Fix blend mask by switch the side of the operand since Blend node uses opposite mask...

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@292066 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 989039f3ee438638464b7ade4fc9358f0394bc3d..31e10dcc2a180a97b4a45c46e1dd8f22805f701d 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -8436,7 +8436,7 @@ static SDValue lowerVectorShuffleAsBlend(const SDLoc &DL, MVT VT, SDValue V1,
       MVT IntegerType =
           MVT::getIntegerVT(std::max((int)VT.getVectorNumElements(), 8));
       SDValue MaskNode = DAG.getConstant(BlendMask, DL, IntegerType);
-      return getVectorMaskingNode(V1, MaskNode, V2, Subtarget, DAG);
+      return getVectorMaskingNode(V2, MaskNode, V1, Subtarget, DAG);
     }
 
     // Attempt to lower to a bitmask if we can. VPAND is faster than VPBLENDVB.
@@ -8485,7 +8485,7 @@ static SDValue lowerVectorShuffleAsBlend(const SDLoc &DL, MVT VT, SDValue V1,
     MVT IntegerType =
         MVT::getIntegerVT(std::max((int)VT.getVectorNumElements(), 8));
     SDValue MaskNode = DAG.getConstant(BlendMask, DL, IntegerType);
-    return getVectorMaskingNode(V1, MaskNode, V2, Subtarget, DAG);
+    return getVectorMaskingNode(V2, MaskNode, V1, Subtarget, DAG);
   }
   default:
     llvm_unreachable("Not a supported integer vector type!");

diff --git a/test/CodeGen/X86/avx512-mask-op.ll b/test/CodeGen/X86/avx512-mask-op.ll
index a2afe843ec811f9404d150449acbfc0a37f3028e..89bd1980e5231dfff057f5c40f3f4a9b33809caa 100644 (file)
--- a/test/CodeGen/X86/avx512-mask-op.ll
+++ b/test/CodeGen/X86/avx512-mask-op.ll
@@ -542,7 +542,7 @@ define <64 x i8> @test16(i64 %x) {
 ; SKX-NEXT:    vpmovm2b %k0, %zmm1
 ; SKX-NEXT:    movl $32, %eax
 ; SKX-NEXT:    kmovd %eax, %k1
-; SKX-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k1}
+; SKX-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1}
 ; SKX-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
 ; SKX-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
 ; SKX-NEXT:    vpmovb2m %zmm0, %k0
@@ -610,7 +610,7 @@ define <64 x i8> @test17(i64 %x, i32 %y, i32 %z) {
 ; SKX-NEXT:    vpmovm2b %k0, %zmm1
 ; SKX-NEXT:    movl $32, %eax
 ; SKX-NEXT:    kmovd %eax, %k1
-; SKX-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k1}
+; SKX-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1}
 ; SKX-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
 ; SKX-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
 ; SKX-NEXT:    vpmovb2m %zmm0, %k0

diff --git a/test/CodeGen/X86/merge-consecutive-loads-512.ll b/test/CodeGen/X86/merge-consecutive-loads-512.ll
index eced69fcb859d150e1720892a34e20a6208f7d91..0111e341c45927e595c0aa6bafc3cd05a62672a3 100644 (file)
--- a/test/CodeGen/X86/merge-consecutive-loads-512.ll
+++ b/test/CodeGen/X86/merge-consecutive-loads-512.ll
@@ -138,17 +138,19 @@ define <8 x double> @merge_8f64_f64_12zzuuzz(double* %ptr) nounwind uwtable noin
 define <8 x double> @merge_8f64_f64_1u3u5zu8(double* %ptr) nounwind uwtable noinline ssp {
 ; ALL-LABEL: merge_8f64_f64_1u3u5zu8:
 ; ALL:       # BB#0:
-; ALL-NEXT:    movb    $32, %al
-; ALL-NEXT:    kmovw   %eax, %k1
-; ALL-NEXT:    vmovupd 8(%rdi), %zmm0 {%k1} {z}
+; ALL-NEXT:    movb $32, %al
+; ALL-NEXT:    kmovw %eax, %k0
+; ALL-NEXT:    knotw %k0, %k1
+; ALL-NEXT:    vmovupd 8(%rdi), %zmm0 {%k1} {z}
 ; ALL-NEXT:    retq
 ;
 ; X32-AVX512F-LABEL: merge_8f64_f64_1u3u5zu8:
 ; X32-AVX512F:       # BB#0:
-; X32-AVX512F-NEXT:    movl    4(%esp), %eax
-; X32-AVX512F-NEXT:    movb    $32, %cl
-; X32-AVX512F-NEXT:    kmovw   %ecx, %k1
-; X32-AVX512F-NEXT:    vmovupd 8(%eax), %zmm0 {%k1} {z}
+; X32-AVX512F-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-AVX512F-NEXT:    movb $32, %cl
+; X32-AVX512F-NEXT:    kmovw %ecx, %k0
+; X32-AVX512F-NEXT:    knotw %k0, %k1
+; X32-AVX512F-NEXT:    vmovupd 8(%eax), %zmm0 {%k1} {z}
 ; X32-AVX512F-NEXT:    retl
   %ptr0 = getelementptr inbounds double, double* %ptr, i64 1
   %ptr2 = getelementptr inbounds double, double* %ptr, i64 3
@@ -223,17 +225,19 @@ define <8 x i64> @merge_8i64_i64_56zz9uzz(i64* %ptr) nounwind uwtable noinline s
