[AVX-512] Add tests for masked palignr/valignd/valignq shuffles, many of which show failures to fold the masking into the operation.

Many of these problems are because shuffle lowering widens element size and reduces element count when possible. This causes the shuffle to become separated from the select by a bitcast. Future patches will work to improve these cases by rewriting the shuffle back to a narrow element type if we think it can result in folding the mask.

llvm-svn: 287503

3 files changed, 361 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll b/llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll
index d13c867b087..02c49744f5b 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll
@@ -369,3 +369,42 @@ define <16 x i32> @mask_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15
   %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> %passthru
   ret <16 x i32> %res
 }
+
+define <16 x i32> @mask_shuffle_v16i32_01_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16(<16 x i32> %a, <16 x i32> %b, <16 x i32> %passthru, i16 %mask) {
+; ALL-LABEL: mask_shuffle_v16i32_01_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16:
+; ALL:       # BB#0:
+; ALL-NEXT:    valignq {{.*#+}} zmm0 = zmm0[1,2,3,4,5,6,7],zmm1[0]
+; ALL-NEXT:    kmovw %edi, %k1
+; ALL-NEXT:    vpblendmd %zmm0, %zmm2, %zmm0 {%k1}
+; ALL-NEXT:    retq
+  %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17>
+  %mask.cast = bitcast i16 %mask to <16 x i1>
+  %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> %passthru
+  ret <16 x i32> %res
+}
+
+define <16 x i32> @maskz_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00_01(<16 x i32> %a, i16 %mask) {
+; ALL-LABEL: maskz_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00_01:
+; ALL:       # BB#0:
+; ALL-NEXT:    valignq {{.*#+}} zmm0 = zmm0[1,2,3,4,5,6,7,0]
+; ALL-NEXT:    kmovw %edi, %k1
+; ALL-NEXT:    vmovdqa32 %zmm0, %zmm0 {%k1} {z}
+; ALL-NEXT:    retq
+  %shuffle = shufflevector <16 x i32> %a, <16 x i32> undef, <16 x i32><i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 0, i32 1>
+  %mask.cast = bitcast i16 %mask to <16 x i1>
+  %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> zeroinitializer
+  ret <16 x i32> %res
+}
+
+define <16 x i32> @maskz_shuffle_v16i32_01_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16(<16 x i32> %a, <16 x i32> %b, i16 %mask) {
+; ALL-LABEL: maskz_shuffle_v16i32_01_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16:
+; ALL:       # BB#0:
+; ALL-NEXT:    valignq {{.*#+}} zmm0 = zmm0[1,2,3,4,5,6,7],zmm1[0]
+; ALL-NEXT:    kmovw %edi, %k1
+; ALL-NEXT:    vmovdqa32 %zmm0, %zmm0 {%k1} {z}
+; ALL-NEXT:    retq
+  %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17>
+  %mask.cast = bitcast i16 %mask to <16 x i1>
+  %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> zeroinitializer
+  ret <16 x i32> %res
+}
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-512-v8.ll b/llvm/test/CodeGen/X86/vector-shuffle-512-v8.ll
index 0f163ba2188..2cb35827561 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-512-v8.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-512-v8.ll
@@ -2305,3 +2305,83 @@ define <8 x i64> @shuffle_v8i64_12345670(<8 x i64> %a) {
   %shuffle = shufflevector <8 x i64> %a, <8 x i64> undef, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0>
   ret <8 x i64> %shuffle
 }
+
+define <8 x i64> @mask_shuffle_v8i64_12345678(<8 x i64> %a, <8 x i64> %b, <16 x i32> %passthru, i8 %mask) {
+;
+; AVX512F-LABEL: mask_shuffle_v8i64_12345678:
+; AVX512F:       # BB#0:
+; AVX512F-NEXT:    kmovw %edi, %k1
+; AVX512F-NEXT:    valignq {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,4,5,6,7],zmm1[0]
+; AVX512F-NEXT:    retq
+;
+; AVX512F-32-LABEL: mask_shuffle_v8i64_12345678:
+; AVX512F-32:       # BB#0:
+; AVX512F-32-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
