aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorSimon Pilgrim <llvm-dev@redking.me.uk>2024-06-26 18:31:24 +0100
committerSimon Pilgrim <llvm-dev@redking.me.uk>2024-06-27 10:45:11 +0100
commit8467cc61cef88818b5ac811176aa6fd7b22c7538 (patch)
treea1a7e4e500e43ce78b0e392ee8052b8135fd0a21
parent2033b1cf16f040e1369d8efba8439dcd3e36ed31 (diff)
downloadllvm-8467cc61cef88818b5ac811176aa6fd7b22c7538.zip
llvm-8467cc61cef88818b5ac811176aa6fd7b22c7538.tar.gz
llvm-8467cc61cef88818b5ac811176aa6fd7b22c7538.tar.bz2
[X86] Add phase ordering test coverage for #58895
-rw-r--r--llvm/test/Transforms/PhaseOrdering/X86/blendv-select.ll574
1 files changed, 574 insertions, 0 deletions
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/blendv-select.ll b/llvm/test/Transforms/PhaseOrdering/X86/blendv-select.ll
new file mode 100644
index 0000000..bccd189
--- /dev/null
+++ b/llvm/test/Transforms/PhaseOrdering/X86/blendv-select.ll
@@ -0,0 +1,574 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -O3 -S -mtriple=x86_64-- -mcpu=x86-64-v2 | FileCheck %s
+; RUN: opt < %s -O3 -S -mtriple=x86_64-- -mcpu=x86-64-v3 | FileCheck %s
+; RUN: opt < %s -O3 -S -mtriple=x86_64-- -mcpu=x86-64-v4 | FileCheck %s
+
+;
+; TODO: PR58895 - replace shuffled _mm_blendv_epi8+icmp with select+icmp
+;
+
+;
+; 256-bits types split to use 128-bit PBLENDVB selects
+;
+
+define <4 x double> @x86_pblendvb_v4f64_v2f64(<4 x double> %a, <4 x double> %b, <4 x double> %c, <4 x double> %d) {
+; CHECK-LABEL: @x86_pblendvb_v4f64_v2f64(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <4 x double> [[A:%.*]] to <32 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <4 x double> [[B:%.*]] to <32 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <4 x double> [[C:%.*]], [[D:%.*]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <4 x i64> [[SEXT]] to <32 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_HI]], <16 x i8> [[B_HI]], <16 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <16 x i8> [[SEL_LO]], <16 x i8> [[SEL_HI]], <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <32 x i8> [[CONCAT]] to <4 x double>
+; CHECK-NEXT: ret <4 x double> [[RES]]
+;
+ %a.bc = bitcast <4 x double> %a to <32 x i8>
+ %b.bc = bitcast <4 x double> %b to <32 x i8>
+ %a.lo = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %a.hi = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.hi = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %cmp = fcmp olt <4 x double> %c, %d
+ %sext = sext <4 x i1> %cmp to <4 x i64>
+ %sext.bc = bitcast <4 x i64> %sext to <32 x i8>
+ %sext.lo = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %sext.hi = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %sel.hi = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.hi, <16 x i8> %b.hi, <16 x i8> %sext.hi)
+ %concat = shufflevector <16 x i8> %sel.lo, <16 x i8> %sel.hi, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %res = bitcast <32 x i8> %concat to <4 x double>
+ ret <4 x double> %res
+}
+
+define <8 x float> @x86_pblendvb_v8f32_v4f32(<8 x float> %a, <8 x float> %b, <8 x float> %c, <8 x float> %d) {
+; CHECK-LABEL: @x86_pblendvb_v8f32_v4f32(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <8 x float> [[A:%.*]] to <32 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <8 x float> [[B:%.*]] to <32 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x float> [[C:%.*]], [[D:%.*]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <8 x i32> [[SEXT]] to <32 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_HI]], <16 x i8> [[B_HI]], <16 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <16 x i8> [[SEL_LO]], <16 x i8> [[SEL_HI]], <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <32 x i8> [[CONCAT]] to <8 x float>
+; CHECK-NEXT: ret <8 x float> [[RES]]
+;
+ %a.bc = bitcast <8 x float> %a to <32 x i8>
+ %b.bc = bitcast <8 x float> %b to <32 x i8>
+ %a.lo = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %a.hi = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.hi = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %cmp = fcmp olt <8 x float> %c, %d
+ %sext = sext <8 x i1> %cmp to <8 x i32>
+ %sext.bc = bitcast <8 x i32> %sext to <32 x i8>