 define <8 x i64> @merge_8i64_i64_1u3u5zu8(i64* %ptr) nounwind uwtable noinline ssp {
 ; ALL-LABEL: merge_8i64_i64_1u3u5zu8:
 ; ALL:       # BB#0:
-; ALL-NEXT:    movb    $32, %al
-; ALL-NEXT:    kmovw   %eax, %k1
-; ALL-NEXT:    vmovdqu64       8(%rdi), %zmm0 {%k1} {z}
+; ALL-NEXT:    movb $32, %al
+; ALL-NEXT:    kmovw %eax, %k0
+; ALL-NEXT:    knotw %k0, %k1
+; ALL-NEXT:    vmovdqu64 8(%rdi), %zmm0 {%k1} {z}
 ; ALL-NEXT:    retq
 ;
 ; X32-AVX512F-LABEL: merge_8i64_i64_1u3u5zu8:
 ; X32-AVX512F:       # BB#0:
-; X32-AVX512F-NEXT:    movl    4(%esp), %eax
-; X32-AVX512F-NEXT:    movb    $32, %cl
-; X32-AVX512F-NEXT:    kmovw   %ecx, %k1
-; X32-AVX512F-NEXT:    vmovdqu64       8(%eax), %zmm0 {%k1} {z}
+; X32-AVX512F-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-AVX512F-NEXT:    movb $32, %cl
+; X32-AVX512F-NEXT:    kmovw %ecx, %k0
+; X32-AVX512F-NEXT:    knotw %k0, %k1
+; X32-AVX512F-NEXT:    vmovdqu64 8(%eax), %zmm0 {%k1} {z}
 ; X32-AVX512F-NEXT:    retl
   %ptr0 = getelementptr inbounds i64, i64* %ptr, i64 1
   %ptr2 = getelementptr inbounds i64, i64* %ptr, i64 3
@@ -444,17 +448,19 @@ define <16 x i32> @merge_16i32_i32_0uu3uuuuuuuuCuEF(i32* %ptr) nounwind uwtable
 define <16 x i32> @merge_16i32_i32_0uu3zzuuuuuzCuEF(i32* %ptr) nounwind uwtable noinline ssp {
 ; ALL-LABEL: merge_16i32_i32_0uu3zzuuuuuzCuEF:
 ; ALL:       # BB#0:
-; ALL-NEXT:    movw    $8240, %ax              # imm = 0x2030
-; ALL-NEXT:    kmovw   %eax, %k1
-; ALL-NEXT:    vmovdqu32       (%rdi), %zmm0 {%k1} {z}
+; ALL-NEXT:    movw $8240, %ax # imm = 0x2030
+; ALL-NEXT:    kmovw %eax, %k0
+; ALL-NEXT:    knotw %k0, %k1
+; ALL-NEXT:    vmovdqu32 (%rdi), %zmm0 {%k1} {z}
 ; ALL-NEXT:    retq
 ;
 ; X32-AVX512F-LABEL: merge_16i32_i32_0uu3zzuuuuuzCuEF:
 ; X32-AVX512F:       # BB#0:
-; X32-AVX512F-NEXT:    movl    4(%esp), %eax
-; X32-AVX512F-NEXT:    movw    $8240, %cx              # imm = 0x2030
-; X32-AVX512F-NEXT:    kmovw   %ecx, %k1
-; X32-AVX512F-NEXT:    vmovdqu32       (%eax), %zmm0 {%k1} {z}
+; X32-AVX512F-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-AVX512F-NEXT:    movw $8240, %cx # imm = 0x2030
+; X32-AVX512F-NEXT:    kmovw %ecx, %k0
+; X32-AVX512F-NEXT:    knotw %k0, %k1
+; X32-AVX512F-NEXT:    vmovdqu32 (%eax), %zmm0 {%k1} {z}
 ; X32-AVX512F-NEXT:    retl
   %ptr0 = getelementptr inbounds i32, i32* %ptr, i64 0
   %ptr3 = getelementptr inbounds i32, i32* %ptr, i64 3

diff --git a/test/CodeGen/X86/sse3-avx-addsub.ll b/test/CodeGen/X86/sse3-avx-addsub.ll
index e7fab2e8ec5ffacdf505548063be744fdb4ce04a..0e0cf4852568d7681a4231019ef0deea0826a6ef 100644 (file)
--- a/test/CodeGen/X86/sse3-avx-addsub.ll
+++ b/test/CodeGen/X86/sse3-avx-addsub.ll
@@ -119,10 +119,10 @@ define <16 x float> @test5(<16 x float> %A, <16 x float> %B) {
 ;
 ; AVX512-LABEL: test5:
 ; AVX512:       # BB#0:
-; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm2
+; AVX512-NEXT:    vsubps %zmm1, %zmm0, %zmm2
 ; AVX512-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; AVX512-NEXT:    kmovw %eax, %k1
-; AVX512-NEXT:    vsubps %zmm1, %zmm0, %zmm2 {%k1}
+; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm2 {%k1}
 ; AVX512-NEXT:    vmovaps %zmm2, %zmm0
 ; AVX512-NEXT:    retq
   %add = fadd <16 x float> %A, %B

diff --git a/test/CodeGen/X86/vector-shuffle-128-v16.ll b/test/CodeGen/X86/vector-shuffle-128-v16.ll
index b5290414b98dc5e8fd4dfc8f042366de7cb9d990..3e19da2cd071ad53c17bc814ece3adf162ffacbd 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-128-v16.ll
+++ b/test/CodeGen/X86/vector-shuffle-128-v16.ll
@@ -425,7 +425,7 @@ define <16 x i8> @shuffle_v16i8_00_17_02_19_04_21_06_23_08_25_10_27_12_29_14_31(
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vpblendmb %xmm0, %xmm1, %xmm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
   ret <16 x i8> %shuffle
@@ -465,7 +465,7 @@ define <16 x i8> @shuffle_v16i8_00_01_02_19_04_05_06_23_08_09_10_27_12_13_14_31(
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movw $-30584, %ax # imm = 0x8888
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vpblendmb %xmm0, %xmm1, %xmm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 19, i32 4, i32 5, i32 6, i32 23, i32 8, i32 9, i32 10, i32 27, i32 12, i32 13, i32 14, i32 31>