+; AVX512F-32-NEXT:    kmovw %eax, %k1
+; AVX512F-32-NEXT:    valignq {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,4,5,6,7],zmm1[0]
+; AVX512F-32-NEXT:    retl
+  %shuffle = shufflevector <8 x i64> %a, <8 x i64> %b, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i64> %shuffle, <8 x i64> zeroinitializer
+  ret <8 x i64> %res
+}
+
+define <8 x i64> @mask_shuffle_v8i64_12345670(<8 x i64> %a, <16 x i32> %passthru, i8 %mask) {
+;
+; AVX512F-LABEL: mask_shuffle_v8i64_12345670:
+; AVX512F:       # BB#0:
+; AVX512F-NEXT:    kmovw %edi, %k1
+; AVX512F-NEXT:    valignq {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,4,5,6,7,0]
+; AVX512F-NEXT:    retq
+;
+; AVX512F-32-LABEL: mask_shuffle_v8i64_12345670:
+; AVX512F-32:       # BB#0:
+; AVX512F-32-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
+; AVX512F-32-NEXT:    kmovw %eax, %k1
+; AVX512F-32-NEXT:    valignq {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,4,5,6,7,0]
+; AVX512F-32-NEXT:    retl
+  %shuffle = shufflevector <8 x i64> %a, <8 x i64> undef, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i64> %shuffle, <8 x i64> zeroinitializer
+  ret <8 x i64> %res
+}
+
+define <8 x i64> @maskz_shuffle_v8i64_12345678(<8 x i64> %a, <8 x i64> %b, i8 %mask) {
+;
+; AVX512F-LABEL: maskz_shuffle_v8i64_12345678:
+; AVX512F:       # BB#0:
+; AVX512F-NEXT:    kmovw %edi, %k1
+; AVX512F-NEXT:    valignq {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,4,5,6,7],zmm1[0]
+; AVX512F-NEXT:    retq
+;
+; AVX512F-32-LABEL: maskz_shuffle_v8i64_12345678:
+; AVX512F-32:       # BB#0:
+; AVX512F-32-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
+; AVX512F-32-NEXT:    kmovw %eax, %k1
+; AVX512F-32-NEXT:    valignq {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,4,5,6,7],zmm1[0]
+; AVX512F-32-NEXT:    retl
+  %shuffle = shufflevector <8 x i64> %a, <8 x i64> %b, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i64> %shuffle, <8 x i64> zeroinitializer
+  ret <8 x i64> %res
+}
+
+define <8 x i64> @maskz_shuffle_v8i64_12345670(<8 x i64> %a, i8 %mask) {
+;
+; AVX512F-LABEL: maskz_shuffle_v8i64_12345670:
+; AVX512F:       # BB#0:
+; AVX512F-NEXT:    kmovw %edi, %k1
+; AVX512F-NEXT:    valignq {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,4,5,6,7,0]
+; AVX512F-NEXT:    retq
+;
+; AVX512F-32-LABEL: maskz_shuffle_v8i64_12345670:
+; AVX512F-32:       # BB#0:
+; AVX512F-32-NEXT:    movzbl {{[0-9]+}}(%esp), %eax
+; AVX512F-32-NEXT:    kmovw %eax, %k1
+; AVX512F-32-NEXT:    valignq {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,4,5,6,7,0]
+; AVX512F-32-NEXT:    retl
+  %shuffle = shufflevector <8 x i64> %a, <8 x i64> undef, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i64> %shuffle, <8 x i64> zeroinitializer
+  ret <8 x i64> %res
+}
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-masked.ll b/llvm/test/CodeGen/X86/vector-shuffle-masked.ll
new file mode 100644
index 00000000000..24cf6f9a966
--- /dev/null
+++ b/llvm/test/CodeGen/X86/vector-shuffle-masked.ll
@@ -0,0 +1,242 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefix=CHECK
+
+define <4 x i32> @mask_shuffle_v4i32_1234(<4 x i32> %a, <4 x i32> %b, <4 x i32> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v4i32_1234:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[4,5,6,7,8,9,10,11,12,13,14,15],xmm1[0,1,2,3]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vpblendmd %xmm0, %xmm2, %xmm0 {%k1}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 1, i32 2, i32 3, i32 4>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %res = select <4 x i1> %mask.extract, <4 x i32> %shuffle, <4 x i32> %passthru
+  ret <4 x i32> %res
+}
+