+ %sext.lo = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %sext.hi = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %sel.hi = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.hi, <16 x i8> %b.hi, <16 x i8> %sext.hi)
+ %concat = shufflevector <16 x i8> %sel.lo, <16 x i8> %sel.hi, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %res = bitcast <32 x i8> %concat to <8 x float>
+ ret <8 x float> %res
+}
+
+define <4 x i64> @x86_pblendvb_v4i64_v2i64(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v4i64_v2i64(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <4 x i64> [[A:%.*]] to <32 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <4 x i64> [[B:%.*]] to <32 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <4 x i64> [[C:%.*]], [[D:%.*]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i64>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <4 x i64> [[SEXT]] to <32 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_HI]], <16 x i8> [[B_HI]], <16 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <16 x i8> [[SEL_LO]], <16 x i8> [[SEL_HI]], <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <32 x i8> [[CONCAT]] to <4 x i64>
+; CHECK-NEXT: ret <4 x i64> [[RES]]
+;
+ %a.bc = bitcast <4 x i64> %a to <32 x i8>
+ %b.bc = bitcast <4 x i64> %b to <32 x i8>
+ %a.lo = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %a.hi = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.hi = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %cmp = icmp slt <4 x i64> %c, %d
+ %sext = sext <4 x i1> %cmp to <4 x i64>
+ %sext.bc = bitcast <4 x i64> %sext to <32 x i8>
+ %sext.lo = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %sext.hi = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %sel.hi = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.hi, <16 x i8> %b.hi, <16 x i8> %sext.hi)
+ %concat = shufflevector <16 x i8> %sel.lo, <16 x i8> %sel.hi, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %res = bitcast <32 x i8> %concat to <4 x i64>
+ ret <4 x i64> %res
+}
+
+define <4 x i64> @x86_pblendvb_v8i32_v4i32(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v8i32_v4i32(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <4 x i64> [[A:%.*]] to <32 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <4 x i64> [[B:%.*]] to <32 x i8>
+; CHECK-NEXT: [[C_BC:%.*]] = bitcast <4 x i64> [[C:%.*]] to <8 x i32>
+; CHECK-NEXT: [[D_BC:%.*]] = bitcast <4 x i64> [[D:%.*]] to <8 x i32>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <8 x i32> [[C_BC]], [[D_BC]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i32>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <8 x i32> [[SEXT]] to <32 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_HI]], <16 x i8> [[B_HI]], <16 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <16 x i8> [[SEL_LO]], <16 x i8> [[SEL_HI]], <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <32 x i8> [[CONCAT]] to <4 x i64>
+; CHECK-NEXT: ret <4 x i64> [[RES]]
+;
+ %a.bc = bitcast <4 x i64> %a to <32 x i8>
+ %b.bc = bitcast <4 x i64> %b to <32 x i8>
+ %c.bc = bitcast <4 x i64> %c to <8 x i32>
+ %d.bc = bitcast <4 x i64> %d to <8 x i32>
+ %a.lo = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %a.hi = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.hi = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %cmp = icmp slt <8 x i32> %c.bc, %d.bc
+ %sext = sext <8 x i1> %cmp to <8 x i32>
+ %sext.bc = bitcast <8 x i32> %sext to <32 x i8>
+ %sext.lo = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %sext.hi = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %sel.hi = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.hi, <16 x i8> %b.hi, <16 x i8> %sext.hi)
+ %concat = shufflevector <16 x i8> %sel.lo, <16 x i8> %sel.hi, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %res = bitcast <32 x i8> %concat to <4 x i64>
+ ret <4 x i64> %res
+}
+
+define <4 x i64> @x86_pblendvb_v16i16_v8i16(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v16i16_v8i16(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <4 x i64> [[A:%.*]] to <32 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <4 x i64> [[B:%.*]] to <32 x i8>
+; CHECK-NEXT: [[C_BC:%.*]] = bitcast <4 x i64> [[C:%.*]] to <16 x i16>