   ret <16 x i8> %shuffle
@@ -524,7 +524,7 @@ define <16 x i8> @shuffle_v16i8_00_01_02_03_20_05_06_23_08_09_10_11_28_13_14_31(
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movw $-28528, %ax # imm = 0x9090
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vpblendmb %xmm0, %xmm1, %xmm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 20, i32 5, i32 6, i32 23, i32 8, i32 9, i32 10, i32 11, i32 28, i32 13, i32 14, i32 31>
   ret <16 x i8> %shuffle
@@ -565,7 +565,7 @@ define <16 x i8> @shuffle_v16i8_16_17_18_19_04_05_06_07_24_25_10_11_28_13_30_15(
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movw $-21264, %ax # imm = 0xACF0
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
+; AVX512VL-NEXT:    vpblendmb %xmm0, %xmm1, %xmm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 24, i32 25, i32 10, i32 11, i32 28, i32 13, i32 30, i32 15>
   ret <16 x i8> %shuffle

diff --git a/test/CodeGen/X86/vector-shuffle-256-v16.ll b/test/CodeGen/X86/vector-shuffle-256-v16.ll
index a1cf97d787abee6212bf46fe9cb0b19357c363fb..682a34d3cdb772aadbdb792a365f4f31b57eb61d 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-256-v16.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v16.ll
@@ -719,7 +719,7 @@ define <16 x i16> @shuffle_v16i16_00_01_02_03_04_05_06_07_08_09_10_11_12_13_14_3
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movw $-32768, %ax # imm = 0x8000
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vpblendmw %ymm0, %ymm1, %ymm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu16 %ymm1, %ymm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 31>
   ret <16 x i16> %shuffle
@@ -744,7 +744,7 @@ define <16 x i16> @shuffle_v16i16_16_01_02_03_04_05_06_07_08_09_10_11_12_13_14_1
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movw $1, %ax
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vpblendmw %ymm0, %ymm1, %ymm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu16 %ymm1, %ymm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 16, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
   ret <16 x i16> %shuffle
@@ -769,7 +769,7 @@ define <16 x i16> @shuffle_v16i16_00_17_02_19_04_21_06_23_24_09_26_11_28_13_30_1
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movw $21930, %ax # imm = 0x55AA
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vpblendmw %ymm0, %ymm1, %ymm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu16 %ymm1, %ymm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
   ret <16 x i16> %shuffle
@@ -794,7 +794,7 @@ define <16 x i16> @shuffle_v16i16_16_01_18_03_20_05_22_07_08_25_10_27_12_29_14_3
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movw $-21931, %ax # imm = 0xAA55
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vpblendmw %ymm0, %ymm1, %ymm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu16 %ymm1, %ymm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
   ret <16 x i16> %shuffle

diff --git a/test/CodeGen/X86/vector-shuffle-256-v32.ll b/test/CodeGen/X86/vector-shuffle-256-v32.ll
index 390ab16699d5bb28328a46c8de2b39b28991f0d3..8b0e25ce43f6d484cb8ac68577e3d62df3df61f0 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-256-v32.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v32.ll
@@ -317,11 +317,13 @@ define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_16_
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_16_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
-; AVX512VL-NEXT:    vpbroadcastb %xmm0, %xmm0
 ; AVX512VL-NEXT:    vpxor %ymm2, %ymm2, %ymm2
+; AVX512VL-NEXT:    vpshufb %ymm2, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpbroadcastb %xmm0, %xmm0
 ; AVX512VL-NEXT:    movl $32767, %eax # imm = 0x7FFF
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vpshufb %ymm2, %ymm1, %ymm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu8 %ymm0, %ymm1 {%k1}
+; AVX512VL-NEXT:    vmovdqa %ymm1, %ymm0
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 16, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