+define <4 x i32> @maskz_shuffle_v4i32_1234(<4 x i32> %a, <4 x i32> %b, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v4i32_1234:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[4,5,6,7,8,9,10,11,12,13,14,15],xmm1[0,1,2,3]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vmovdqa32 %xmm0, %xmm0 {%k1} {z}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 1, i32 2, i32 3, i32 4>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %res = select <4 x i1> %mask.extract, <4 x i32> %shuffle, <4 x i32> zeroinitializer
+  ret <4 x i32> %res
+}
+
+define <4 x i32> @mask_shuffle_v4i32_2345(<4 x i32> %a, <4 x i32> %b, <4 x i32> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v4i32_2345:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vpblendmd %xmm0, %xmm2, %xmm0 {%k1}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %res = select <4 x i1> %mask.extract, <4 x i32> %shuffle, <4 x i32> %passthru
+  ret <4 x i32> %res
+}
+
+define <4 x i32> @maskz_shuffle_v4i32_2345(<4 x i32> %a, <4 x i32> %b, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v4i32_2345:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vmovdqa32 %xmm0, %xmm0 {%k1} {z}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %res = select <4 x i1> %mask.extract, <4 x i32> %shuffle, <4 x i32> zeroinitializer
+  ret <4 x i32> %res
+}
+
+define <2 x i64> @mask_shuffle_v2i64_12(<2 x i64> %a, <2 x i64> %b, <2 x i64> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v2i64_12:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vpblendmq %xmm0, %xmm2, %xmm0 {%k1}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 1, i32 2>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
+  %res = select <2 x i1> %mask.extract, <2 x i64> %shuffle, <2 x i64> %passthru
+  ret <2 x i64> %res
+}
+
+define <2 x i64> @maskz_shuffle_v2i64_12(<2 x i64> %a, <2 x i64> %b, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v2i64_12:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vmovdqa64 %xmm0, %xmm0 {%k1} {z}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 1, i32 2>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
+  %res = select <2 x i1> %mask.extract, <2 x i64> %shuffle, <2 x i64> zeroinitializer
+  ret <2 x i64> %res
+}
+
+define <4 x i64> @mask_shuffle_v4i64_1234(<4 x i64> %a, <4 x i64> %b, <4 x i64> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v4i64_1234:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    valignq {{.*#+}} ymm2 {%k1} = ymm0[1,2,3],ymm1[0]
+; CHECK-NEXT:    vmovdqa64 %ymm2, %ymm0
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 3, i32 4>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %res = select <4 x i1> %mask.extract, <4 x i64> %shuffle, <4 x i64> %passthru
+  ret <4 x i64> %res
+}
+
+define <4 x i64> @maskz_shuffle_v4i64_1234(<4 x i64> %a, <4 x i64> %b, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v4i64_1234:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    valignq {{.*#+}} ymm0 {%k1} {z} = ymm0[1,2,3],ymm1[0]
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 2, i32 3, i32 4>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %res = select <4 x i1> %mask.extract, <4 x i64> %shuffle, <4 x i64> zeroinitializer
+  ret <4 x i64> %res
+}
+
+define <4 x i64> @mask_shuffle_v4i64_1230(<4 x i64> %a, <4 x i64> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v4i64_1230:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vpermq {{.*#+}} ymm1 {%k1} = ymm0[1,2,3,0]
+; CHECK-NEXT:    vmovdqa64 %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> undef, <4 x i32> <i32 1, i32 2, i32 3, i32 0>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %res = select <4 x i1> %mask.extract, <4 x i64> %shuffle, <4 x i64> %passthru
+  ret <4 x i64> %res
+}
+
+define <4 x i64> @maskz_shuffle_v4i64_1230(<4 x i64> %a, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v4i64_1230:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vpermq {{.*#+}} ymm0 {%k1} {z} = ymm0[1,2,3,0]