+; CHECK-NEXT: [[D_BC:%.*]] = bitcast <4 x i64> [[D:%.*]] to <16 x i16>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <16 x i16> [[C_BC]], [[D_BC]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i16>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <16 x i16> [[SEXT]] to <32 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_HI]], <16 x i8> [[B_HI]], <16 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <16 x i8> [[SEL_LO]], <16 x i8> [[SEL_HI]], <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <32 x i8> [[CONCAT]] to <4 x i64>
+; CHECK-NEXT: ret <4 x i64> [[RES]]
+;
+ %a.bc = bitcast <4 x i64> %a to <32 x i8>
+ %b.bc = bitcast <4 x i64> %b to <32 x i8>
+ %c.bc = bitcast <4 x i64> %c to <16 x i16>
+ %d.bc = bitcast <4 x i64> %d to <16 x i16>
+ %a.lo = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %a.hi = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.hi = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %cmp = icmp slt <16 x i16> %c.bc, %d.bc
+ %sext = sext <16 x i1> %cmp to <16 x i16>
+ %sext.bc = bitcast <16 x i16> %sext to <32 x i8>
+ %sext.lo = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %sext.hi = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %sel.hi = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.hi, <16 x i8> %b.hi, <16 x i8> %sext.hi)
+ %concat = shufflevector <16 x i8> %sel.lo, <16 x i8> %sel.hi, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %res = bitcast <32 x i8> %concat to <4 x i64>
+ ret <4 x i64> %res
+}
+
+define <4 x i64> @x86_pblendvb_v32i8_v16i8(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v32i8_v16i8(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <4 x i64> [[A:%.*]] to <32 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <4 x i64> [[B:%.*]] to <32 x i8>
+; CHECK-NEXT: [[C_BC:%.*]] = bitcast <4 x i64> [[C:%.*]] to <32 x i8>
+; CHECK-NEXT: [[D_BC:%.*]] = bitcast <4 x i64> [[D:%.*]] to <32 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <32 x i8> [[C_BC]], [[D_BC]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <32 x i8> [[SEXT]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <32 x i8> [[SEXT]], <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_HI]], <16 x i8> [[B_HI]], <16 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <16 x i8> [[SEL_LO]], <16 x i8> [[SEL_HI]], <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <32 x i8> [[CONCAT]] to <4 x i64>
+; CHECK-NEXT: ret <4 x i64> [[RES]]
+;
+ %a.bc = bitcast <4 x i64> %a to <32 x i8>
+ %b.bc = bitcast <4 x i64> %b to <32 x i8>
+ %c.bc = bitcast <4 x i64> %c to <32 x i8>
+ %d.bc = bitcast <4 x i64> %d to <32 x i8>
+ %a.lo = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %a.hi = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.hi = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %cmp = icmp slt <32 x i8> %c.bc, %d.bc
+ %sext = sext <32 x i1> %cmp to <32 x i8>
+ %sext.lo = shufflevector <32 x i8> %sext, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %sext.hi = shufflevector <32 x i8> %sext, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %sel.hi = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.hi, <16 x i8> %b.hi, <16 x i8> %sext.hi)
+ %concat = shufflevector <16 x i8> %sel.lo, <16 x i8> %sel.hi, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %res = bitcast <32 x i8> %concat to <4 x i64>
+ ret <4 x i64> %res
+}
+
+;
+; 512-bits types split to use 256-bit PBLENDVB selects
+;
+
+define <8 x double> @x86_pblendvb_v8f64_v4f64(<8 x double> %a, <8 x double> %b, <8 x double> %c, <8 x double> %d) {
+; CHECK-LABEL: @x86_pblendvb_v8f64_v4f64(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <8 x double> [[A:%.*]] to <64 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <8 x double> [[B:%.*]] to <64 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <8 x double> [[C:%.*]], [[D:%.*]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i64>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <8 x i64> [[SEXT]] to <64 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_LO]], <32 x i8> [[B_LO]], <32 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_HI]], <32 x i8> [[B_HI]], <32 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <32 x i8> [[SEL_LO]], <32 x i8> [[SEL_HI]], <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <64 x i8> [[CONCAT]] to <8 x double>