@@ -351,8 +353,8 @@ define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_17_00_
 ; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    movl $1, %eax
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k1}
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
+; AVX512VL-NEXT:    vmovdqu8 %ymm0, %ymm1 {%k1}
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm1[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 17, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
@@ -382,8 +384,8 @@ define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_18_00_00_
 ; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    movw $1, %ax
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vmovdqu16 %ymm1, %ymm0 {%k1}
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
+; AVX512VL-NEXT:    vmovdqu16 %ymm0, %ymm1 {%k1}
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm1[0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 18, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
@@ -413,8 +415,8 @@ define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_19_00_00_00_
 ; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    movw $1, %ax
 ; AVX512VL-NEXT:    kmovw %eax, %k1
-; AVX512VL-NEXT:    vmovdqu16 %ymm1, %ymm0 {%k1}
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,0,0,0,3,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
+; AVX512VL-NEXT:    vmovdqu16 %ymm0, %ymm1 {%k1}
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm1[0,0,0,0,0,0,0,0,0,0,0,0,3,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 19, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
@@ -1036,7 +1038,7 @@ define <32 x i8> @shuffle_v32i8_00_33_02_35_04_37_06_39_08_41_10_43_12_45_14_47_
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 33, i32 2, i32 35, i32 4, i32 37, i32 6, i32 39, i32 8, i32 41, i32 10, i32 43, i32 12, i32 45, i32 14, i32 47, i32 16, i32 49, i32 18, i32 51, i32 20, i32 53, i32 22, i32 55, i32 24, i32 57, i32 26, i32 59, i32 28, i32 61, i32 30, i32 63>
   ret <32 x i8> %shuffle
@@ -1061,7 +1063,7 @@ define <32 x i8> @shuffle_v32i8_32_01_34_03_36_05_38_07_40_09_42_11_44_13_46_15_
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k1}
+; AVX512VL-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 32, i32 1, i32 34, i32 3, i32 36, i32 5, i32 38, i32 7, i32 40, i32 9, i32 42, i32 11, i32 44, i32 13, i32 46, i32 15, i32 48, i32 17, i32 50, i32 19, i32 52, i32 21, i32 54, i32 23, i32 56, i32 25, i32 58, i32 27, i32 60, i32 29, i32 62, i32 31>
   ret <32 x i8> %shuffle
@@ -1076,8 +1078,7 @@ define <32 x i8> @shuffle_v32i8_zz_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_
 ; AVX512VL-LABEL: shuffle_v32i8_zz_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_zz_17_zz_19_zz_21_zz_23_zz_25_zz_27_zz_29_zz_31:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
-; AVX512VL-NEXT:    kmovd %eax, %k0
-; AVX512VL-NEXT:    knotd %k0, %k1
+; AVX512VL-NEXT:    kmovd %eax, %k1
 ; AVX512VL-NEXT:    vmovdqu8 %ymm0, %ymm0 {%k1} {z}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> zeroinitializer, <32 x i32> <i32 32, i32 1, i32 34, i32 3, i32 36, i32 5, i32 38, i32 7, i32 40, i32 9, i32 42, i32 11, i32 44, i32 13, i32 46, i32 15, i32 48, i32 17, i32 50, i32 19, i32 52, i32 21, i32 54, i32 23, i32 56, i32 25, i32 58, i32 27, i32 60, i32 29, i32 62, i32 31>
@@ -1144,14 +1145,12 @@ define <32 x i8> @shuffle_v32i8_00_32_00_32_00_32_00_32_00_32_00_32_00_32_00_32_
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_32_00_32_00_32_00_32_00_32_00_32_00_32_00_32_16_48_16_48_16_48_16_48_16_48_16_48_16_48_16_48:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpxor %ymm2, %ymm2, %ymm2
-; AVX512VL-NEXT:    vpshufb %ymm2, %ymm1, %ymm1
 ; AVX512VL-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,0,0,0,4,5,6,7,8,8,8,8,12,13,14,15]
 ; AVX512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,0,1,1,4,4,5,5]
+; AVX512VL-NEXT:    vpxor %ymm2, %ymm2, %ymm2
 ; AVX512VL-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vmovdqu8 %ymm0, %ymm1 {%k1}
-; AVX512VL-NEXT:    vmovdqa %ymm1, %ymm0
