, %xmm1, %xmm8
+; AVX512-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm7 = xmm7[0],xmm8[0],xmm7[1],xmm8[1],xmm7[2],xmm8[2],xmm7[3],xmm8[3]
+; AVX512-SLOW-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
+; AVX512-SLOW-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
+; AVX512-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
+; AVX512-SLOW-NEXT:    vmovd %xmm2, (%rsi)
+; AVX512-SLOW-NEXT:    vmovd %xmm4, (%rdx)
+; AVX512-SLOW-NEXT:    vmovd %xmm6, (%rcx)
+; AVX512-SLOW-NEXT:    vpextrd $2, %xmm5, (%r8)
+; AVX512-SLOW-NEXT:    vmovd %xmm7, (%r9)
+; AVX512-SLOW-NEXT:    vmovd %xmm3, (%r10)
+; AVX512-SLOW-NEXT:    vmovd %xmm0, (%rax)
+; AVX512-SLOW-NEXT:    retq
 ;
 ; AVX512F-FAST-LABEL: load_i16_stride7_vf2:
 ; AVX512F-FAST:       # %bb.0:
@@ -224,36 +224,6 @@ define void @load_i16_stride7_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX512F-FAST-NEXT:    vmovd %xmm0, (%rax)
 ; AVX512F-FAST-NEXT:    retq
 ;
-; AVX512BW-SLOW-LABEL: load_i16_stride7_vf2:
-; AVX512BW-SLOW:       # %bb.0:
-; AVX512BW-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512BW-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %r10
-; AVX512BW-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,3,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpsrld $16, %xmm0, %xmm3
-; AVX512BW-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
-; AVX512BW-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm5 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm6 = xmm5[2,1,2,3]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm6 = xmm6[0,3,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,6,5,6,7]
-; AVX512BW-SLOW-NEXT:    vpbroadcastw 8(%rdi), %xmm7
-; AVX512BW-SLOW-NEXT:    vpsrlq $48, %xmm1, %xmm8
-; AVX512BW-SLOW-NEXT:    vpunpcklwd {{.*#+}} xmm7 = xmm7[0],xmm8[0],xmm7[1],xmm8[1],xmm7[2],xmm8[2],xmm7[3],xmm8[3]
-; AVX512BW-SLOW-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; AVX512BW-SLOW-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
-; AVX512BW-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,3,2,3,4,5,6,7]
-; AVX512BW-SLOW-NEXT:    vmovd %xmm2, (%rsi)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm4, (%rdx)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm6, (%rcx)
-; AVX512BW-SLOW-NEXT:    vpextrd $2, %xmm5, (%r8)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm7, (%r9)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm3, (%r10)
-; AVX512BW-SLOW-NEXT:    vmovd %xmm0, (%rax)
-; AVX512BW-SLOW-NEXT:    retq
-;
 ; AVX512BW-FAST-LABEL: load_i16_stride7_vf2:
 ; AVX512BW-FAST:       # %bb.0:
 ; AVX512BW-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %rax
@@ -16204,8 +16174,10 @@ define void @load_i16_stride7_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX2: {{.*}}
 ; AVX2-ONLY: {{.*}}
 ; AVX512: {{.*}}
+; AVX512-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
+; AVX512BW-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}
 ; AVX512DQBW-SLOW: {{.*}}
 ; AVX512F: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i16-stride-8.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i16-stride-8.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -9893,6 +9893,8 @@ define void @load_i16_stride8_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX2-FAST-PERLANE: {{.*}}
 ; AVX2-SLOW: {{.*}}
 ; AVX512: {{.*}}
+; AVX512-FAST: {{.*}}
+; AVX512-SLOW: {{.*}}
 ; AVX512BW-FAST: {{.*}}
 ; AVX512BW-SLOW: {{.*}}
 ; AVX512DQ-FAST: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-2.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-2.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -124,49 +124,27 @@ define void @load_i32_stride2_vf8(ptr %in.vec, ptr %out.vec0, ptr %out.vec1) nou
 ; AVX2-ONLY-NEXT:    vzeroupper
 ; AVX2-ONLY-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i32_stride2_vf8:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    vmovdqa64 (%rdi), %zmm0
-; AVX512F-SLOW-NEXT:    vmovaps (%rdi), %ymm1
-; AVX512F-SLOW-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[1,3],mem[1,3],ymm1[5,7],mem[5,7]
-; AVX512F-SLOW-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,2,1,3]
-; AVX512F-SLOW-NEXT:    vpmovqd %zmm0, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovaps %ymm1, (%rdx)
-; AVX512F-SLOW-NEXT:    vzeroupper
-; AVX512F-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i32_stride2_vf8:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    vmovdqa64 (%rdi), %zmm0
+; AVX512-SLOW-NEXT:    vmovaps (%rdi), %ymm1
+; AVX512-SLOW-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[1,3],mem[1,3],ymm1[5,7],mem[5,7]
+; AVX512-SLOW-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,2,1,3]
+; AVX512-SLOW-NEXT:    vpmovqd %zmm0, (%rsi)
+; AVX512-SLOW-NEXT:    vmovaps %ymm1, (%rdx)
+; AVX512-SLOW-NEXT:    vzeroupper
+; AVX512-SLOW-NEXT:    retq
 ;
-; AVX512F-FAST-LABEL: load_i32_stride2_vf8:
-; AVX512F-FAST:       # %bb.0:
-; AVX512F-FAST-NEXT:    vmovdqa64 (%rdi), %zmm0
-; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %ymm1
-; AVX512F-FAST-NEXT:    vmovdqa {{.*#+}} ymm2 = [1,3,5,7,9,11,13,15]
-; AVX512F-FAST-NEXT:    vpermi2d 32(%rdi), %ymm1, %ymm2
-; AVX512F-FAST-NEXT:    vpmovqd %zmm0, (%rsi)
-; AVX512F-FAST-NEXT:    vmovdqa %ymm2, (%rdx)
-; AVX512F-FAST-NEXT:    vzeroupper
-; AVX512F-FAST-NEXT:    retq
-;
-; AVX512BW-SLOW-LABEL: load_i32_stride2_vf8:
-; AVX512BW-SLOW:       # %bb.0:
-; AVX512BW-SLOW-NEXT:    vmovdqa64 (%rdi), %zmm0
-; AVX512BW-SLOW-NEXT:    vmovaps (%rdi), %ymm1
-; AVX512BW-SLOW-NEXT:    vshufps {{.*#+}} ymm1 = ymm1[1,3],mem[1,3],ymm1[5,7],mem[5,7]
-; AVX512BW-SLOW-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,2,1,3]
-; AVX512BW-SLOW-NEXT:    vpmovqd %zmm0, (%rsi)
-; AVX512BW-SLOW-NEXT:    vmovaps %ymm1, (%rdx)
-; AVX512BW-SLOW-NEXT:    vzeroupper
-; AVX512BW-SLOW-NEXT:    retq
-;