+; CHECK-NEXT: ret <8 x double> [[RES]]
+;
+ %a.bc = bitcast <8 x double> %a to <64 x i8>
+ %b.bc = bitcast <8 x double> %b to <64 x i8>
+ %a.lo = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.lo = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %a.hi = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %b.hi = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %cmp = fcmp olt <8 x double> %c, %d
+ %sext = sext <8 x i1> %cmp to <8 x i64>
+ %sext.bc = bitcast <8 x i64> %sext to <64 x i8>
+ %sext.lo = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sext.hi = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %sel.lo = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.lo, <32 x i8> %b.lo, <32 x i8> %sext.lo)
+ %sel.hi = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.hi, <32 x i8> %b.hi, <32 x i8> %sext.hi)
+ %concat = shufflevector <32 x i8> %sel.lo, <32 x i8> %sel.hi, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %res = bitcast <64 x i8> %concat to <8 x double>
+ ret <8 x double> %res
+}
+
+define <16 x float> @x86_pblendvb_v16f32_v8f32(<16 x float> %a, <16 x float> %b, <16 x float> %c, <16 x float> %d) {
+; CHECK-LABEL: @x86_pblendvb_v16f32_v8f32(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <16 x float> [[A:%.*]] to <64 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <16 x float> [[B:%.*]] to <64 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <16 x float> [[C:%.*]], [[D:%.*]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i32>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <16 x i32> [[SEXT]] to <64 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_LO]], <32 x i8> [[B_LO]], <32 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_HI]], <32 x i8> [[B_HI]], <32 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <32 x i8> [[SEL_LO]], <32 x i8> [[SEL_HI]], <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <64 x i8> [[CONCAT]] to <16 x float>
+; CHECK-NEXT: ret <16 x float> [[RES]]
+;
+ %a.bc = bitcast <16 x float> %a to <64 x i8>
+ %b.bc = bitcast <16 x float> %b to <64 x i8>
+ %a.lo = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.lo = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %a.hi = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %b.hi = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %cmp = fcmp olt <16 x float> %c, %d
+ %sext = sext <16 x i1> %cmp to <16 x i32>
+ %sext.bc = bitcast <16 x i32> %sext to <64 x i8>
+ %sext.lo = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sext.hi = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %sel.lo = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.lo, <32 x i8> %b.lo, <32 x i8> %sext.lo)
+ %sel.hi = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.hi, <32 x i8> %b.hi, <32 x i8> %sext.hi)
+ %concat = shufflevector <32 x i8> %sel.lo, <32 x i8> %sel.hi, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %res = bitcast <64 x i8> %concat to <16 x float>
+ ret <16 x float> %res
+}
+
+define <8 x i64> @x86_pblendvb_v8i64_v4i64(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v8i64_v4i64(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <8 x i64> [[A:%.*]] to <64 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <8 x i64> [[B:%.*]] to <64 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <8 x i64> [[C:%.*]], [[D:%.*]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <8 x i1> [[CMP]] to <8 x i64>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <8 x i64> [[SEXT]] to <64 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_LO]], <32 x i8> [[B_LO]], <32 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_HI]], <32 x i8> [[B_HI]], <32 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <32 x i8> [[SEL_LO]], <32 x i8> [[SEL_HI]], <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <64 x i8> [[CONCAT]] to <8 x i64>
+; CHECK-NEXT: ret <8 x i64> [[RES]]
+;
+ %a.bc = bitcast <8 x i64> %a to <64 x i8>
+ %b.bc = bitcast <8 x i64> %b to <64 x i8>
+ %a.lo = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.lo = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %a.hi = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %b.hi = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %cmp = icmp slt <8 x i64> %c, %d