+; AVX512VL-NEXT:    vpshufb %ymm2, %ymm1, %ymm0 {%k1}
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 32, i32 0, i32 32, i32 0, i32 32, i32 0, i32 32, i32 0, i32 32, i32 0, i32 32, i32 0, i32 32, i32 0, i32 32, i32 16, i32 48, i32 16, i32 48, i32 16, i32 48, i32 16, i32 48, i32 16, i32 48, i32 16, i32 48, i32 16, i32 48, i32 16, i32 48>
   ret <32 x i8> %shuffle
@@ -1394,11 +1393,10 @@ define <32 x i8> @shuffle_v32i8_00_32_01_33_02_34_03_35_04_36_05_37_06_38_07_39_
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_32_01_33_02_34_03_35_04_36_05_37_06_38_07_39_24_56_25_57_26_58_27_59_28_60_29_61_30_62_31_63:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,0,u,1,u,2,u,3,u,4,u,5,u,6,u,7,u,24,u,25,u,26,u,27,u,28,u,29,u,30,u,31]
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,u,1,u,2,u,3,u,4,u,5,u,6,u,7,u,24,u,25,u,26,u,27,u,28,u,29,u,30,u,31,u]
 ; AVX512VL-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 {%k1} = ymm0[0,u,1,u,2,u,3,u,4,u,5,u,6,u,7,u,24,u,25,u,26,u,27,u,28,u,29,u,30,u,31,u]
-; AVX512VL-NEXT:    vmovdqa %ymm1, %ymm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 {%k1} = ymm1[u,0,u,1,u,2,u,3,u,4,u,5,u,6,u,7,u,24,u,25,u,26,u,27,u,28,u,29,u,30,u,31]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 4, i32 36, i32 5, i32 37, i32 6, i32 38, i32 7, i32 39, i32 24, i32 56, i32 25, i32 57, i32 26, i32 58, i32 27, i32 59, i32 28, i32 60, i32 29, i32 61, i32 30, i32 62, i32 31, i32 63>
   ret <32 x i8> %shuffle
@@ -1424,11 +1422,10 @@ define <32 x i8> @shuffle_v32i8_08_40_09_41_10_42_11_43_12_44_13_45_14_46_15_47_
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_08_40_09_41_10_42_11_43_12_44_13_45_14_46_15_47_16_48_17_49_18_50_19_51_20_52_21_53_22_54_23_55:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,8,u,9,u,10,u,11,u,12,u,13,u,14,u,15,u,16,u,17,u,18,u,19,u,20,u,21,u,22,u,23]
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[8,u,9,u,10,u,11,u,12,u,13,u,14,u,15,u,16,u,17,u,18,u,19,u,20,u,21,u,22,u,23,u]
 ; AVX512VL-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 {%k1} = ymm0[8,u,9,u,10,u,11,u,12,u,13,u,14,u,15,u,16,u,17,u,18,u,19,u,20,u,21,u,22,u,23,u]
-; AVX512VL-NEXT:    vmovdqa %ymm1, %ymm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 {%k1} = ymm1[u,8,u,9,u,10,u,11,u,12,u,13,u,14,u,15,u,16,u,17,u,18,u,19,u,20,u,21,u,22,u,23]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 8, i32 40, i32 9, i32 41, i32 10, i32 42, i32 11, i32 43, i32 12, i32 44, i32 13, i32 45, i32 14, i32 46, i32 15, i32 47, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 20, i32 52, i32 21, i32 53, i32 22, i32 54, i32 23, i32 55>
   ret <32 x i8> %shuffle
@@ -1681,17 +1678,18 @@ define <32 x i8> @shuffle_v32i8_42_45_12_13_35_35_60_40_17_22_29_44_33_12_48_51_
 ; AVX512VL-LABEL: shuffle_v32i8_42_45_12_13_35_35_60_40_17_22_29_44_33_12_48_51_20_19_52_19_49_54_37_32_48_42_59_07_36_34_36_39:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[u,u,u,u,u,u,12,u,u,u,u,u,u,u,0,3,u,u,u,u,u,u,21,16,u,26,u,u,20,18,20,23]
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[10,13,u,u,3,3,u,8,u,u,u,12,1,u,u,u,u,u,20,u,17,22,u,u,16,u,27,u,u,u,u,u]
 ; AVX512VL-NEXT:    movl $-222248896, %eax # imm = 0xF2C0C040
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm2 {%k1} = ymm1[10,13,u,u,3,3,u,8,u,u,u,12,1,u,u,u,u,u,20,u,17,22,u,u,16,u,27,u,u,u,u,u]
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,1,6,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,23,u,u,u,u]
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 {%k1} = ymm2[u,u,u,u,u,u,12,u,u,u,u,u,u,u,0,3,u,u,u,u,u,u,21,16,u,26,u,u,20,18,20,23]
+; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[u,u,u,u,u,u,u,u,1,6,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,23,u,u,u,u]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[u,u,12,13,u,u,u,u,u,u,u,u,u,12,u,u,20,19,u,19,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3,4,5],ymm1[6],ymm0[7]
+; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm2[2],ymm0[3,4,5],ymm2[6],ymm0[7]
 ; AVX512VL-NEXT:    movl $134948620, %eax # imm = 0x80B270C
 ; AVX512VL-NEXT:    kmovd %eax, %k1