-; AVX512BW-FAST-LABEL: load_i32_stride2_vf8:
-; AVX512BW-FAST:       # %bb.0:
-; AVX512BW-FAST-NEXT:    vmovdqa64 (%rdi), %zmm0
-; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %ymm1
-; AVX512BW-FAST-NEXT:    vmovdqa {{.*#+}} ymm2 = [1,3,5,7,9,11,13,15]
-; AVX512BW-FAST-NEXT:    vpermi2d 32(%rdi), %ymm1, %ymm2
-; AVX512BW-FAST-NEXT:    vpmovqd %zmm0, (%rsi)
-; AVX512BW-FAST-NEXT:    vmovdqa %ymm2, (%rdx)
-; AVX512BW-FAST-NEXT:    vzeroupper
-; AVX512BW-FAST-NEXT:    retq
+; AVX512-FAST-LABEL: load_i32_stride2_vf8:
+; AVX512-FAST:       # %bb.0:
+; AVX512-FAST-NEXT:    vmovdqa64 (%rdi), %zmm0
+; AVX512-FAST-NEXT:    vmovdqa (%rdi), %ymm1
+; AVX512-FAST-NEXT:    vmovdqa {{.*#+}} ymm2 = [1,3,5,7,9,11,13,15]
+; AVX512-FAST-NEXT:    vpermi2d 32(%rdi), %ymm1, %ymm2
+; AVX512-FAST-NEXT:    vpmovqd %zmm0, (%rsi)
+; AVX512-FAST-NEXT:    vmovdqa %ymm2, (%rdx)
+; AVX512-FAST-NEXT:    vzeroupper
+; AVX512-FAST-NEXT:    retq
   %wide.vec = load <16 x i32>, ptr %in.vec, align 64
   %strided.vec0 = shufflevector <16 x i32> %wide.vec, <16 x i32> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
   %strided.vec1 = shufflevector <16 x i32> %wide.vec, <16 x i32> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
@@ -761,15 +739,19 @@ define void @load_i32_stride2_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1) no
 ; AVX2-FAST-PERLANE: {{.*}}
 ; AVX2-SLOW: {{.*}}
 ; AVX512BW: {{.*}}
+; AVX512BW-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
+; AVX512BW-SLOW: {{.*}}
 ; AVX512DQ-FAST: {{.*}}
 ; AVX512DQ-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}
 ; AVX512DQBW-SLOW: {{.*}}
 ; AVX512F: {{.*}}
+; AVX512F-FAST: {{.*}}
 ; AVX512F-ONLY-FAST: {{.*}}
 ; AVX512F-ONLY-SLOW: {{.*}}
+; AVX512F-SLOW: {{.*}}
 ; FALLBACK0: {{.*}}
 ; FALLBACK1: {{.*}}
 ; FALLBACK10: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-3.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-3.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -59,61 +59,33 @@ define void @load_i32_stride3_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX2-ONLY-NEXT:    vmovlps %xmm1, (%rcx)
 ; AVX2-ONLY-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i32_stride3_vf2:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    vmovaps (%rdi), %xmm0
-; AVX512F-SLOW-NEXT:    vmovaps 16(%rdi), %xmm1
-; AVX512F-SLOW-NEXT:    vshufps {{.*#+}} xmm2 = xmm0[0,3,2,3]
-; AVX512F-SLOW-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
-; AVX512F-SLOW-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[1,0,2,3]
-; AVX512F-SLOW-NEXT:    vbroadcastss 8(%rdi), %xmm3
-; AVX512F-SLOW-NEXT:    vblendps {{.*#+}} xmm1 = xmm3[0],xmm1[1],xmm3[2,3]
-; AVX512F-SLOW-NEXT:    vmovlps %xmm2, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovlps %xmm0, (%rdx)
-; AVX512F-SLOW-NEXT:    vmovlps %xmm1, (%rcx)
-; AVX512F-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i32_stride3_vf2:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    vmovaps (%rdi), %xmm0
+; AVX512-SLOW-NEXT:    vmovaps 16(%rdi), %xmm1
+; AVX512-SLOW-NEXT:    vshufps {{.*#+}} xmm2 = xmm0[0,3,2,3]
+; AVX512-SLOW-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
+; AVX512-SLOW-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[1,0,2,3]
+; AVX512-SLOW-NEXT:    vbroadcastss 8(%rdi), %xmm3
+; AVX512-SLOW-NEXT:    vblendps {{.*#+}} xmm1 = xmm3[0],xmm1[1],xmm3[2,3]
+; AVX512-SLOW-NEXT:    vmovlps %xmm2, (%rsi)
+; AVX512-SLOW-NEXT:    vmovlps %xmm0, (%rdx)
+; AVX512-SLOW-NEXT:    vmovlps %xmm1, (%rcx)
+; AVX512-SLOW-NEXT:    retq
 ;
-; AVX512F-FAST-LABEL: load_i32_stride3_vf2:
-; AVX512F-FAST:       # %bb.0:
-; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-FAST-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [5,0,5,0]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm0, %xmm1, %xmm3
-; AVX512F-FAST-NEXT:    vpbroadcastd 8(%rdi), %xmm0
-; AVX512F-FAST-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
-; AVX512F-FAST-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512F-FAST-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512F-FAST-NEXT:    vmovq %xmm0, (%rcx)
-; AVX512F-FAST-NEXT:    retq
-;
-; AVX512BW-SLOW-LABEL: load_i32_stride3_vf2:
-; AVX512BW-SLOW:       # %bb.0:
-; AVX512BW-SLOW-NEXT:    vmovaps (%rdi), %xmm0
-; AVX512BW-SLOW-NEXT:    vmovaps 16(%rdi), %xmm1
-; AVX512BW-SLOW-NEXT:    vshufps {{.*#+}} xmm2 = xmm0[0,3,2,3]
-; AVX512BW-SLOW-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
-; AVX512BW-SLOW-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[1,0,2,3]
-; AVX512BW-SLOW-NEXT:    vbroadcastss 8(%rdi), %xmm3
-; AVX512BW-SLOW-NEXT:    vblendps {{.*#+}} xmm1 = xmm3[0],xmm1[1],xmm3[2,3]
-; AVX512BW-SLOW-NEXT:    vmovlps %xmm2, (%rsi)
-; AVX512BW-SLOW-NEXT:    vmovlps %xmm0, (%rdx)
-; AVX512BW-SLOW-NEXT:    vmovlps %xmm1, (%rcx)
-; AVX512BW-SLOW-NEXT:    retq
-;
-; AVX512BW-FAST-LABEL: load_i32_stride3_vf2:
-; AVX512BW-FAST:       # %bb.0:
-; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-FAST-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [5,0,5,0]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm0, %xmm1, %xmm3
-; AVX512BW-FAST-NEXT:    vpbroadcastd 8(%rdi), %xmm0
-; AVX512BW-FAST-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
-; AVX512BW-FAST-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512BW-FAST-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512BW-FAST-NEXT:    vmovq %xmm0, (%rcx)
-; AVX512BW-FAST-NEXT:    retq
+; AVX512-FAST-LABEL: load_i32_stride3_vf2:
+; AVX512-FAST:       # %bb.0:
+; AVX512-FAST-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-FAST-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,3,2,3]
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [5,0,5,0]
+; AVX512-FAST-NEXT:    vpermi2d %xmm0, %xmm1, %xmm3
+; AVX512-FAST-NEXT:    vpbroadcastd 8(%rdi), %xmm0