+ %sext = sext <8 x i1> %cmp to <8 x i64>
+ %sext.bc = bitcast <8 x i64> %sext to <64 x i8>
+ %sext.lo = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sext.hi = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %sel.lo = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.lo, <32 x i8> %b.lo, <32 x i8> %sext.lo)
+ %sel.hi = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.hi, <32 x i8> %b.hi, <32 x i8> %sext.hi)
+ %concat = shufflevector <32 x i8> %sel.lo, <32 x i8> %sel.hi, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %res = bitcast <64 x i8> %concat to <8 x i64>
+ ret <8 x i64> %res
+}
+
+define <8 x i64> @x86_pblendvb_v16i32_v8i32(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v16i32_v8i32(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <8 x i64> [[A:%.*]] to <64 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <8 x i64> [[B:%.*]] to <64 x i8>
+; CHECK-NEXT: [[C_BC:%.*]] = bitcast <8 x i64> [[A]] to <16 x i32>
+; CHECK-NEXT: [[D_BC:%.*]] = bitcast <8 x i64> [[B]] to <16 x i32>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <16 x i32> [[C_BC]], [[D_BC]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i32>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <16 x i32> [[SEXT]] to <64 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_LO]], <32 x i8> [[B_LO]], <32 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_HI]], <32 x i8> [[B_HI]], <32 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <32 x i8> [[SEL_LO]], <32 x i8> [[SEL_HI]], <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <64 x i8> [[CONCAT]] to <8 x i64>
+; CHECK-NEXT: ret <8 x i64> [[RES]]
+;
+ %a.bc = bitcast <8 x i64> %a to <64 x i8>
+ %b.bc = bitcast <8 x i64> %b to <64 x i8>
+ %c.bc = bitcast <8 x i64> %a to <16 x i32>
+ %d.bc = bitcast <8 x i64> %b to <16 x i32>
+ %a.lo = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.lo = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %a.hi = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %b.hi = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %cmp = icmp slt <16 x i32> %c.bc, %d.bc
+ %sext = sext <16 x i1> %cmp to <16 x i32>
+ %sext.bc = bitcast <16 x i32> %sext to <64 x i8>
+ %sext.lo = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sext.hi = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %sel.lo = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.lo, <32 x i8> %b.lo, <32 x i8> %sext.lo)
+ %sel.hi = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.hi, <32 x i8> %b.hi, <32 x i8> %sext.hi)
+ %concat = shufflevector <32 x i8> %sel.lo, <32 x i8> %sel.hi, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %res = bitcast <64 x i8> %concat to <8 x i64>
+ ret <8 x i64> %res
+}
+
+define <8 x i64> @x86_pblendvb_v32i16_v16i16(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v32i16_v16i16(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <8 x i64> [[A:%.*]] to <64 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <8 x i64> [[B:%.*]] to <64 x i8>
+; CHECK-NEXT: [[C_BC:%.*]] = bitcast <8 x i64> [[C:%.*]] to <32 x i16>
+; CHECK-NEXT: [[D_BC:%.*]] = bitcast <8 x i64> [[D:%.*]] to <32 x i16>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <32 x i16> [[C_BC]], [[D_BC]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i16>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <32 x i16> [[SEXT]] to <64 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <64 x i8> [[SEXT_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_LO]], <32 x i8> [[B_LO]], <32 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_HI]], <32 x i8> [[B_HI]], <32 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <32 x i8> [[SEL_LO]], <32 x i8> [[SEL_HI]], <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <64 x i8> [[CONCAT]] to <8 x i64>
+; CHECK-NEXT: ret <8 x i64> [[RES]]
+;
+ %a.bc = bitcast <8 x i64> %a to <64 x i8>
+ %b.bc = bitcast <8 x i64> %b to <64 x i8>
+ %c.bc = bitcast <8 x i64> %c to <32 x i16>
+ %d.bc = bitcast <8 x i64> %d to <32 x i16>
+ %a.lo = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.lo = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %a.hi = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %b.hi = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %cmp = icmp slt <32 x i16> %c.bc, %d.bc