-; AVX512VL-NEXT:    vmovdqu8 %ymm2, %ymm0 {%k1}
+; AVX512VL-NEXT:    vmovdqu8 %ymm0, %ymm1 {%k1}
+; AVX512VL-NEXT:    vmovdqa %ymm1, %ymm0
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 42, i32 45, i32 12, i32 13, i32 35, i32 35, i32 60, i32 40, i32 17, i32 22, i32 29, i32 44, i32 33, i32 12, i32 48, i32 51, i32 20, i32 19, i32 52, i32 19, i32 49, i32 54, i32 37, i32 32, i32 48, i32 42, i32 59, i32 7, i32 36, i32 34, i32 36, i32 39>
   ret <32 x i8> %shuffle

diff --git a/test/CodeGen/X86/vector-shuffle-512-v16.ll b/test/CodeGen/X86/vector-shuffle-512-v16.ll
index 628cb9358288b3732893489172c33e77c74a1ff5..983360184fb1b880500f08edad78a98896a1db70 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-512-v16.ll
+++ b/test/CodeGen/X86/vector-shuffle-512-v16.ll
@@ -254,7 +254,7 @@ define <16 x i32> @shuffle_v16i32_0_1_2_19_u_u_u_u_u_u_u_u_u_u_u_u(<16 x i32> %a
 ; ALL:       # BB#0:
 ; ALL-NEXT:    movw $8, %ax
 ; ALL-NEXT:    kmovw %eax, %k1
-; ALL-NEXT:    vpblendmd %zmm0, %zmm1, %zmm0 {%k1}
+; ALL-NEXT:    vmovdqa32 %zmm1, %zmm0 {%k1}
 ; ALL-NEXT:    retq
   %c = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 19, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
   ret <16 x i32> %c

diff --git a/test/CodeGen/X86/vector-shuffle-512-v32.ll b/test/CodeGen/X86/vector-shuffle-512-v32.ll
index 18041dac2c66612cbde439411d92c38b94d5f463..26cd7301fe6049f8537bbf44e3125b4f972939b2 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-512-v32.ll
+++ b/test/CodeGen/X86/vector-shuffle-512-v32.ll
@@ -110,10 +110,9 @@ define <32 x i16> @shuffle_v32i16_1_1_0_0_5_5_4_4_9_9_11_11_13_13_12_12_17_17_19
 define <32 x i16> @shuffle_v32i16_0zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz(<32 x i16> %a) {
 ; ALL-LABEL: shuffle_v32i16_0zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz:
 ; ALL:       # BB#0:
-; ALL-NEXT:    movl    $1, %eax
-; ALL-NEXT:    kmovd   %eax, %k0
-; ALL-NEXT:    knotd   %k0, %k1
-; ALL-NEXT:    vmovdqu16       %zmm0, %zmm0 {%k1} {z}
+; ALL-NEXT:    movl $1, %eax
+; ALL-NEXT:    kmovd %eax, %k1
+; ALL-NEXT:    vmovdqu16 %zmm0, %zmm0 {%k1} {z}
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <32 x i16> %a, <32 x i16> zeroinitializer, <32 x i32> <i32 0, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32>
   ret <32 x i16> %shuffle

diff --git a/test/CodeGen/X86/vector-shuffle-512-v8.ll b/test/CodeGen/X86/vector-shuffle-512-v8.ll
index 59a6629081f175ec3253cbc0001956c82e6e4f38..26e08d365e065472cef9aeba7f548876edc19ea5 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-512-v8.ll
+++ b/test/CodeGen/X86/vector-shuffle-512-v8.ll
@@ -1179,17 +1179,16 @@ define <8 x i64> @shuffle_v8i64_81a3c5e7(<8 x i64> %a, <8 x i64> %b) {
 ;
 ; AVX512F-LABEL: shuffle_v8i64_81a3c5e7:
 ; AVX512F:       # BB#0:
-; AVX512F-NEXT:   movb $-86, %al
-; AVX512F-NEXT:   kmovw %eax, %k1
-; AVX512F-NEXT:   vmovdqa64 %zmm1, %zmm0 {%k1}
-; AVX512F-NEXT:   retq  
-;
+; AVX512F-NEXT:    movb $-86, %al
+; AVX512F-NEXT:    kmovw %eax, %k1
+; AVX512F-NEXT:    vpblendmq %zmm0, %zmm1, %zmm0 {%k1}
+; AVX512F-NEXT:    retq
 ;
 ; AVX512F-32-LABEL: shuffle_v8i64_81a3c5e7:
 ; AVX512F-32:       # BB#0:
 ; AVX512F-32-NEXT:    movb $-86, %al
 ; AVX512F-32-NEXT:    kmovw %eax, %k1
-; AVX512F-32-NEXT:    vmovdqa64 %zmm1, %zmm0 {%k1}
+; AVX512F-32-NEXT:    vpblendmq %zmm0, %zmm1, %zmm0 {%k1}
 ; AVX512F-32-NEXT:    retl
   %shuffle = shufflevector <8 x i64> %a, <8 x i64> %b, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
   ret <8 x i64> %shuffle

diff --git a/test/CodeGen/X86/vector-shuffle-avx512.ll b/test/CodeGen/X86/vector-shuffle-avx512.ll
index 5cef69a0e226702c849d422401b07cfe69309717..4098d16d288cc7b065387329e0348e9780eab5d1 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-avx512.ll
+++ b/test/CodeGen/X86/vector-shuffle-avx512.ll
@@ -595,7 +595,7 @@ define <64 x i8> @test_mm512_mask_blend_epi8(<64 x i8> %A, <64 x i8> %W){
 ; SKX64:       # BB#0: # %entry
 ; SKX64-NEXT:    movabsq $-6148914691236517206, %rax # imm = 0xAAAAAAAAAAAAAAAA
 ; SKX64-NEXT:    kmovq %rax, %k1
-; SKX64-NEXT:    vmovdqu8 %zmm1, %zmm0 {%k1}