+; AVX512-FAST-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX512-FAST-NEXT:    vmovq %xmm2, (%rsi)
+; AVX512-FAST-NEXT:    vmovq %xmm3, (%rdx)
+; AVX512-FAST-NEXT:    vmovq %xmm0, (%rcx)
+; AVX512-FAST-NEXT:    retq
   %wide.vec = load <6 x i32>, ptr %in.vec, align 64
   %strided.vec0 = shufflevector <6 x i32> %wide.vec, <6 x i32> poison, <2 x i32> <i32 0, i32 3>
   %strided.vec1 = shufflevector <6 x i32> %wide.vec, <6 x i32> poison, <2 x i32> <i32 1, i32 4>
@@ -3049,15 +3021,19 @@ define void @load_i32_stride3_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX1: {{.*}}
 ; AVX2: {{.*}}
 ; AVX512BW: {{.*}}
+; AVX512BW-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
+; AVX512BW-SLOW: {{.*}}
 ; AVX512DQ-FAST: {{.*}}
 ; AVX512DQ-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}
 ; AVX512DQBW-SLOW: {{.*}}
 ; AVX512F: {{.*}}
+; AVX512F-FAST: {{.*}}
 ; AVX512F-ONLY-FAST: {{.*}}
 ; AVX512F-ONLY-SLOW: {{.*}}
+; AVX512F-SLOW: {{.*}}
 ; FALLBACK0: {{.*}}
 ; FALLBACK1: {{.*}}
 ; FALLBACK10: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-4.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-4.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -59,61 +59,33 @@ define void @load_i32_stride4_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX2-ONLY-NEXT:    vpextrq $1, %xmm0, (%r8)
 ; AVX2-ONLY-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i32_stride4_vf2:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-SLOW-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
-; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2,3]
-; AVX512F-SLOW-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512F-SLOW-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512F-SLOW-NEXT:    vmovq %xmm0, (%rcx)
-; AVX512F-SLOW-NEXT:    vpextrq $1, %xmm0, (%r8)
-; AVX512F-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i32_stride4_vf2:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-SLOW-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
+; AVX512-SLOW-NEXT:    vpblendd {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2,3]
+; AVX512-SLOW-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-SLOW-NEXT:    vmovq %xmm2, (%rsi)
+; AVX512-SLOW-NEXT:    vmovq %xmm3, (%rdx)
+; AVX512-SLOW-NEXT:    vmovq %xmm0, (%rcx)
+; AVX512-SLOW-NEXT:    vpextrq $1, %xmm0, (%r8)
+; AVX512-SLOW-NEXT:    retq
 ;
-; AVX512F-FAST-LABEL: load_i32_stride4_vf2:
-; AVX512F-FAST:       # %bb.0:
-; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-FAST-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [1,5,1,5]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm3
-; AVX512F-FAST-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512F-FAST-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512F-FAST-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512F-FAST-NEXT:    vmovq %xmm0, (%rcx)
-; AVX512F-FAST-NEXT:    vpextrq $1, %xmm0, (%r8)
-; AVX512F-FAST-NEXT:    retq
-;
-; AVX512BW-SLOW-LABEL: load_i32_stride4_vf2:
-; AVX512BW-SLOW:       # %bb.0:
-; AVX512BW-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-SLOW-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
-; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2,3]
-; AVX512BW-SLOW-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512BW-SLOW-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm0, (%rcx)
-; AVX512BW-SLOW-NEXT:    vpextrq $1, %xmm0, (%r8)
-; AVX512BW-SLOW-NEXT:    retq
-;
-; AVX512BW-FAST-LABEL: load_i32_stride4_vf2:
-; AVX512BW-FAST:       # %bb.0:
-; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-FAST-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [1,5,1,5]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm3
-; AVX512BW-FAST-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX512BW-FAST-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512BW-FAST-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512BW-FAST-NEXT:    vmovq %xmm0, (%rcx)
-; AVX512BW-FAST-NEXT:    vpextrq $1, %xmm0, (%r8)
-; AVX512BW-FAST-NEXT:    retq
+; AVX512-FAST-LABEL: load_i32_stride4_vf2:
+; AVX512-FAST:       # %bb.0:
+; AVX512-FAST-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-FAST-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [1,5,1,5]
+; AVX512-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm3
+; AVX512-FAST-NEXT:    vpunpckhdq {{.*#+}} xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-FAST-NEXT:    vmovq %xmm2, (%rsi)
+; AVX512-FAST-NEXT:    vmovq %xmm3, (%rdx)
+; AVX512-FAST-NEXT:    vmovq %xmm0, (%rcx)
+; AVX512-FAST-NEXT:    vpextrq $1, %xmm0, (%r8)
+; AVX512-FAST-NEXT:    retq
   %wide.vec = load <8 x i32>, ptr %in.vec, align 64
   %strided.vec0 = shufflevector <8 x i32> %wide.vec, <8 x i32> poison, <2 x i32> <i32 0, i32 4>
   %strided.vec1 = shufflevector <8 x i32> %wide.vec, <8 x i32> poison, <2 x i32> <i32 1, i32 5>
@@ -3466,15 +3438,19 @@ define void @load_i32_stride4_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX2-FAST-PERLANE: {{.*}}
 ; AVX2-SLOW: {{.*}}
 ; AVX512BW: {{.*}}
+; AVX512BW-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
+; AVX512BW-SLOW: {{.*}}
 ; AVX512DQ-FAST: {{.*}}
 ; AVX512DQ-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}
 ; AVX512DQBW-SLOW: {{.*}}
 ; AVX512F: {{.*}}
+; AVX512F-FAST: {{.*}}
 ; AVX512F-ONLY-FAST: {{.*}}
 ; AVX512F-ONLY-SLOW: {{.*}}
+; AVX512F-SLOW: {{.*}}
 ; FALLBACK0: {{.*}}
 ; FALLBACK1: {{.*}}
 ; FALLBACK10: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-5.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-5.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -78,93 +78,49 @@ define void @load_i32_stride5_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX2-ONLY-NEXT:    vzeroupper
 ; AVX2-ONLY-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i32_stride5_vf2:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm2
-; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
-; AVX512F-SLOW-NEXT:    vpextrd $2, %xmm1, %eax
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm4 = xmm0[1,1,1,1]
-; AVX512F-SLOW-NEXT:    vpinsrd $1, %eax, %xmm4, %xmm4
-; AVX512F-SLOW-NEXT:    vpbroadcastd 8(%rdi), %xmm5
-; AVX512F-SLOW-NEXT:    vpextrd $3, %xmm1, %eax
-; AVX512F-SLOW-NEXT:    vpinsrd $1, %eax, %xmm5, %xmm1
-; AVX512F-SLOW-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
-; AVX512F-SLOW-NEXT:    vpbroadcastd 16(%rdi), %ymm5
-; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm2 = xmm5[0],xmm2[1],xmm5[2,3]
-; AVX512F-SLOW-NEXT:    vmovq %xmm3, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovq %xmm4, (%rdx)
-; AVX512F-SLOW-NEXT:    vmovq %xmm1, (%rcx)
-; AVX512F-SLOW-NEXT:    vmovq %xmm0, (%r8)
-; AVX512F-SLOW-NEXT:    vmovq %xmm2, (%r9)
-; AVX512F-SLOW-NEXT:    vzeroupper
-; AVX512F-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i32_stride5_vf2:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm2
+; AVX512-SLOW-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX512-SLOW-NEXT:    vpextrd $2, %xmm1, %eax
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm4 = xmm0[1,1,1,1]
+; AVX512-SLOW-NEXT:    vpinsrd $1, %eax, %xmm4, %xmm4
+; AVX512-SLOW-NEXT:    vpbroadcastd 8(%rdi), %xmm5
+; AVX512-SLOW-NEXT:    vpextrd $3, %xmm1, %eax
+; AVX512-SLOW-NEXT:    vpinsrd $1, %eax, %xmm5, %xmm1
+; AVX512-SLOW-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
+; AVX512-SLOW-NEXT:    vpbroadcastd 16(%rdi), %ymm5
+; AVX512-SLOW-NEXT:    vpblendd {{.*#+}} xmm2 = xmm5[0],xmm2[1],xmm5[2,3]
+; AVX512-SLOW-NEXT:    vmovq %xmm3, (%rsi)
+; AVX512-SLOW-NEXT:    vmovq %xmm4, (%rdx)
+; AVX512-SLOW-NEXT:    vmovq %xmm1, (%rcx)
+; AVX512-SLOW-NEXT:    vmovq %xmm0, (%r8)
+; AVX512-SLOW-NEXT:    vmovq %xmm2, (%r9)
+; AVX512-SLOW-NEXT:    vzeroupper
+; AVX512-SLOW-NEXT:    retq
 ;
-; AVX512F-FAST-LABEL: load_i32_stride5_vf2:
-; AVX512F-FAST:       # %bb.0:
-; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512F-FAST-NEXT:    vmovdqa 32(%rdi), %xmm2
-; AVX512F-FAST-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm4 = [1,6,1,6]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm4
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [2,7,2,7]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm5
-; AVX512F-FAST-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
-; AVX512F-FAST-NEXT:    vpbroadcastd 16(%rdi), %ymm1
-; AVX512F-FAST-NEXT:    vpblendd {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2,3]
-; AVX512F-FAST-NEXT:    vmovq %xmm3, (%rsi)
-; AVX512F-FAST-NEXT:    vmovq %xmm4, (%rdx)
-; AVX512F-FAST-NEXT:    vmovq %xmm5, (%rcx)
-; AVX512F-FAST-NEXT:    vmovq %xmm0, (%r8)
-; AVX512F-FAST-NEXT:    vmovq %xmm1, (%r9)
-; AVX512F-FAST-NEXT:    vzeroupper
-; AVX512F-FAST-NEXT:    retq
-;
-; AVX512BW-SLOW-LABEL: load_i32_stride5_vf2:
-; AVX512BW-SLOW:       # %bb.0:
-; AVX512BW-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-SLOW-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm2
-; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
-; AVX512BW-SLOW-NEXT:    vpextrd $2, %xmm1, %eax
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm4 = xmm0[1,1,1,1]
-; AVX512BW-SLOW-NEXT:    vpinsrd $1, %eax, %xmm4, %xmm4
-; AVX512BW-SLOW-NEXT:    vpbroadcastd 8(%rdi), %xmm5
-; AVX512BW-SLOW-NEXT:    vpextrd $3, %xmm1, %eax
-; AVX512BW-SLOW-NEXT:    vpinsrd $1, %eax, %xmm5, %xmm1
-; AVX512BW-SLOW-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
-; AVX512BW-SLOW-NEXT:    vpbroadcastd 16(%rdi), %ymm5
-; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm2 = xmm5[0],xmm2[1],xmm5[2,3]
-; AVX512BW-SLOW-NEXT:    vmovq %xmm3, (%rsi)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm4, (%rdx)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm1, (%rcx)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm0, (%r8)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm2, (%r9)
-; AVX512BW-SLOW-NEXT:    vzeroupper
-; AVX512BW-SLOW-NEXT:    retq
-;
-; AVX512BW-FAST-LABEL: load_i32_stride5_vf2:
-; AVX512BW-FAST:       # %bb.0:
-; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX512BW-FAST-NEXT:    vmovdqa 32(%rdi), %xmm2
-; AVX512BW-FAST-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm4 = [1,6,1,6]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm4
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [2,7,2,7]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm5
-; AVX512BW-FAST-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
-; AVX512BW-FAST-NEXT:    vpbroadcastd 16(%rdi), %ymm1
-; AVX512BW-FAST-NEXT:    vpblendd {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2,3]
-; AVX512BW-FAST-NEXT:    vmovq %xmm3, (%rsi)
-; AVX512BW-FAST-NEXT:    vmovq %xmm4, (%rdx)
-; AVX512BW-FAST-NEXT:    vmovq %xmm5, (%rcx)
-; AVX512BW-FAST-NEXT:    vmovq %xmm0, (%r8)
-; AVX512BW-FAST-NEXT:    vmovq %xmm1, (%r9)
-; AVX512BW-FAST-NEXT:    vzeroupper
-; AVX512BW-FAST-NEXT:    retq
+; AVX512-FAST-LABEL: load_i32_stride5_vf2:
+; AVX512-FAST:       # %bb.0:
+; AVX512-FAST-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-FAST-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX512-FAST-NEXT:    vmovdqa 32(%rdi), %xmm2
+; AVX512-FAST-NEXT:    vpblendd {{.*#+}} xmm3 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm4 = [1,6,1,6]
+; AVX512-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm4
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [2,7,2,7]
+; AVX512-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm5
+; AVX512-FAST-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[12,13,14,15],xmm2[0,1,2,3,4,5,6,7,8,9,10,11]
+; AVX512-FAST-NEXT:    vpbroadcastd 16(%rdi), %ymm1
+; AVX512-FAST-NEXT:    vpblendd {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2,3]
+; AVX512-FAST-NEXT:    vmovq %xmm3, (%rsi)