+ %sext = sext <32 x i1> %cmp to <32 x i16>
+ %sext.bc = bitcast <32 x i16> %sext to <64 x i8>
+ %sext.lo = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sext.hi = shufflevector <64 x i8> %sext.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %sel.lo = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.lo, <32 x i8> %b.lo, <32 x i8> %sext.lo)
+ %sel.hi = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.hi, <32 x i8> %b.hi, <32 x i8> %sext.hi)
+ %concat = shufflevector <32 x i8> %sel.lo, <32 x i8> %sel.hi, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %res = bitcast <64 x i8> %concat to <8 x i64>
+ ret <8 x i64> %res
+}
+
+define <8 x i64> @x86_pblendvb_v64i8_v32i8(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v64i8_v32i8(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <8 x i64> [[A:%.*]] to <64 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <8 x i64> [[B:%.*]] to <64 x i8>
+; CHECK-NEXT: [[C_BC:%.*]] = bitcast <8 x i64> [[C:%.*]] to <64 x i8>
+; CHECK-NEXT: [[D_BC:%.*]] = bitcast <8 x i64> [[D:%.*]] to <64 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[A_HI:%.*]] = shufflevector <64 x i8> [[A_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[B_HI:%.*]] = shufflevector <64 x i8> [[B_BC]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <64 x i8> [[C_BC]], [[D_BC]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <64 x i1> [[CMP]] to <64 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <64 x i8> [[SEXT]], <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; CHECK-NEXT: [[SEXT_HI:%.*]] = shufflevector <64 x i8> [[SEXT]], <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_LO]], <32 x i8> [[B_LO]], <32 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[SEL_HI:%.*]] = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> [[A_HI]], <32 x i8> [[B_HI]], <32 x i8> [[SEXT_HI]])
+; CHECK-NEXT: [[CONCAT:%.*]] = shufflevector <32 x i8> [[SEL_LO]], <32 x i8> [[SEL_HI]], <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; CHECK-NEXT: [[RES:%.*]] = bitcast <64 x i8> [[CONCAT]] to <8 x i64>
+; CHECK-NEXT: ret <8 x i64> [[RES]]
+;
+ %a.bc = bitcast <8 x i64> %a to <64 x i8>
+ %b.bc = bitcast <8 x i64> %b to <64 x i8>
+ %c.bc = bitcast <8 x i64> %c to <64 x i8>
+ %d.bc = bitcast <8 x i64> %d to <64 x i8>
+ %a.lo = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %b.lo = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %a.hi = shufflevector <64 x i8> %a.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %b.hi = shufflevector <64 x i8> %b.bc, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %cmp = icmp slt <64 x i8> %c.bc, %d.bc
+ %sext = sext <64 x i1> %cmp to <64 x i8>
+ %sext.lo = shufflevector <64 x i8> %sext, <64 x i8> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+ %sext.hi = shufflevector <64 x i8> %sext, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %sel.lo = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.lo, <32 x i8> %b.lo, <32 x i8> %sext.lo)
+ %sel.hi = tail call <32 x i8> @llvm.x86.avx2.pblendvb(<32 x i8> %a.hi, <32 x i8> %b.hi, <32 x i8> %sext.hi)
+ %concat = shufflevector <32 x i8> %sel.lo, <32 x i8> %sel.hi, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+ %res = bitcast <64 x i8> %concat to <8 x i64>
+ ret <8 x i64> %res
+}
+
+;
+; Negative Tests
+;
+
+; non-pow2 source
+define <2 x double> @x86_pblendvb_v3f64_v2f64(<3 x double> %a, <3 x double> %b, <3 x double> %c, <3 x double> %d) {
+; CHECK-LABEL: @x86_pblendvb_v3f64_v2f64(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <3 x double> [[A:%.*]] to <24 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <3 x double> [[B:%.*]] to <24 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <24 x i8> [[A_BC]], <24 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <24 x i8> [[B_BC]], <24 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[CMP:%.*]] = fcmp olt <3 x double> [[C:%.*]], [[D:%.*]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <3 x i1> [[CMP]] to <3 x i64>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <3 x i64> [[SEXT]] to <24 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <24 x i8> [[SEXT_BC]], <24 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[RES:%.*]] = bitcast <16 x i8> [[SEL_LO]] to <2 x double>