+; SKX64-NEXT:    vpblendmb %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: test_mm512_mask_blend_epi8:
@@ -610,7 +610,7 @@ define <64 x i8> @test_mm512_mask_blend_epi8(<64 x i8> %A, <64 x i8> %W){
 ; SKX32-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; SKX32-NEXT:    kmovd %eax, %k0
 ; SKX32-NEXT:    kunpckdq %k0, %k0, %k1
-; SKX32-NEXT:    vmovdqu8 %zmm1, %zmm0 {%k1}
+; SKX32-NEXT:    vpblendmb %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: test_mm512_mask_blend_epi8:
@@ -641,7 +641,7 @@ define <32 x i16> @test_mm512_mask_blend_epi16(<32 x i16> %A, <32 x i16> %W){
 ; SKX64:       # BB#0: # %entry
 ; SKX64-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; SKX64-NEXT:    kmovd %eax, %k1
-; SKX64-NEXT:    vmovdqu16 %zmm1, %zmm0 {%k1}
+; SKX64-NEXT:    vpblendmw %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: test_mm512_mask_blend_epi16:
@@ -654,7 +654,7 @@ define <32 x i16> @test_mm512_mask_blend_epi16(<32 x i16> %A, <32 x i16> %W){
 ; SKX32:       # BB#0: # %entry
 ; SKX32-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; SKX32-NEXT:    kmovd %eax, %k1
-; SKX32-NEXT:    vmovdqu16 %zmm1, %zmm0 {%k1}
+; SKX32-NEXT:    vpblendmw %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: test_mm512_mask_blend_epi16:
@@ -684,28 +684,28 @@ define <16 x i32> @test_mm512_mask_blend_epi32(<16 x i32> %A, <16 x i32> %W){
 ; SKX64:       # BB#0: # %entry
 ; SKX64-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; SKX64-NEXT:    kmovw %eax, %k1
-; SKX64-NEXT:    vmovdqa32 %zmm1, %zmm0 {%k1}
+; SKX64-NEXT:    vpblendmd %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: test_mm512_mask_blend_epi32:
 ; KNL64:       # BB#0: # %entry
 ; KNL64-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; KNL64-NEXT:    kmovw %eax, %k1
-; KNL64-NEXT:    vmovdqa32 %zmm1, %zmm0 {%k1}
+; KNL64-NEXT:    vpblendmd %zmm0, %zmm1, %zmm0 {%k1}
 ; KNL64-NEXT:    retq
 ;
 ; SKX32-LABEL: test_mm512_mask_blend_epi32:
 ; SKX32:       # BB#0: # %entry
 ; SKX32-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; SKX32-NEXT:    kmovw %eax, %k1
-; SKX32-NEXT:    vmovdqa32 %zmm1, %zmm0 {%k1}
+; SKX32-NEXT:    vpblendmd %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: test_mm512_mask_blend_epi32:
 ; KNL32:       # BB#0: # %entry
 ; KNL32-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; KNL32-NEXT:    kmovw %eax, %k1
-; KNL32-NEXT:    vmovdqa32 %zmm1, %zmm0 {%k1}
+; KNL32-NEXT:    vpblendmd %zmm0, %zmm1, %zmm0 {%k1}
 ; KNL32-NEXT:    retl
 entry:
   %0 = shufflevector <16 x i32> %A, <16 x i32> %W, <16 x i32>  <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
@@ -717,28 +717,28 @@ define <8 x i64> @test_mm512_mask_blend_epi64(<8 x i64> %A, <8 x i64> %W){
 ; SKX64:       # BB#0: # %entry
 ; SKX64-NEXT:    movb $-86, %al
 ; SKX64-NEXT:    kmovb %eax, %k1
-; SKX64-NEXT:    vmovdqa64 %zmm1, %zmm0 {%k1}
+; SKX64-NEXT:    vpblendmq %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: test_mm512_mask_blend_epi64:
 ; KNL64:       # BB#0: # %entry
 ; KNL64-NEXT:    movb $-86, %al
 ; KNL64-NEXT:    kmovw %eax, %k1
-; KNL64-NEXT:    vmovdqa64 %zmm1, %zmm0 {%k1}
+; KNL64-NEXT:    vpblendmq %zmm0, %zmm1, %zmm0 {%k1}
 ; KNL64-NEXT:    retq
 ;
 ; SKX32-LABEL: test_mm512_mask_blend_epi64:
 ; SKX32:       # BB#0: # %entry
 ; SKX32-NEXT:    movb $-86, %al
 ; SKX32-NEXT:    kmovb %eax, %k1
-; SKX32-NEXT:    vmovdqa64 %zmm1, %zmm0 {%k1}
+; SKX32-NEXT:    vpblendmq %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: test_mm512_mask_blend_epi64:
 ; KNL32:       # BB#0: # %entry
 ; KNL32-NEXT:    movb $-86, %al
 ; KNL32-NEXT:    kmovw %eax, %k1
-; KNL32-NEXT:    vmovdqa64 %zmm1, %zmm0 {%k1}
+; KNL32-NEXT:    vpblendmq %zmm0, %zmm1, %zmm0 {%k1}
 ; KNL32-NEXT:    retl
 entry:
   %0 = shufflevector <8 x i64> %A, <8 x i64> %W, <8 x i32>  <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
@@ -750,28 +750,28 @@ define <16 x float> @test_mm512_mask_blend_ps(<16 x float> %A, <16 x float> %W){
 ; SKX64:       # BB#0: # %entry
 ; SKX64-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; SKX64-NEXT:    kmovw %eax, %k1
-; SKX64-NEXT:    vmovaps %zmm1, %zmm0 {%k1}