+; AVX512-FAST-NEXT:    vmovq %xmm4, (%rdx)
+; AVX512-FAST-NEXT:    vmovq %xmm5, (%rcx)
+; AVX512-FAST-NEXT:    vmovq %xmm0, (%r8)
+; AVX512-FAST-NEXT:    vmovq %xmm1, (%r9)
+; AVX512-FAST-NEXT:    vzeroupper
+; AVX512-FAST-NEXT:    retq
   %wide.vec = load <10 x i32>, ptr %in.vec, align 64
   %strided.vec0 = shufflevector <10 x i32> %wide.vec, <10 x i32> poison, <2 x i32> <i32 0, i32 5>
   %strided.vec1 = shufflevector <10 x i32> %wide.vec, <10 x i32> poison, <2 x i32> <i32 1, i32 6>
@@ -5295,14 +5251,18 @@ define void @load_i32_stride5_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX2-FAST: {{.*}}
 ; AVX2-FAST-PERLANE: {{.*}}
 ; AVX2-SLOW: {{.*}}
+; AVX512BW-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
+; AVX512BW-SLOW: {{.*}}
 ; AVX512DQ-FAST: {{.*}}
 ; AVX512DQ-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}
 ; AVX512DQBW-SLOW: {{.*}}
+; AVX512F-FAST: {{.*}}
 ; AVX512F-ONLY-FAST: {{.*}}
 ; AVX512F-ONLY-SLOW: {{.*}}
+; AVX512F-SLOW: {{.*}}
 ; FALLBACK0: {{.*}}
 ; FALLBACK1: {{.*}}
 ; FALLBACK10: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-6.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-6.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -102,69 +102,69 @@ define void @load_i32_stride6_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX2-ONLY-NEXT:    vzeroupper
 ; AVX2-ONLY-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i32_stride6_vf2:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-SLOW-NEXT:    vmovaps 16(%rdi), %xmm1
-; AVX512F-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm2
-; AVX512F-SLOW-NEXT:    vextractps $2, %xmm1, %r10d
-; AVX512F-SLOW-NEXT:    vpinsrd $1, %r10d, %xmm0, %xmm3
-; AVX512F-SLOW-NEXT:    vextractps $3, %xmm1, %r10d
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
-; AVX512F-SLOW-NEXT:    vpinsrd $1, %r10d, %xmm1, %xmm1
-; AVX512F-SLOW-NEXT:    vpbroadcastd 8(%rdi), %xmm4
-; AVX512F-SLOW-NEXT:    vmovd %xmm2, %r10d
-; AVX512F-SLOW-NEXT:    vpinsrd $1, %r10d, %xmm4, %xmm4
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2,3]
-; AVX512F-SLOW-NEXT:    vmovddup {{.*#+}} xmm2 = [4,2,4,2]
-; AVX512F-SLOW-NEXT:    # xmm2 = mem[0,0]
-; AVX512F-SLOW-NEXT:    vmovaps 32(%rdi), %ymm5
-; AVX512F-SLOW-NEXT:    vblendps {{.*#+}} ymm5 = ymm5[0,1,2,3],mem[4,5,6,7]
-; AVX512F-SLOW-NEXT:    vpermps %ymm5, %ymm2, %ymm2
-; AVX512F-SLOW-NEXT:    vmovddup {{.*#+}} xmm6 = [5,3,5,3]
-; AVX512F-SLOW-NEXT:    # xmm6 = mem[0,0]
-; AVX512F-SLOW-NEXT:    vpermps %ymm5, %ymm6, %ymm5
-; AVX512F-SLOW-NEXT:    vmovq %xmm3, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovq %xmm1, (%rdx)
-; AVX512F-SLOW-NEXT:    vmovq %xmm4, (%rcx)
-; AVX512F-SLOW-NEXT:    vmovq %xmm0, (%r8)
-; AVX512F-SLOW-NEXT:    vmovlps %xmm2, (%r9)
-; AVX512F-SLOW-NEXT:    vmovlps %xmm5, (%rax)
-; AVX512F-SLOW-NEXT:    vzeroupper
-; AVX512F-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i32_stride6_vf2:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
+; AVX512-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-SLOW-NEXT:    vmovaps 16(%rdi), %xmm1
+; AVX512-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm2
+; AVX512-SLOW-NEXT:    vextractps $2, %xmm1, %r10d
+; AVX512-SLOW-NEXT:    vpinsrd $1, %r10d, %xmm0, %xmm3
+; AVX512-SLOW-NEXT:    vextractps $3, %xmm1, %r10d
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,1,1]
+; AVX512-SLOW-NEXT:    vpinsrd $1, %r10d, %xmm1, %xmm1
+; AVX512-SLOW-NEXT:    vpbroadcastd 8(%rdi), %xmm4
+; AVX512-SLOW-NEXT:    vmovd %xmm2, %r10d
+; AVX512-SLOW-NEXT:    vpinsrd $1, %r10d, %xmm4, %xmm4
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,3,3]
+; AVX512-SLOW-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2,3]
+; AVX512-SLOW-NEXT:    vmovddup {{.*#+}} xmm2 = [4,2,4,2]
+; AVX512-SLOW-NEXT:    # xmm2 = mem[0,0]
+; AVX512-SLOW-NEXT:    vmovaps 32(%rdi), %ymm5
+; AVX512-SLOW-NEXT:    vblendps {{.*#+}} ymm5 = ymm5[0,1,2,3],mem[4,5,6,7]
+; AVX512-SLOW-NEXT:    vpermps %ymm5, %ymm2, %ymm2
+; AVX512-SLOW-NEXT:    vmovddup {{.*#+}} xmm6 = [5,3,5,3]
+; AVX512-SLOW-NEXT:    # xmm6 = mem[0,0]
+; AVX512-SLOW-NEXT:    vpermps %ymm5, %ymm6, %ymm5
+; AVX512-SLOW-NEXT:    vmovq %xmm3, (%rsi)
+; AVX512-SLOW-NEXT:    vmovq %xmm1, (%rdx)
+; AVX512-SLOW-NEXT:    vmovq %xmm4, (%rcx)
+; AVX512-SLOW-NEXT:    vmovq %xmm0, (%r8)
+; AVX512-SLOW-NEXT:    vmovlps %xmm2, (%r9)
+; AVX512-SLOW-NEXT:    vmovlps %xmm5, (%rax)
+; AVX512-SLOW-NEXT:    vzeroupper
+; AVX512-SLOW-NEXT:    retq
 ;
-; AVX512F-FAST-LABEL: load_i32_stride6_vf2:
-; AVX512F-FAST:       # %bb.0:
-; AVX512F-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm0 = [0,6,0,6]
-; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %xmm1
-; AVX512F-FAST-NEXT:    vmovdqa 16(%rdi), %xmm2
-; AVX512F-FAST-NEXT:    vmovdqa 32(%rdi), %xmm3
-; AVX512F-FAST-NEXT:    vpermi2d %xmm2, %xmm1, %xmm0
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm4 = [1,7,1,7]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm2, %xmm1, %xmm4
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm2 = [2,4,2,4]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm3, %xmm1, %xmm2
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [3,5,3,5]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm3, %xmm1, %xmm5
-; AVX512F-FAST-NEXT:    vmovddup {{.*#+}} xmm1 = [4,2,4,2]
-; AVX512F-FAST-NEXT:    # xmm1 = mem[0,0]
-; AVX512F-FAST-NEXT:    vmovaps 32(%rdi), %ymm3
-; AVX512F-FAST-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],mem[4,5,6,7]
-; AVX512F-FAST-NEXT:    vpermps %ymm3, %ymm1, %ymm1
-; AVX512F-FAST-NEXT:    vmovddup {{.*#+}} xmm6 = [5,3,5,3]
-; AVX512F-FAST-NEXT:    # xmm6 = mem[0,0]
-; AVX512F-FAST-NEXT:    vpermps %ymm3, %ymm6, %ymm3
-; AVX512F-FAST-NEXT:    vmovq %xmm0, (%rsi)
-; AVX512F-FAST-NEXT:    vmovq %xmm4, (%rdx)
-; AVX512F-FAST-NEXT:    vmovq %xmm2, (%rcx)