+; CHECK-NEXT: ret <2 x double> [[RES]]
+;
+ %a.bc = bitcast <3 x double> %a to <24 x i8>
+ %b.bc = bitcast <3 x double> %b to <24 x i8>
+ %a.lo = shufflevector <24 x i8> %a.bc, <24 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <24 x i8> %b.bc, <24 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %cmp = fcmp olt <3 x double> %c, %d
+ %sext = sext <3 x i1> %cmp to <3 x i64>
+ %sext.bc = bitcast <3 x i64> %sext to <24 x i8>
+ %sext.lo = shufflevector <24 x i8> %sext.bc, <24 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %res = bitcast <16 x i8> %sel.lo to <2 x double>
+ ret <2 x double> %res
+}
+
+; shuffle masks contain undef
+define <2 x i64> @x86_pblendvb_v32i8_v16i8_undefs(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v32i8_v16i8_undefs(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <4 x i64> [[A:%.*]] to <32 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <4 x i64> [[B:%.*]] to <32 x i8>
+; CHECK-NEXT: [[C_BC:%.*]] = bitcast <4 x i64> [[C:%.*]] to <32 x i8>
+; CHECK-NEXT: [[D_BC:%.*]] = bitcast <4 x i64> [[D:%.*]] to <32 x i8>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <32 x i8> [[C_BC]], [[D_BC]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <32 x i1> [[CMP]] to <32 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <32 x i8> [[SEXT]], <32 x i8> poison, <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 poison, i32 poison, i32 poison, i32 poison>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[RES:%.*]] = bitcast <16 x i8> [[SEL_LO]] to <2 x i64>
+; CHECK-NEXT: ret <2 x i64> [[RES]]
+;
+ %a.bc = bitcast <4 x i64> %a to <32 x i8>
+ %b.bc = bitcast <4 x i64> %b to <32 x i8>
+ %c.bc = bitcast <4 x i64> %c to <32 x i8>
+ %d.bc = bitcast <4 x i64> %d to <32 x i8>
+ %a.lo = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %cmp = icmp slt <32 x i8> %c.bc, %d.bc
+ %sext = sext <32 x i1> %cmp to <32 x i8>
+ %sext.lo = shufflevector <32 x i8> %sext, <32 x i8> poison, <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 undef, i32 undef, i32 undef, i32 undef>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %res = bitcast <16 x i8> %sel.lo to <2 x i64>
+ ret <2 x i64> %res
+}
+
+; masks elements too small
+define <2 x i64> @x86_pblendvb_v32i8_v32i4(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
+; CHECK-LABEL: @x86_pblendvb_v32i8_v32i4(
+; CHECK-NEXT: [[A_BC:%.*]] = bitcast <4 x i64> [[A:%.*]] to <32 x i8>
+; CHECK-NEXT: [[B_BC:%.*]] = bitcast <4 x i64> [[B:%.*]] to <32 x i8>
+; CHECK-NEXT: [[C_BC:%.*]] = bitcast <4 x i64> [[C:%.*]] to <64 x i4>
+; CHECK-NEXT: [[D_BC:%.*]] = bitcast <4 x i64> [[D:%.*]] to <64 x i4>
+; CHECK-NEXT: [[A_LO:%.*]] = shufflevector <32 x i8> [[A_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[B_LO:%.*]] = shufflevector <32 x i8> [[B_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt <64 x i4> [[C_BC]], [[D_BC]]
+; CHECK-NEXT: [[SEXT:%.*]] = sext <64 x i1> [[CMP]] to <64 x i4>
+; CHECK-NEXT: [[SEXT_BC:%.*]] = bitcast <64 x i4> [[SEXT]] to <32 x i8>
+; CHECK-NEXT: [[SEXT_LO:%.*]] = shufflevector <32 x i8> [[SEXT_BC]], <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: [[SEL_LO:%.*]] = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> [[A_LO]], <16 x i8> [[B_LO]], <16 x i8> [[SEXT_LO]])
+; CHECK-NEXT: [[RES:%.*]] = bitcast <16 x i8> [[SEL_LO]] to <2 x i64>
+; CHECK-NEXT: ret <2 x i64> [[RES]]
+;
+ %a.bc = bitcast <4 x i64> %a to <32 x i8>
+ %b.bc = bitcast <4 x i64> %b to <32 x i8>
+ %c.bc = bitcast <4 x i64> %c to <64 x i4>
+ %d.bc = bitcast <4 x i64> %d to <64 x i4>
+ %a.lo = shufflevector <32 x i8> %a.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %b.lo = shufflevector <32 x i8> %b.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %cmp = icmp slt <64 x i4> %c.bc, %d.bc
+ %sext = sext <64 x i1> %cmp to <64 x i4>
+ %sext.bc = bitcast <64 x i4> %sext to <32 x i8>
+ %sext.lo = shufflevector <32 x i8> %sext.bc, <32 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ %sel.lo = tail call <16 x i8> @llvm.x86.sse41.pblendvb(<16 x i8> %a.lo, <16 x i8> %b.lo, <16 x i8> %sext.lo)
+ %res = bitcast <16 x i8> %sel.lo to <2 x i64>
+ ret <2 x i64> %res
+}