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> undef, <4 x i32> <i32 1, i32 2, i32 3, i32 0>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %mask.extract = shufflevector <8 x i1> %mask.cast, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %res = select <4 x i1> %mask.extract, <4 x i64> %shuffle, <4 x i64> zeroinitializer
+  ret <4 x i64> %res
+}
+
+define <8 x i32> @mask_shuffle_v8i32_12345678(<8 x i32> %a, <8 x i32> %b, <8 x i32> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v8i32_12345678:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    valignd {{.*#+}} ymm2 {%k1} = ymm0[1,2,3,4,5,6,7],ymm1[0]
+; CHECK-NEXT:    vmovdqa64 %ymm2, %ymm0
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i32> %shuffle, <8 x i32> %passthru
+  ret <8 x i32> %res
+}
+
+define <8 x i32> @maskz_shuffle_v8i32_12345678(<8 x i32> %a, <8 x i32> %b, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v8i32_12345678:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    valignd {{.*#+}} ymm0 {%k1} {z} = ymm0[1,2,3,4,5,6,7],ymm1[0]
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i32> %shuffle, <8 x i32> zeroinitializer
+  ret <8 x i32> %res
+}
+
+define <8 x i32> @mask_shuffle_v8i32_23456789(<8 x i32> %a, <8 x i32> %b, <8 x i32> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v8i32_23456789:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    valignq {{.*#+}} ymm0 = ymm0[1,2,3],ymm1[0]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vpblendmd %ymm0, %ymm2, %ymm0 {%k1}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i32> %shuffle, <8 x i32> %passthru
+  ret <8 x i32> %res
+}
+
+define <8 x i32> @maskz_shuffle_v8i32_23456789(<8 x i32> %a, <8 x i32> %b, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v8i32_23456789:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    valignq {{.*#+}} ymm0 = ymm0[1,2,3],ymm1[0]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vmovdqa32 %ymm0, %ymm0 {%k1} {z}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i32> %shuffle, <8 x i32> zeroinitializer
+  ret <8 x i32> %res
+}
+
+define <8 x i32> @mask_shuffle_v8i32_12345670(<8 x i32> %a, <8 x i32> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v8i32_12345670:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    valignd {{.*#+}} ymm1 {%k1} = ymm0[1,2,3,4,5,6,7,0]
+; CHECK-NEXT:    vmovdqa64 %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> undef, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i32> %shuffle, <8 x i32> %passthru
+  ret <8 x i32> %res
+}
+
+define <8 x i32> @maskz_shuffle_v8i32_12345670(<8 x i32> %a, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v8i32_12345670:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    valignd {{.*#+}} ymm0 {%k1} {z} = ymm0[1,2,3,4,5,6,7,0]
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> undef, <8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i32> %shuffle, <8 x i32> zeroinitializer
+  ret <8 x i32> %res
+}
+
+define <8 x i32> @mask_shuffle_v8i32_23456701(<8 x i32> %a, <8 x i32> %passthru, i8 %mask) {
+; CHECK-LABEL: mask_shuffle_v8i32_23456701:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[1,2,3,0]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vpblendmd %ymm0, %ymm1, %ymm0 {%k1}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i32> %shuffle, <8 x i32> %passthru
+  ret <8 x i32> %res
+}
+
+define <8 x i32> @maskz_shuffle_v8i32_23456701(<8 x i32> %a, i8 %mask) {
+; CHECK-LABEL: maskz_shuffle_v8i32_23456701:
+; CHECK:       # BB#0:
+; CHECK-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[1,2,3,0]
+; CHECK-NEXT:    kmovb %edi, %k1
+; CHECK-NEXT:    vmovdqa32 %ymm0, %ymm0 {%k1} {z}
+; CHECK-NEXT:    retq
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1>
+  %mask.cast = bitcast i8 %mask to <8 x i1>
+  %res = select <8 x i1> %mask.cast, <8 x i32> %shuffle, <8 x i32> zeroinitializer
+  ret <8 x i32> %res
+}