-; AVX512F-FAST-NEXT:    vmovq %xmm5, (%r8)
-; AVX512F-FAST-NEXT:    vmovlps %xmm1, (%r9)
-; AVX512F-FAST-NEXT:    vmovlps %xmm3, (%rax)
-; AVX512F-FAST-NEXT:    vzeroupper
-; AVX512F-FAST-NEXT:    retq
+; AVX512-FAST-LABEL: load_i32_stride6_vf2:
+; AVX512-FAST:       # %bb.0:
+; AVX512-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %rax
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm0 = [0,6,0,6]
+; AVX512-FAST-NEXT:    vmovdqa (%rdi), %xmm1
+; AVX512-FAST-NEXT:    vmovdqa 16(%rdi), %xmm2
+; AVX512-FAST-NEXT:    vmovdqa 32(%rdi), %xmm3
+; AVX512-FAST-NEXT:    vpermi2d %xmm2, %xmm1, %xmm0
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm4 = [1,7,1,7]
+; AVX512-FAST-NEXT:    vpermi2d %xmm2, %xmm1, %xmm4
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm2 = [2,4,2,4]
+; AVX512-FAST-NEXT:    vpermi2d %xmm3, %xmm1, %xmm2
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [3,5,3,5]
+; AVX512-FAST-NEXT:    vpermi2d %xmm3, %xmm1, %xmm5
+; AVX512-FAST-NEXT:    vmovddup {{.*#+}} xmm1 = [4,2,4,2]
+; AVX512-FAST-NEXT:    # xmm1 = mem[0,0]
+; AVX512-FAST-NEXT:    vmovaps 32(%rdi), %ymm3
+; AVX512-FAST-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],mem[4,5,6,7]
+; AVX512-FAST-NEXT:    vpermps %ymm3, %ymm1, %ymm1
+; AVX512-FAST-NEXT:    vmovddup {{.*#+}} xmm6 = [5,3,5,3]
+; AVX512-FAST-NEXT:    # xmm6 = mem[0,0]
+; AVX512-FAST-NEXT:    vpermps %ymm3, %ymm6, %ymm3
+; AVX512-FAST-NEXT:    vmovq %xmm0, (%rsi)
+; AVX512-FAST-NEXT:    vmovq %xmm4, (%rdx)
+; AVX512-FAST-NEXT:    vmovq %xmm2, (%rcx)
+; AVX512-FAST-NEXT:    vmovq %xmm5, (%r8)
+; AVX512-FAST-NEXT:    vmovlps %xmm1, (%r9)
+; AVX512-FAST-NEXT:    vmovlps %xmm3, (%rax)
+; AVX512-FAST-NEXT:    vzeroupper
+; AVX512-FAST-NEXT:    retq
 ;
 ; AVX512BW-SLOW-LABEL: load_i32_stride6_vf2:
 ; AVX512BW-SLOW:       # %bb.0:
@@ -198,37 +198,6 @@ define void @load_i32_stride6_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX512BW-SLOW-NEXT:    vmovlps %xmm5, (%rax)
 ; AVX512BW-SLOW-NEXT:    vzeroupper
 ; AVX512BW-SLOW-NEXT:    retq
-;
-; AVX512BW-FAST-LABEL: load_i32_stride6_vf2:
-; AVX512BW-FAST:       # %bb.0:
-; AVX512BW-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm0 = [0,6,0,6]
-; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %xmm1
-; AVX512BW-FAST-NEXT:    vmovdqa 16(%rdi), %xmm2
-; AVX512BW-FAST-NEXT:    vmovdqa 32(%rdi), %xmm3
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm2, %xmm1, %xmm0
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm4 = [1,7,1,7]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm2, %xmm1, %xmm4
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm2 = [2,4,2,4]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm3, %xmm1, %xmm2
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [3,5,3,5]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm3, %xmm1, %xmm5
-; AVX512BW-FAST-NEXT:    vmovddup {{.*#+}} xmm1 = [4,2,4,2]
-; AVX512BW-FAST-NEXT:    # xmm1 = mem[0,0]
-; AVX512BW-FAST-NEXT:    vmovaps 32(%rdi), %ymm3
-; AVX512BW-FAST-NEXT:    vblendps {{.*#+}} ymm3 = ymm3[0,1,2,3],mem[4,5,6,7]
-; AVX512BW-FAST-NEXT:    vpermps %ymm3, %ymm1, %ymm1
-; AVX512BW-FAST-NEXT:    vmovddup {{.*#+}} xmm6 = [5,3,5,3]
-; AVX512BW-FAST-NEXT:    # xmm6 = mem[0,0]
-; AVX512BW-FAST-NEXT:    vpermps %ymm3, %ymm6, %ymm3
-; AVX512BW-FAST-NEXT:    vmovq %xmm0, (%rsi)
-; AVX512BW-FAST-NEXT:    vmovq %xmm4, (%rdx)
-; AVX512BW-FAST-NEXT:    vmovq %xmm2, (%rcx)
-; AVX512BW-FAST-NEXT:    vmovq %xmm5, (%r8)
-; AVX512BW-FAST-NEXT:    vmovlps %xmm1, (%r9)
-; AVX512BW-FAST-NEXT:    vmovlps %xmm3, (%rax)
-; AVX512BW-FAST-NEXT:    vzeroupper
-; AVX512BW-FAST-NEXT:    retq
   %wide.vec = load <12 x i32>, ptr %in.vec, align 64
   %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>
@@ -10329,14 +10298,17 @@ define void @load_i32_stride6_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX: {{.*}}
 ; AVX1: {{.*}}
 ; AVX2: {{.*}}
+; AVX512BW-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
 ; AVX512DQ-FAST: {{.*}}
 ; AVX512DQ-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}
 ; AVX512DQBW-SLOW: {{.*}}
+; AVX512F-FAST: {{.*}}
 ; AVX512F-ONLY-FAST: {{.*}}
 ; AVX512F-ONLY-SLOW: {{.*}}
+; AVX512F-SLOW: {{.*}}
 ; FALLBACK0: {{.*}}
 ; FALLBACK1: {{.*}}
 ; FALLBACK10: {{.*}}

diff --git a/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-7.ll b/llvm/test/CodeGen/X86/vector-interleaved-load-i32-stride-7.ll
@@ -4,14 +4,14 @@
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-SLOW,FALLBACK2
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST,FALLBACK3
 ; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX1,AVX2,AVX2-ONLY,AVX2-FAST-PERLANE,FALLBACK4
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512F-ONLY-SLOW,FALLBACK5
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512F-ONLY-FAST,FALLBACK6
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-SLOW,AVX512F-SLOW,AVX512DQ-SLOW,FALLBACK7
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512F,AVX512-FAST,AVX512F-FAST,AVX512DQ-FAST,FALLBACK8
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512BW-ONLY-SLOW,FALLBACK9
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512BW-ONLY-FAST,FALLBACK10
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-SLOW,AVX512BW-SLOW,AVX512DQBW-SLOW,FALLBACK11
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512dq,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX,AVX2,AVX512,AVX512BW,AVX512-FAST,AVX512BW-FAST,AVX512DQBW-FAST,FALLBACK12
 
 ; These patterns are produced by LoopVectorizer for interleaved loads.
 
@@ -120,143 +120,74 @@ define void @load_i32_stride7_vf2(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, ptr
 ; AVX2-ONLY-NEXT:    vzeroupper
 ; AVX2-ONLY-NEXT:    retq
 ;
-; AVX512F-SLOW-LABEL: load_i32_stride7_vf2:
-; AVX512F-SLOW:       # %bb.0:
-; AVX512F-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512F-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %r10
-; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm1
-; AVX512F-SLOW-NEXT:    vpinsrd $1, 28(%rdi), %xmm0, %xmm2
-; AVX512F-SLOW-NEXT:    vmovd %xmm1, %r11d
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
-; AVX512F-SLOW-NEXT:    vpinsrd $1, %r11d, %xmm3, %xmm3
-; AVX512F-SLOW-NEXT:    vpbroadcastd 8(%rdi), %xmm4
-; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm4 = xmm4[0],xmm1[1],xmm4[2,3]
-; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[3]
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,2,2,3]
-; AVX512F-SLOW-NEXT:    vpbroadcastq {{.*#+}} xmm1 = [4,11,4,11]
-; AVX512F-SLOW-NEXT:    vmovdqa 32(%rdi), %ymm5
-; AVX512F-SLOW-NEXT:    vmovdqa (%rdi), %ymm6
-; AVX512F-SLOW-NEXT:    vpermi2d %ymm5, %ymm6, %ymm1
-; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} ymm7 = ymm5[0],ymm6[1],ymm5[2,3,4],ymm6[5],ymm5[6,7]
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} ymm7 = ymm7[1,0,2,3,5,4,6,7]
-; AVX512F-SLOW-NEXT:    vextracti128 $1, %ymm7, %xmm7
-; AVX512F-SLOW-NEXT:    vpshufd {{.*#+}} ymm6 = ymm6[2,3,2,3,6,7,6,7]
-; AVX512F-SLOW-NEXT:    vpblendd {{.*#+}} ymm5 = ymm6[0],ymm5[1],ymm6[2,3,4],ymm5[5],ymm6[6,7]
-; AVX512F-SLOW-NEXT:    vextracti128 $1, %ymm5, %xmm5
-; AVX512F-SLOW-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512F-SLOW-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512F-SLOW-NEXT:    vmovq %xmm4, (%rcx)
-; AVX512F-SLOW-NEXT:    vmovq %xmm0, (%r8)
-; AVX512F-SLOW-NEXT:    vmovq %xmm1, (%r9)
-; AVX512F-SLOW-NEXT:    vmovq %xmm7, (%r10)
-; AVX512F-SLOW-NEXT:    vmovq %xmm5, (%rax)
-; AVX512F-SLOW-NEXT:    vzeroupper
-; AVX512F-SLOW-NEXT:    retq
+; AVX512-SLOW-LABEL: load_i32_stride7_vf2:
+; AVX512-SLOW:       # %bb.0:
+; AVX512-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
+; AVX512-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %r10
+; AVX512-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm1
+; AVX512-SLOW-NEXT:    vpinsrd $1, 28(%rdi), %xmm0, %xmm2
+; AVX512-SLOW-NEXT:    vmovd %xmm1, %r11d
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
+; AVX512-SLOW-NEXT:    vpinsrd $1, %r11d, %xmm3, %xmm3
+; AVX512-SLOW-NEXT:    vpbroadcastd 8(%rdi), %xmm4
+; AVX512-SLOW-NEXT:    vpblendd {{.*#+}} xmm4 = xmm4[0],xmm1[1],xmm4[2,3]
+; AVX512-SLOW-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[3]
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,2,2,3]
+; AVX512-SLOW-NEXT:    vpbroadcastq {{.*#+}} xmm1 = [4,11,4,11]
+; AVX512-SLOW-NEXT:    vmovdqa 32(%rdi), %ymm5
+; AVX512-SLOW-NEXT:    vmovdqa (%rdi), %ymm6
+; AVX512-SLOW-NEXT:    vpermi2d %ymm5, %ymm6, %ymm1
+; AVX512-SLOW-NEXT:    vpblendd {{.*#+}} ymm7 = ymm5[0],ymm6[1],ymm5[2,3,4],ymm6[5],ymm5[6,7]
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} ymm7 = ymm7[1,0,2,3,5,4,6,7]
+; AVX512-SLOW-NEXT:    vextracti128 $1, %ymm7, %xmm7
+; AVX512-SLOW-NEXT:    vpshufd {{.*#+}} ymm6 = ymm6[2,3,2,3,6,7,6,7]
+; AVX512-SLOW-NEXT:    vpblendd {{.*#+}} ymm5 = ymm6[0],ymm5[1],ymm6[2,3,4],ymm5[5],ymm6[6,7]
+; AVX512-SLOW-NEXT:    vextracti128 $1, %ymm5, %xmm5
+; AVX512-SLOW-NEXT:    vmovq %xmm2, (%rsi)
+; AVX512-SLOW-NEXT:    vmovq %xmm3, (%rdx)
+; AVX512-SLOW-NEXT:    vmovq %xmm4, (%rcx)
+; AVX512-SLOW-NEXT:    vmovq %xmm0, (%r8)
+; AVX512-SLOW-NEXT:    vmovq %xmm1, (%r9)
+; AVX512-SLOW-NEXT:    vmovq %xmm7, (%r10)
+; AVX512-SLOW-NEXT:    vmovq %xmm5, (%rax)
+; AVX512-SLOW-NEXT:    vzeroupper
+; AVX512-SLOW-NEXT:    retq
 ;
-; AVX512F-FAST-LABEL: load_i32_stride7_vf2:
-; AVX512F-FAST:       # %bb.0:
-; AVX512F-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512F-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %r10
-; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512F-FAST-NEXT:    vmovdqa 32(%rdi), %xmm1
-; AVX512F-FAST-NEXT:    vpinsrd $1, 28(%rdi), %xmm0, %xmm2
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [1,4,1,4]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm3
-; AVX512F-FAST-NEXT:    vpbroadcastd 8(%rdi), %xmm4
-; AVX512F-FAST-NEXT:    vpblendd {{.*#+}} xmm4 = xmm4[0],xmm1[1],xmm4[2,3]
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [7,2,7,2]
-; AVX512F-FAST-NEXT:    vpermi2d %xmm0, %xmm1, %xmm5
-; AVX512F-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm0 = [4,11,4,11]
-; AVX512F-FAST-NEXT:    vmovdqa 32(%rdi), %ymm1
-; AVX512F-FAST-NEXT:    vmovdqa (%rdi), %ymm6
-; AVX512F-FAST-NEXT:    vpermi2d %ymm1, %ymm6, %ymm0
-; AVX512F-FAST-NEXT:    vbroadcasti128 {{.*#+}} ymm7 = [13,4,6,7,13,4,6,7]
-; AVX512F-FAST-NEXT:    # ymm7 = mem[0,1,0,1]
-; AVX512F-FAST-NEXT:    vpermi2d %ymm6, %ymm1, %ymm7
-; AVX512F-FAST-NEXT:    vextracti128 $1, %ymm7, %xmm7
-; AVX512F-FAST-NEXT:    vbroadcasti128 {{.*#+}} ymm8 = [6,13,6,7,6,13,6,7]
-; AVX512F-FAST-NEXT:    # ymm8 = mem[0,1,0,1]
-; AVX512F-FAST-NEXT:    vpermi2d %ymm1, %ymm6, %ymm8
-; AVX512F-FAST-NEXT:    vextracti128 $1, %ymm8, %xmm1
-; AVX512F-FAST-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512F-FAST-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512F-FAST-NEXT:    vmovq %xmm4, (%rcx)
-; AVX512F-FAST-NEXT:    vmovq %xmm5, (%r8)
-; AVX512F-FAST-NEXT:    vmovq %xmm0, (%r9)
-; AVX512F-FAST-NEXT:    vmovq %xmm7, (%r10)
-; AVX512F-FAST-NEXT:    vmovq %xmm1, (%rax)
-; AVX512F-FAST-NEXT:    vzeroupper
-; AVX512F-FAST-NEXT:    retq
-;
-; AVX512BW-SLOW-LABEL: load_i32_stride7_vf2:
-; AVX512BW-SLOW:       # %bb.0:
-; AVX512BW-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512BW-SLOW-NEXT:    movq {{[0-9]+}}(%rsp), %r10
-; AVX512BW-SLOW-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-SLOW-NEXT:    vmovdqa 32(%rdi), %xmm1
-; AVX512BW-SLOW-NEXT:    vpinsrd $1, 28(%rdi), %xmm0, %xmm2
-; AVX512BW-SLOW-NEXT:    vmovd %xmm1, %r11d
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,1,1]
-; AVX512BW-SLOW-NEXT:    vpinsrd $1, %r11d, %xmm3, %xmm3
-; AVX512BW-SLOW-NEXT:    vpbroadcastd 8(%rdi), %xmm4
-; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm4 = xmm4[0],xmm1[1],xmm4[2,3]
-; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[3]
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,2,2,3]
-; AVX512BW-SLOW-NEXT:    vpbroadcastq {{.*#+}} xmm1 = [4,11,4,11]
-; AVX512BW-SLOW-NEXT:    vmovdqa 32(%rdi), %ymm5
-; AVX512BW-SLOW-NEXT:    vmovdqa (%rdi), %ymm6
-; AVX512BW-SLOW-NEXT:    vpermi2d %ymm5, %ymm6, %ymm1
-; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} ymm7 = ymm5[0],ymm6[1],ymm5[2,3,4],ymm6[5],ymm5[6,7]
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} ymm7 = ymm7[1,0,2,3,5,4,6,7]
-; AVX512BW-SLOW-NEXT:    vextracti128 $1, %ymm7, %xmm7
-; AVX512BW-SLOW-NEXT:    vpshufd {{.*#+}} ymm6 = ymm6[2,3,2,3,6,7,6,7]
-; AVX512BW-SLOW-NEXT:    vpblendd {{.*#+}} ymm5 = ymm6[0],ymm5[1],ymm6[2,3,4],ymm5[5],ymm6[6,7]
-; AVX512BW-SLOW-NEXT:    vextracti128 $1, %ymm5, %xmm5
-; AVX512BW-SLOW-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm4, (%rcx)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm0, (%r8)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm1, (%r9)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm7, (%r10)
-; AVX512BW-SLOW-NEXT:    vmovq %xmm5, (%rax)
-; AVX512BW-SLOW-NEXT:    vzeroupper
-; AVX512BW-SLOW-NEXT:    retq
-;
-; AVX512BW-FAST-LABEL: load_i32_stride7_vf2:
-; AVX512BW-FAST:       # %bb.0:
-; AVX512BW-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %rax
-; AVX512BW-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %r10
-; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %xmm0
-; AVX512BW-FAST-NEXT:    vmovdqa 32(%rdi), %xmm1
-; AVX512BW-FAST-NEXT:    vpinsrd $1, 28(%rdi), %xmm0, %xmm2
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [1,4,1,4]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm3
-; AVX512BW-FAST-NEXT:    vpbroadcastd 8(%rdi), %xmm4
-; AVX512BW-FAST-NEXT:    vpblendd {{.*#+}} xmm4 = xmm4[0],xmm1[1],xmm4[2,3]
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [7,2,7,2]
-; AVX512BW-FAST-NEXT:    vpermi2d %xmm0, %xmm1, %xmm5
-; AVX512BW-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm0 = [4,11,4,11]
-; AVX512BW-FAST-NEXT:    vmovdqa 32(%rdi), %ymm1
-; AVX512BW-FAST-NEXT:    vmovdqa (%rdi), %ymm6
-; AVX512BW-FAST-NEXT:    vpermi2d %ymm1, %ymm6, %ymm0
-; AVX512BW-FAST-NEXT:    vbroadcasti128 {{.*#+}} ymm7 = [13,4,6,7,13,4,6,7]
-; AVX512BW-FAST-NEXT:    # ymm7 = mem[0,1,0,1]
-; AVX512BW-FAST-NEXT:    vpermi2d %ymm6, %ymm1, %ymm7
-; AVX512BW-FAST-NEXT:    vextracti128 $1, %ymm7, %xmm7
-; AVX512BW-FAST-NEXT:    vbroadcasti128 {{.*#+}} ymm8 = [6,13,6,7,6,13,6,7]
-; AVX512BW-FAST-NEXT:    # ymm8 = mem[0,1,0,1]
-; AVX512BW-FAST-NEXT:    vpermi2d %ymm1, %ymm6, %ymm8
-; AVX512BW-FAST-NEXT:    vextracti128 $1, %ymm8, %xmm1
-; AVX512BW-FAST-NEXT:    vmovq %xmm2, (%rsi)
-; AVX512BW-FAST-NEXT:    vmovq %xmm3, (%rdx)
-; AVX512BW-FAST-NEXT:    vmovq %xmm4, (%rcx)
-; AVX512BW-FAST-NEXT:    vmovq %xmm5, (%r8)
-; AVX512BW-FAST-NEXT:    vmovq %xmm0, (%r9)
-; AVX512BW-FAST-NEXT:    vmovq %xmm7, (%r10)
-; AVX512BW-FAST-NEXT:    vmovq %xmm1, (%rax)
-; AVX512BW-FAST-NEXT:    vzeroupper
-; AVX512BW-FAST-NEXT:    retq
+; AVX512-FAST-LABEL: load_i32_stride7_vf2:
+; AVX512-FAST:       # %bb.0:
+; AVX512-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %rax
+; AVX512-FAST-NEXT:    movq {{[0-9]+}}(%rsp), %r10
+; AVX512-FAST-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX512-FAST-NEXT:    vmovdqa 32(%rdi), %xmm1
+; AVX512-FAST-NEXT:    vpinsrd $1, 28(%rdi), %xmm0, %xmm2
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm3 = [1,4,1,4]
+; AVX512-FAST-NEXT:    vpermi2d %xmm1, %xmm0, %xmm3
+; AVX512-FAST-NEXT:    vpbroadcastd 8(%rdi), %xmm4
+; AVX512-FAST-NEXT:    vpblendd {{.*#+}} xmm4 = xmm4[0],xmm1[1],xmm4[2,3]
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm5 = [7,2,7,2]
+; AVX512-FAST-NEXT:    vpermi2d %xmm0, %xmm1, %xmm5
+; AVX512-FAST-NEXT:    vpbroadcastq {{.*#+}} xmm0 = [4,11,4,11]
+; AVX512-FAST-NEXT:    vmovdqa 32(%rdi), %ymm1
+; AVX512-FAST-NEXT:    vmovdqa (%rdi), %ymm6
+; AVX512-FAST-NEXT:    vpermi2d %ymm1, %ymm6, %ymm0
+; AVX512-FAST-NEXT:    vbroadcasti128 {{.*#+}} ymm7 = [13,4,6,7,13,4,6,7]
+; AVX512-FAST-NEXT:    # ymm7 = mem[0,1,0,1]
+; AVX512-FAST-NEXT:    vpermi2d %ymm6, %ymm1, %ymm7
+; AVX512-FAST-NEXT:    vextracti128 $1, %ymm7, %xmm7
+; AVX512-FAST-NEXT:    vbroadcasti128 {{.*#+}} ymm8 = [6,13,6,7,6,13,6,7]
+; AVX512-FAST-NEXT:    # ymm8 = mem[0,1,0,1]
+; AVX512-FAST-NEXT:    vpermi2d %ymm1, %ymm6, %ymm8
+; AVX512-FAST-NEXT:    vextracti128 $1, %ymm8, %xmm1
+; AVX512-FAST-NEXT:    vmovq %xmm2, (%rsi)
+; AVX512-FAST-NEXT:    vmovq %xmm3, (%rdx)
+; AVX512-FAST-NEXT:    vmovq %xmm4, (%rcx)
+; AVX512-FAST-NEXT:    vmovq %xmm5, (%r8)
+; AVX512-FAST-NEXT:    vmovq %xmm0, (%r9)
+; AVX512-FAST-NEXT:    vmovq %xmm7, (%r10)
+; AVX512-FAST-NEXT:    vmovq %xmm1, (%rax)
+; AVX512-FAST-NEXT:    vzeroupper
+; AVX512-FAST-NEXT:    retq
   %wide.vec = load <14 x i32>, ptr %in.vec, align 64
   %strided.vec0 = shufflevector <14 x i32> %wide.vec, <14 x i32> poison, <2 x i32> <i32 0, i32 7>
   %strided.vec1 = shufflevector <14 x i32> %wide.vec, <14 x i32> poison, <2 x i32> <i32 1, i32 8>
@@ -12700,14 +12631,18 @@ define void @load_i32_stride7_vf64(ptr %in.vec, ptr %out.vec0, ptr %out.vec1, pt
 ; AVX: {{.*}}
 ; AVX1: {{.*}}
 ; AVX2: {{.*}}
+; AVX512BW-FAST: {{.*}}
 ; AVX512BW-ONLY-FAST: {{.*}}
 ; AVX512BW-ONLY-SLOW: {{.*}}
+; AVX512BW-SLOW: {{.*}}
 ; AVX512DQ-FAST: {{.*}}
 ; AVX512DQ-SLOW: {{.*}}
 ; AVX512DQBW-FAST: {{.*}}
 ; AVX512DQBW-SLOW: {{.*}}
+; AVX512F-FAST: {{.*}}
 ; AVX512F-ONLY-FAST: {{.*}}
 ; AVX512F-ONLY-SLOW: {{.*}}
+; AVX512F-SLOW: {{.*}}
 ; FALLBACK0: {{.*}}
 ; FALLBACK1: {{.*}}
 ; FALLBACK10: {{.*}}