+; SKX64-NEXT:    vblendmps %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: test_mm512_mask_blend_ps:
 ; KNL64:       # BB#0: # %entry
 ; KNL64-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; KNL64-NEXT:    kmovw %eax, %k1
-; KNL64-NEXT:    vmovaps %zmm1, %zmm0 {%k1}
+; KNL64-NEXT:    vblendmps %zmm0, %zmm1, %zmm0 {%k1}
 ; KNL64-NEXT:    retq
 ;
 ; SKX32-LABEL: test_mm512_mask_blend_ps:
 ; SKX32:       # BB#0: # %entry
 ; SKX32-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; SKX32-NEXT:    kmovw %eax, %k1
-; SKX32-NEXT:    vmovaps %zmm1, %zmm0 {%k1}
+; SKX32-NEXT:    vblendmps %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: test_mm512_mask_blend_ps:
 ; KNL32:       # BB#0: # %entry
 ; KNL32-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; KNL32-NEXT:    kmovw %eax, %k1
-; KNL32-NEXT:    vmovaps %zmm1, %zmm0 {%k1}
+; KNL32-NEXT:    vblendmps %zmm0, %zmm1, %zmm0 {%k1}
 ; KNL32-NEXT:    retl
 entry:
   %0 = shufflevector <16 x float> %A, <16 x float> %W, <16 x i32>  <i32 16, i32 1, i32 18, i32 3, i32 20, i32 5, i32 22, i32 7, i32 24, i32 9, i32 26, i32 11, i32 28, i32 13, i32 30, i32 15>
@@ -783,28 +783,28 @@ define <8 x double> @test_mm512_mask_blend_pd(<8 x double> %A, <8 x double> %W){
 ; SKX64:       # BB#0: # %entry
 ; SKX64-NEXT:    movb $-88, %al
 ; SKX64-NEXT:    kmovb %eax, %k1
-; SKX64-NEXT:    vmovapd %zmm1, %zmm0 {%k1}
+; SKX64-NEXT:    vblendmpd %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: test_mm512_mask_blend_pd:
 ; KNL64:       # BB#0: # %entry
 ; KNL64-NEXT:    movb $-88, %al
 ; KNL64-NEXT:    kmovw %eax, %k1
-; KNL64-NEXT:    vmovapd %zmm1, %zmm0 {%k1}
+; KNL64-NEXT:    vblendmpd %zmm0, %zmm1, %zmm0 {%k1}
 ; KNL64-NEXT:    retq
 ;
 ; SKX32-LABEL: test_mm512_mask_blend_pd:
 ; SKX32:       # BB#0: # %entry
 ; SKX32-NEXT:    movb $-88, %al
 ; SKX32-NEXT:    kmovb %eax, %k1
-; SKX32-NEXT:    vmovapd %zmm1, %zmm0 {%k1}
+; SKX32-NEXT:    vblendmpd %zmm0, %zmm1, %zmm0 {%k1}
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: test_mm512_mask_blend_pd:
 ; KNL32:       # BB#0: # %entry
 ; KNL32-NEXT:    movb $-88, %al
 ; KNL32-NEXT:    kmovw %eax, %k1
-; KNL32-NEXT:    vmovapd %zmm1, %zmm0 {%k1}
+; KNL32-NEXT:    vblendmpd %zmm0, %zmm1, %zmm0 {%k1}
 ; KNL32-NEXT:    retl
 entry:
   %0 = shufflevector <8 x double> %A, <8 x double> %W, <8 x i32>  <i32 8, i32 9, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
@@ -817,7 +817,7 @@ define <32 x i8> @test_mm256_mask_blend_epi8(<32 x i8> %A, <32 x i8> %W){
 ; SKX64:       # BB#0: # %entry
 ; SKX64-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; SKX64-NEXT:    kmovd %eax, %k1
-; SKX64-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k1}
+; SKX64-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1}
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: test_mm256_mask_blend_epi8:
@@ -830,7 +830,7 @@ define <32 x i8> @test_mm256_mask_blend_epi8(<32 x i8> %A, <32 x i8> %W){
 ; SKX32:       # BB#0: # %entry
 ; SKX32-NEXT:    movl $-1431655766, %eax # imm = 0xAAAAAAAA
 ; SKX32-NEXT:    kmovd %eax, %k1
-; SKX32-NEXT:    vmovdqu8 %ymm1, %ymm0 {%k1}
+; SKX32-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1}
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: test_mm256_mask_blend_epi8:
@@ -848,7 +848,7 @@ define <16 x i8> @test_mm_mask_blend_epi8(<16 x i8> %A, <16 x i8> %W){
 ; SKX64:       # BB#0: # %entry
 ; SKX64-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; SKX64-NEXT:    kmovw %eax, %k1
-; SKX64-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
+; SKX64-NEXT:    vpblendmb %xmm0, %xmm1, %xmm0 {%k1}
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: test_mm_mask_blend_epi8:
@@ -861,7 +861,7 @@ define <16 x i8> @test_mm_mask_blend_epi8(<16 x i8> %A, <16 x i8> %W){
 ; SKX32:       # BB#0: # %entry
 ; SKX32-NEXT:    movw $-21846, %ax # imm = 0xAAAA
 ; SKX32-NEXT:    kmovw %eax, %k1
-; SKX32-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
+; SKX32-NEXT:    vpblendmb %xmm0, %xmm1, %xmm0 {%k1}
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: test_mm_mask_blend_epi8: