summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--llvm/test/CodeGen/X86/shuffle-vs-trunc-128.ll481
-rw-r--r--llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll629
-rw-r--r--llvm/test/CodeGen/X86/shuffle-vs-trunc-512.ll537
3 files changed, 1647 insertions, 0 deletions
diff --git a/llvm/test/CodeGen/X86/shuffle-vs-trunc-128.ll b/llvm/test/CodeGen/X86/shuffle-vs-trunc-128.ll
new file mode 100644
index 00000000000..f9fe97b21ee
--- /dev/null
+++ b/llvm/test/CodeGen/X86/shuffle-vs-trunc-128.ll
@@ -0,0 +1,481 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=AVX --check-prefix=AVX1
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX --check-prefix=AVX2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512F
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512VL
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw,+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BWVL
+
+; PR31551
+; Pairs of shufflevector:trunc functions with functional equivalence.
+; Ideally, the shuffles should be lowered to code with the same quality as the truncates.
+
+define void @shuffle_v16i8_to_v8i8(<16 x i8>* %L, <8 x i8>* %S) nounwind {
+; AVX-LABEL: shuffle_v16i8_to_v8i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %xmm0
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v16i8_to_v8i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v16i8_to_v8i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512VL-NEXT: vmovq %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v16i8_to_v8i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v16i8_to_v8i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovwb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i8>, <16 x i8>* %L
+ %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+ store <8 x i8> %strided.vec, <8 x i8>* %S
+ ret void
+}
+
+define void @trunc_v8i16_to_v8i8(<16 x i8>* %L, <8 x i8>* %S) nounwind {
+; AVX-LABEL: trunc_v8i16_to_v8i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %xmm0
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v8i16_to_v8i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v8i16_to_v8i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512VL-NEXT: vmovq %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v8i16_to_v8i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v8i16_to_v8i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovwb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i8>, <16 x i8>* %L
+ %bc = bitcast <16 x i8> %vec to <8 x i16>
+ %strided.vec = trunc <8 x i16> %bc to <8 x i8>
+ store <8 x i8> %strided.vec, <8 x i8>* %S
+ ret void
+}
+
+define void @shuffle_v8i16_to_v4i16(<8 x i16>* %L, <4 x i16>* %S) nounwind {
+; AVX-LABEL: shuffle_v8i16_to_v4i16:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %xmm0
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v8i16_to_v4i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v8i16_to_v4i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovdw %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v8i16_to_v4i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v8i16_to_v4i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovdw %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <8 x i16>, <8 x i16>* %L
+ %strided.vec = shufflevector <8 x i16> %vec, <8 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+ store <4 x i16> %strided.vec, <4 x i16>* %S
+ ret void
+}
+
+define void @trunc_v4i32_to_v4i16(<8 x i16>* %L, <4 x i16>* %S) nounwind {
+; AVX-LABEL: trunc_v4i32_to_v4i16:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %xmm0
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v4i32_to_v4i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v4i32_to_v4i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovdw %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v4i32_to_v4i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v4i32_to_v4i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovdw %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <8 x i16>, <8 x i16>* %L
+ %bc = bitcast <8 x i16> %vec to <4 x i32>
+ %strided.vec = trunc <4 x i32> %bc to <4 x i16>
+ store <4 x i16> %strided.vec, <4 x i16>* %S
+ ret void
+}
+
+define void @shuffle_v4i32_to_v2i32(<4 x i32>* %L, <2 x i32>* %S) nounwind {
+; AVX-LABEL: shuffle_v4i32_to_v2i32:
+; AVX: # BB#0:
+; AVX-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v4i32_to_v2i32:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v4i32_to_v2i32:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovqd %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v4i32_to_v2i32:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v4i32_to_v2i32:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovqd %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <4 x i32>, <4 x i32>* %L
+ %strided.vec = shufflevector <4 x i32> %vec, <4 x i32> undef, <2 x i32> <i32 0, i32 2>
+ store <2 x i32> %strided.vec, <2 x i32>* %S
+ ret void
+}
+
+define void @trunc_v2i64_to_v2i32(<4 x i32>* %L, <2 x i32>* %S) nounwind {
+; AVX-LABEL: trunc_v2i64_to_v2i32:
+; AVX: # BB#0:
+; AVX-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v2i64_to_v2i32:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v2i64_to_v2i32:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovqd %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v2i64_to_v2i32:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v2i64_to_v2i32:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovqd %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <4 x i32>, <4 x i32>* %L
+ %bc = bitcast <4 x i32> %vec to <2 x i64>
+ %strided.vec = trunc <2 x i64> %bc to <2 x i32>
+ store <2 x i32> %strided.vec, <2 x i32>* %S
+ ret void
+}
+
+define void @shuffle_v16i8_to_v4i8(<16 x i8>* %L, <4 x i8>* %S) nounwind {
+; AVX-LABEL: shuffle_v16i8_to_v4i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %xmm0
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT: vmovd %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v16i8_to_v4i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vmovd %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v16i8_to_v4i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovdb %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v16i8_to_v4i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vmovd %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v16i8_to_v4i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovdb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i8>, <16 x i8>* %L
+ %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
+ store <4 x i8> %strided.vec, <4 x i8>* %S
+ ret void
+}
+
+define void @trunc_v4i32_to_v4i8(<16 x i8>* %L, <4 x i8>* %S) nounwind {
+; AVX-LABEL: trunc_v4i32_to_v4i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %xmm0
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT: vmovd %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v4i32_to_v4i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vmovd %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v4i32_to_v4i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovdb %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v4i32_to_v4i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vmovd %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v4i32_to_v4i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovdb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i8>, <16 x i8>* %L
+ %bc = bitcast <16 x i8> %vec to <4 x i32>
+ %strided.vec = trunc <4 x i32> %bc to <4 x i8>
+ store <4 x i8> %strided.vec, <4 x i8>* %S
+ ret void
+}
+
+define void @shuffle_v8i16_to_v2i16(<8 x i16>* %L, <2 x i16>* %S) nounwind {
+; AVX-LABEL: shuffle_v8i16_to_v2i16:
+; AVX: # BB#0:
+; AVX-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX-NEXT: vmovd %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v8i16_to_v2i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX512F-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT: vmovd %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v8i16_to_v2i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovqw %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v8i16_to_v2i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX512BW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512BW-NEXT: vmovd %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v8i16_to_v2i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovqw %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <8 x i16>, <8 x i16>* %L
+ %strided.vec = shufflevector <8 x i16> %vec, <8 x i16> undef, <2 x i32> <i32 0, i32 4>
+ store <2 x i16> %strided.vec, <2 x i16>* %S
+ ret void
+}
+
+define void @trunc_v2i64_to_v2i16(<8 x i16>* %L, <2 x i16>* %S) nounwind {
+; AVX-LABEL: trunc_v2i64_to_v2i16:
+; AVX: # BB#0:
+; AVX-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX-NEXT: vmovd %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v2i64_to_v2i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX512F-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT: vmovd %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v2i64_to_v2i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovqw %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v2i64_to_v2i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3]
+; AVX512BW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512BW-NEXT: vmovd %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v2i64_to_v2i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovqw %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <8 x i16>, <8 x i16>* %L
+ %bc = bitcast <8 x i16> %vec to <2 x i64>
+ %strided.vec = trunc <2 x i64> %bc to <2 x i16>
+ store <2 x i16> %strided.vec, <2 x i16>* %S
+ ret void
+}
+
+define void @shuffle_v16i8_to_v2i8(<16 x i8>* %L, <2 x i8>* %S) nounwind {
+; AVX-LABEL: shuffle_v16i8_to_v2i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %xmm0
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT: vpextrw $0, %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v16i8_to_v2i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vpextrw $0, %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v16i8_to_v2i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovqb %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v16i8_to_v2i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vpextrw $0, %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v16i8_to_v2i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovqb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i8>, <16 x i8>* %L
+ %strided.vec = shufflevector <16 x i8> %vec, <16 x i8> undef, <2 x i32> <i32 0, i32 8>
+ store <2 x i8> %strided.vec, <2 x i8>* %S
+ ret void
+}
+
+define void @trunc_v2i64_to_v2i8(<16 x i8>* %L, <2 x i8>* %S) nounwind {
+; AVX-LABEL: trunc_v2i64_to_v2i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %xmm0
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT: vpextrw $0, %xmm0, (%rsi)
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v2i64_to_v2i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vpextrw $0, %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v2i64_to_v2i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512VL-NEXT: vpmovqb %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v2i64_to_v2i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vpextrw $0, %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v2i64_to_v2i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0
+; AVX512BWVL-NEXT: vpmovqb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i8>, <16 x i8>* %L
+ %bc = bitcast <16 x i8> %vec to <2 x i64>
+ %strided.vec = trunc <2 x i64> %bc to <2 x i8>
+ store <2 x i8> %strided.vec, <2 x i8>* %S
+ ret void
+}
diff --git a/llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll b/llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll
new file mode 100644
index 00000000000..893f96e6fb2
--- /dev/null
+++ b/llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll
@@ -0,0 +1,629 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX --check-prefix=AVX2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512F
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512VL
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw,+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BWVL
+
+; PR31551
+; Pairs of shufflevector:trunc functions with functional equivalence.
+; Ideally, the shuffles should be lowered to code with the same quality as the truncates.
+
+define void @shuffle_v32i8_to_v16i8(<32 x i8>* %L, <16 x i8>* %S) nounwind {
+; AVX-LABEL: shuffle_v32i8_to_v16i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %ymm0
+; AVX-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v32i8_to_v16i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512F-NEXT: vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX512F-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512F-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512F-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512F-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v32i8_to_v16i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512VL-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v32i8_to_v16i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BW-NEXT: vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX512BW-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512BW-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512BW-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512BW-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v32i8_to_v16i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu (%rdi), %ymm0
+; AVX512BWVL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BWVL-NEXT: vmovdqu {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX512BWVL-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512BWVL-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512BWVL-NEXT: vmovdqu %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <32 x i8>, <32 x i8>* %L
+ %strided.vec = shufflevector <32 x i8> %vec, <32 x i8> undef, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+ store <16 x i8> %strided.vec, <16 x i8>* %S
+ ret void
+}
+
+define void @trunc_v16i16_to_v16i8(<32 x i8>* %L, <16 x i8>* %S) nounwind {
+; AVX-LABEL: trunc_v16i16_to_v16i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %ymm0
+; AVX-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
+; AVX-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v16i16_to_v16i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vpmovsxwd (%rdi), %zmm0
+; AVX512F-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512F-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v16i16_to_v16i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vpmovsxwd (%rdi), %zmm0
+; AVX512VL-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512VL-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v16i16_to_v16i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
+; AVX512BW-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v16i16_to_v16i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu (%rdi), %ymm0
+; AVX512BWVL-NEXT: vpmovwb %ymm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <32 x i8>, <32 x i8>* %L
+ %bc = bitcast <32 x i8> %vec to <16 x i16>
+ %strided.vec = trunc <16 x i16> %bc to <16 x i8>
+ store <16 x i8> %strided.vec, <16 x i8>* %S
+ ret void
+}
+
+define void @shuffle_v16i16_to_v8i16(<16 x i16>* %L, <8 x i16>* %S) nounwind {
+; AVX-LABEL: shuffle_v16i16_to_v8i16:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %ymm0
+; AVX-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v16i16_to_v8i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512F-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX512F-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512F-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512F-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512F-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v16i16_to_v8i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512VL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; AVX512VL-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,6,7]
+; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; AVX512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512VL-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
+; AVX512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512VL-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512VL-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v16i16_to_v8i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BW-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX512BW-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512BW-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512BW-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512BW-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v16i16_to_v8i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu (%rdi), %ymm0
+; AVX512BWVL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BWVL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; AVX512BWVL-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,6,7]
+; AVX512BWVL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; AVX512BWVL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512BWVL-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
+; AVX512BWVL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BWVL-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512BWVL-NEXT: vmovdqu %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i16>, <16 x i16>* %L
+ %strided.vec = shufflevector <16 x i16> %vec, <16 x i16> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+ store <8 x i16> %strided.vec, <8 x i16>* %S
+ ret void
+}
+
+define void @trunc_v8i32_to_v8i16(<16 x i16>* %L, <8 x i16>* %S) nounwind {
+; AVX-LABEL: trunc_v8i32_to_v8i16:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %ymm0
+; AVX-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,20,21,24,25,28,29],zero,zero,zero,zero,zero,zero,zero,zero
+; AVX-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v8i32_to_v8i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vpmovdw %zmm0, %ymm0
+; AVX512F-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v8i32_to_v8i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vpmovdw %ymm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v8i32_to_v8i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vpmovdw %zmm0, %ymm0
+; AVX512BW-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v8i32_to_v8i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BWVL-NEXT: vpmovdw %ymm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i16>, <16 x i16>* %L
+ %bc = bitcast <16 x i16> %vec to <8 x i32>
+ %strided.vec = trunc <8 x i32> %bc to <8 x i16>
+ store <8 x i16> %strided.vec, <8 x i16>* %S
+ ret void
+}
+
+define void @shuffle_v8i32_to_v4i32(<8 x i32>* %L, <4 x i32>* %S) nounwind {
+; AVX-LABEL: shuffle_v8i32_to_v4i32:
+; AVX: # BB#0:
+; AVX-NEXT: vmovaps (%rdi), %ymm0
+; AVX-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX-NEXT: vmovaps %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v8i32_to_v4i32:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovaps (%rdi), %ymm0
+; AVX512F-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX512F-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX512F-NEXT: vmovaps %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v8i32_to_v4i32:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512VL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX512VL-NEXT: vmovaps %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v8i32_to_v4i32:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovaps (%rdi), %ymm0
+; AVX512BW-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX512BW-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX512BW-NEXT: vmovaps %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v8i32_to_v4i32:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BWVL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BWVL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX512BWVL-NEXT: vmovaps %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <8 x i32>, <8 x i32>* %L
+ %strided.vec = shufflevector <8 x i32> %vec, <8 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+ store <4 x i32> %strided.vec, <4 x i32>* %S
+ ret void
+}
+
+define void @trunc_v4i64_to_v4i32(<8 x i32>* %L, <4 x i32>* %S) nounwind {
+; AVX-LABEL: trunc_v4i64_to_v4i32:
+; AVX: # BB#0:
+; AVX-NEXT: vpshufd {{.*#+}} ymm0 = mem[0,2,2,3,4,6,6,7]
+; AVX-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v4i64_to_v4i32:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vpmovqd %zmm0, %ymm0
+; AVX512F-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v4i64_to_v4i32:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vpmovqd %ymm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v4i64_to_v4i32:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0
+; AVX512BW-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v4i64_to_v4i32:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BWVL-NEXT: vpmovqd %ymm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <8 x i32>, <8 x i32>* %L
+ %bc = bitcast <8 x i32> %vec to <4 x i64>
+ %strided.vec = trunc <4 x i64> %bc to <4 x i32>
+ store <4 x i32> %strided.vec, <4 x i32>* %S
+ ret void
+}
+
+define void @shuffle_v32i8_to_v8i8(<32 x i8>* %L, <8 x i8>* %S) nounwind {
+; AVX-LABEL: shuffle_v32i8_to_v8i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %ymm0
+; AVX-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v32i8_to_v8i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512F-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512F-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512F-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512F-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v32i8_to_v8i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512VL-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX512VL-NEXT: vmovq %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v32i8_to_v8i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BW-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512BW-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512BW-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512BW-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v32i8_to_v8i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu (%rdi), %ymm0
+; AVX512BWVL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BWVL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; AVX512BWVL-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,6,7]
+; AVX512BWVL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; AVX512BWVL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512BWVL-NEXT: vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
+; AVX512BWVL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BWVL-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX512BWVL-NEXT: vpmovwb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <32 x i8>, <32 x i8>* %L
+ %strided.vec = shufflevector <32 x i8> %vec, <32 x i8> undef, <8 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28>
+ store <8 x i8> %strided.vec, <8 x i8>* %S
+ ret void
+}
+
+define void @trunc_v8i32_to_v8i8(<32 x i8>* %L, <8 x i8>* %S) nounwind {
+; AVX-LABEL: trunc_v8i32_to_v8i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %ymm0
+; AVX-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,20,21,24,25,28,29],zero,zero,zero,zero,zero,zero,zero,zero
+; AVX-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v8i32_to_v8i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vpmovdw %zmm0, %ymm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v8i32_to_v8i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vpmovdb %ymm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v8i32_to_v8i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vpmovdw %zmm0, %ymm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v8i32_to_v8i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BWVL-NEXT: vpmovdb %ymm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <32 x i8>, <32 x i8>* %L
+ %bc = bitcast <32 x i8> %vec to <8 x i32>
+ %strided.vec = trunc <8 x i32> %bc to <8 x i8>
+ store <8 x i8> %strided.vec, <8 x i8>* %S
+ ret void
+}
+
+define void @shuffle_v16i16_to_v4i16(<16 x i16>* %L, <4 x i16>* %S) nounwind {
+; AVX-LABEL: shuffle_v16i16_to_v4i16:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %ymm0
+; AVX-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; AVX-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; AVX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v16i16_to_v4i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; AVX512F-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v16i16_to_v4i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512VL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX512VL-NEXT: vpmovdw %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v16i16_to_v4i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BW-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; AVX512BW-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; AVX512BW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512BW-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v16i16_to_v4i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu (%rdi), %ymm0
+; AVX512BWVL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BWVL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX512BWVL-NEXT: vpmovdw %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i16>, <16 x i16>* %L
+ %strided.vec = shufflevector <16 x i16> %vec, <16 x i16> undef, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
+ store <4 x i16> %strided.vec, <4 x i16>* %S
+ ret void
+}
+
+define void @trunc_v4i64_to_v4i16(<16 x i16>* %L, <4 x i16>* %S) nounwind {
+; AVX-LABEL: trunc_v4i64_to_v4i16:
+; AVX: # BB#0:
+; AVX-NEXT: vpshufd {{.*#+}} ymm0 = mem[0,2,2,3,4,6,6,7]
+; AVX-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX-NEXT: vmovq %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v4i64_to_v4i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vpmovqd %zmm0, %ymm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v4i64_to_v4i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vpmovqw %ymm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v4i64_to_v4i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v4i64_to_v4i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BWVL-NEXT: vpmovqw %ymm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <16 x i16>, <16 x i16>* %L
+ %bc = bitcast <16 x i16> %vec to <4 x i64>
+ %strided.vec = trunc <4 x i64> %bc to <4 x i16>
+ store <4 x i16> %strided.vec, <4 x i16>* %S
+ ret void
+}
+
+define void @shuffle_v32i8_to_v4i8(<32 x i8>* %L, <4 x i8>* %S) nounwind {
+; AVX-LABEL: shuffle_v32i8_to_v4i8:
+; AVX: # BB#0:
+; AVX-NEXT: vmovdqa (%rdi), %ymm0
+; AVX-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX-NEXT: vmovd %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: shuffle_v32i8_to_v4i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512F-NEXT: vmovdqa {{.*#+}} xmm2 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512F-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512F-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512F-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512F-NEXT: vmovd %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v32i8_to_v4i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512VL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX512VL-NEXT: vpmovdb %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v32i8_to_v4i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BW-NEXT: vmovdqa {{.*#+}} xmm2 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512BW-NEXT: vpshufb %xmm2, %xmm1, %xmm1
+; AVX512BW-NEXT: vpshufb %xmm2, %xmm0, %xmm0
+; AVX512BW-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512BW-NEXT: vmovd %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v32i8_to_v4i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu (%rdi), %ymm0
+; AVX512BWVL-NEXT: vextracti128 $1, %ymm0, %xmm1
+; AVX512BWVL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX512BWVL-NEXT: vpmovdb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <32 x i8>, <32 x i8>* %L
+ %strided.vec = shufflevector <32 x i8> %vec, <32 x i8> undef, <4 x i32> <i32 0, i32 8, i32 16, i32 24>
+ store <4 x i8> %strided.vec, <4 x i8>* %S
+ ret void
+}
+
+define void @trunc_v4i64_to_v4i8(<32 x i8>* %L, <4 x i8>* %S) nounwind {
+; AVX-LABEL: trunc_v4i64_to_v4i8:
+; AVX: # BB#0:
+; AVX-NEXT: vpshufd {{.*#+}} ymm0 = mem[0,2,2,3,4,6,6,7]
+; AVX-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
+; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT: vmovd %xmm0, (%rsi)
+; AVX-NEXT: vzeroupper
+; AVX-NEXT: retq
+;
+; AVX512F-LABEL: trunc_v4i64_to_v4i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vpmovqd %zmm0, %ymm0
+; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vmovd %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v4i64_to_v4i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vpmovqb %ymm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v4i64_to_v4i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0
+; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vmovd %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v4i64_to_v4i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512BWVL-NEXT: vpmovqb %ymm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <32 x i8>, <32 x i8>* %L
+ %bc = bitcast <32 x i8> %vec to <4 x i64>
+ %strided.vec = trunc <4 x i64> %bc to <4 x i8>
+ store <4 x i8> %strided.vec, <4 x i8>* %S
+ ret void
+}
diff --git a/llvm/test/CodeGen/X86/shuffle-vs-trunc-512.ll b/llvm/test/CodeGen/X86/shuffle-vs-trunc-512.ll
new file mode 100644
index 00000000000..923290411ae
--- /dev/null
+++ b/llvm/test/CodeGen/X86/shuffle-vs-trunc-512.ll
@@ -0,0 +1,537 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512F
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512VL
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw,+avx512vl | FileCheck %s --check-prefix=AVX512 --check-prefix=AVX512BWVL
+
+; PR31551
+; Pairs of shufflevector:trunc functions with functional equivalence.
+; Ideally, the shuffles should be lowered to code with the same quality as the truncates.
+
+define void @shuffle_v64i8_to_v32i8(<64 x i8>* %L, <32 x i8>* %S) nounwind {
+; AVX512F-LABEL: shuffle_v64i8_to_v32i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512F-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30]
+; AVX512F-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
+; AVX512F-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512F-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v64i8_to_v32i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512VL-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30]
+; AVX512VL-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u]
+; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
+; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512VL-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v64i8_to_v32i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqu8 (%rdi), %zmm0
+; AVX512BW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512BW-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30]
+; AVX512BW-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
+; AVX512BW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512BW-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v64i8_to_v32i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu8 (%rdi), %zmm0
+; AVX512BWVL-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512BWVL-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30]
+; AVX512BWVL-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u,16,18,20,22,24,26,28,30,u,u,u,u,u,u,u,u]
+; AVX512BWVL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
+; AVX512BWVL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512BWVL-NEXT: vmovdqu %ymm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <64 x i8>, <64 x i8>* %L
+ %strided.vec = shufflevector <64 x i8> %vec, <64 x i8> undef, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
+ store <32 x i8> %strided.vec, <32 x i8>* %S
+ ret void
+}
+
+define void @trunc_v32i16_to_v32i8(<64 x i8>* %L, <32 x i8>* %S) nounwind {
+; AVX512F-LABEL: trunc_v32i16_to_v32i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vpmovsxwd (%rdi), %zmm0
+; AVX512F-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512F-NEXT: vpmovsxwd 32(%rdi), %zmm1
+; AVX512F-NEXT: vpmovdb %zmm1, %xmm1
+; AVX512F-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX512F-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: trunc_v32i16_to_v32i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vpmovsxwd (%rdi), %zmm0
+; AVX512VL-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512VL-NEXT: vpmovsxwd 32(%rdi), %zmm1
+; AVX512VL-NEXT: vpmovdb %zmm1, %xmm1
+; AVX512VL-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX512VL-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: trunc_v32i16_to_v32i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqu16 (%rdi), %zmm0
+; AVX512BW-NEXT: vpmovwb %zmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: trunc_v32i16_to_v32i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu16 (%rdi), %zmm0
+; AVX512BWVL-NEXT: vpmovwb %zmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <64 x i8>, <64 x i8>* %L
+ %bc = bitcast <64 x i8> %vec to <32 x i16>
+ %strided.vec = trunc <32 x i16> %bc to <32 x i8>
+ store <32 x i8> %strided.vec, <32 x i8>* %S
+ ret void
+}
+
+define void @shuffle_v32i16_to_v16i16(<32 x i16>* %L, <16 x i16>* %S) nounwind {
+; AVX512F-LABEL: shuffle_v32i16_to_v16i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512F-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,12,13,16,17,20,21,20,21,22,23,16,17,20,21,24,25,28,29]
+; AVX512F-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
+; AVX512F-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
+; AVX512F-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512F-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v32i16_to_v16i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vpshuflw {{.*#+}} ymm0 = mem[0,2,2,3,4,5,6,7,8,10,10,11,12,13,14,15]
+; AVX512VL-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,4,6,6,7,8,9,10,11,12,14,14,15]
+; AVX512VL-NEXT: vpshufd {{.*#+}} ymm0 = ymm0[0,1,0,2,4,5,4,6]
+; AVX512VL-NEXT: vpshuflw {{.*#+}} ymm1 = mem[0,2,2,3,4,5,6,7,8,10,10,11,12,13,14,15]
+; AVX512VL-NEXT: vpshufhw {{.*#+}} ymm1 = ymm1[0,1,2,3,4,6,6,7,8,9,10,11,12,14,14,15]
+; AVX512VL-NEXT: vpshufd {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7]
+; AVX512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
+; AVX512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512VL-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v32i16_to_v16i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqu16 (%rdi), %zmm0
+; AVX512BW-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512BW-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,12,13,16,17,20,21,20,21,22,23,16,17,20,21,24,25,28,29]
+; AVX512BW-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
+; AVX512BW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
+; AVX512BW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512BW-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v32i16_to_v16i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu16 (%rdi), %zmm0
+; AVX512BWVL-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512BWVL-NEXT: vmovdqu {{.*#+}} ymm2 = [0,2,4,6,16,18,20,22,8,10,12,14,24,26,28,30]
+; AVX512BWVL-NEXT: vpermi2w %ymm1, %ymm0, %ymm2
+; AVX512BWVL-NEXT: vpermq {{.*#+}} ymm0 = ymm2[0,2,1,3]
+; AVX512BWVL-NEXT: vmovdqu %ymm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <32 x i16>, <32 x i16>* %L
+ %strided.vec = shufflevector <32 x i16> %vec, <32 x i16> undef, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+ store <16 x i16> %strided.vec, <16 x i16>* %S
+ ret void
+}
+
+define void @trunc_v16i32_to_v16i16(<32 x i16>* %L, <16 x i16>* %S) nounwind {
+; AVX512-LABEL: trunc_v16i32_to_v16i16:
+; AVX512: # BB#0:
+; AVX512-NEXT: vmovdqa32 (%rdi), %zmm0
+; AVX512-NEXT: vpmovdw %zmm0, (%rsi)
+; AVX512-NEXT: retq
+ %vec = load <32 x i16>, <32 x i16>* %L
+ %bc = bitcast <32 x i16> %vec to <16 x i32>
+ %strided.vec = trunc <16 x i32> %bc to <16 x i16>
+ store <16 x i16> %strided.vec, <16 x i16>* %S
+ ret void
+}
+
+define void @shuffle_v16i32_to_v8i32(<16 x i32>* %L, <8 x i32>* %S) nounwind {
+; AVX512-LABEL: shuffle_v16i32_to_v8i32:
+; AVX512: # BB#0:
+; AVX512-NEXT: vmovdqa32 (%rdi), %zmm0
+; AVX512-NEXT: vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT: vshufps {{.*#+}} ymm0 = ymm0[0,2],ymm1[0,2],ymm0[4,6],ymm1[4,6]
+; AVX512-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512-NEXT: vmovdqa %ymm0, (%rsi)
+; AVX512-NEXT: retq
+ %vec = load <16 x i32>, <16 x i32>* %L
+ %strided.vec = shufflevector <16 x i32> %vec, <16 x i32> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+ store <8 x i32> %strided.vec, <8 x i32>* %S
+ ret void
+}
+
+define void @trunc_v8i64_to_v8i32(<16 x i32>* %L, <8 x i32>* %S) nounwind {
+; AVX512-LABEL: trunc_v8i64_to_v8i32:
+; AVX512: # BB#0:
+; AVX512-NEXT: vmovdqa64 (%rdi), %zmm0
+; AVX512-NEXT: vpmovqd %zmm0, (%rsi)
+; AVX512-NEXT: retq
+ %vec = load <16 x i32>, <16 x i32>* %L
+ %bc = bitcast <16 x i32> %vec to <8 x i64>
+ %strided.vec = trunc <8 x i64> %bc to <8 x i32>
+ store <8 x i32> %strided.vec, <8 x i32>* %S
+ ret void
+}
+
+define void @shuffle_v64i8_to_v16i8(<64 x i8>* %L, <16 x i8>* %S) nounwind {
+; AVX512F-LABEL: shuffle_v64i8_to_v16i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512F-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX512F-NEXT: vmovdqa {{.*#+}} xmm3 = <u,u,u,u,0,4,8,12,u,u,u,u,u,u,u,u>
+; AVX512F-NEXT: vpshufb %xmm3, %xmm2, %xmm2
+; AVX512F-NEXT: vpshufb %xmm3, %xmm1, %xmm1
+; AVX512F-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm2
+; AVX512F-NEXT: vmovdqa {{.*#+}} xmm3 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512F-NEXT: vpshufb %xmm3, %xmm2, %xmm2
+; AVX512F-NEXT: vpshufb %xmm3, %xmm0, %xmm0
+; AVX512F-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; AVX512F-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
+; AVX512F-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v64i8_to_v16i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512VL-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm3 = <u,u,u,u,0,4,8,12,u,u,u,u,u,u,u,u>
+; AVX512VL-NEXT: vpshufb %xmm3, %xmm2, %xmm2
+; AVX512VL-NEXT: vpshufb %xmm3, %xmm1, %xmm1
+; AVX512VL-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm2
+; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm3 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512VL-NEXT: vpshufb %xmm3, %xmm2, %xmm2
+; AVX512VL-NEXT: vpshufb %xmm3, %xmm0, %xmm0
+; AVX512VL-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; AVX512VL-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
+; AVX512VL-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v64i8_to_v16i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqu8 (%rdi), %zmm0
+; AVX512BW-NEXT: vpextrb $4, %xmm0, %eax
+; AVX512BW-NEXT: vpextrb $0, %xmm0, %ecx
+; AVX512BW-NEXT: vmovd %ecx, %xmm1
+; AVX512BW-NEXT: vpinsrb $1, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $8, %xmm0, %eax
+; AVX512BW-NEXT: vpinsrb $2, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $12, %xmm0, %eax
+; AVX512BW-NEXT: vpinsrb $3, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vextracti32x4 $1, %zmm0, %xmm2
+; AVX512BW-NEXT: vpextrb $0, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrb $4, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $4, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrb $5, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $8, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrb $6, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $12, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrb $7, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vextracti32x4 $2, %zmm0, %xmm2
+; AVX512BW-NEXT: vpextrb $0, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrb $8, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $4, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrb $9, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $8, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrb $10, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $12, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrb $11, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vextracti32x4 $3, %zmm0, %xmm0
+; AVX512BW-NEXT: vpextrb $0, %xmm0, %eax
+; AVX512BW-NEXT: vpinsrb $12, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $4, %xmm0, %eax
+; AVX512BW-NEXT: vpinsrb $13, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $8, %xmm0, %eax
+; AVX512BW-NEXT: vpinsrb $14, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrb $12, %xmm0, %eax
+; AVX512BW-NEXT: vpinsrb $15, %eax, %xmm1, %xmm0
+; AVX512BW-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v64i8_to_v16i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu8 (%rdi), %zmm0
+; AVX512BWVL-NEXT: vpextrb $4, %xmm0, %eax
+; AVX512BWVL-NEXT: vpextrb $0, %xmm0, %ecx
+; AVX512BWVL-NEXT: vmovd %ecx, %xmm1
+; AVX512BWVL-NEXT: vpinsrb $1, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $8, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $2, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $12, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $3, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $1, %zmm0, %xmm2
+; AVX512BWVL-NEXT: vpextrb $0, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $4, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $4, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $5, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $8, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $6, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $12, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $7, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $2, %zmm0, %xmm2
+; AVX512BWVL-NEXT: vpextrb $0, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $8, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $4, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $9, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $8, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $10, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $12, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $11, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $3, %zmm0, %xmm0
+; AVX512BWVL-NEXT: vpextrb $0, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $12, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $4, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $13, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $8, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $14, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $12, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $15, %eax, %xmm1, %xmm0
+; AVX512BWVL-NEXT: vmovdqu %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <64 x i8>, <64 x i8>* %L
+ %strided.vec = shufflevector <64 x i8> %vec, <64 x i8> undef, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 32, i32 36, i32 40, i32 44, i32 48, i32 52, i32 56, i32 60>
+ store <16 x i8> %strided.vec, <16 x i8>* %S
+ ret void
+}
+
+define void @trunc_v16i32_to_v16i8(<64 x i8>* %L, <16 x i8>* %S) nounwind {
+; AVX512-LABEL: trunc_v16i32_to_v16i8:
+; AVX512: # BB#0:
+; AVX512-NEXT: vmovdqa32 (%rdi), %zmm0
+; AVX512-NEXT: vpmovdb %zmm0, (%rsi)
+; AVX512-NEXT: retq
+ %vec = load <64 x i8>, <64 x i8>* %L
+ %bc = bitcast <64 x i8> %vec to <16 x i32>
+ %strided.vec = trunc <16 x i32> %bc to <16 x i8>
+ store <16 x i8> %strided.vec, <16 x i8>* %S
+ ret void
+}
+
+define void @shuffle_v32i16_to_v8i16(<32 x i16>* %L, <8 x i16>* %S) nounwind {
+; AVX512F-LABEL: shuffle_v32i16_to_v8i16:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512F-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
+; AVX512F-NEXT: vpshuflw {{.*#+}} xmm2 = xmm2[0,1,0,2,4,5,6,7]
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; AVX512F-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,1,0,2,4,5,6,7]
+; AVX512F-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm2
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
+; AVX512F-NEXT: vpshuflw {{.*#+}} xmm2 = xmm2[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; AVX512F-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
+; AVX512F-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v32i16_to_v8i16:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512VL-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX512VL-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
+; AVX512VL-NEXT: vpshuflw {{.*#+}} xmm2 = xmm2[0,1,0,2,4,5,6,7]
+; AVX512VL-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; AVX512VL-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,1,0,2,4,5,6,7]
+; AVX512VL-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm2
+; AVX512VL-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
+; AVX512VL-NEXT: vpshuflw {{.*#+}} xmm2 = xmm2[0,2,2,3,4,5,6,7]
+; AVX512VL-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512VL-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512VL-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; AVX512VL-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
+; AVX512VL-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v32i16_to_v8i16:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqu16 (%rdi), %zmm0
+; AVX512BW-NEXT: vpextrw $4, %xmm0, %eax
+; AVX512BW-NEXT: vpinsrw $1, %eax, %xmm0, %xmm1
+; AVX512BW-NEXT: vextracti32x4 $1, %zmm0, %xmm2
+; AVX512BW-NEXT: vmovd %xmm2, %eax
+; AVX512BW-NEXT: vpinsrw $2, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrw $4, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrw $3, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vextracti32x4 $2, %zmm0, %xmm2
+; AVX512BW-NEXT: vmovd %xmm2, %eax
+; AVX512BW-NEXT: vpinsrw $4, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrw $4, %xmm2, %eax
+; AVX512BW-NEXT: vpinsrw $5, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vextracti32x4 $3, %zmm0, %xmm0
+; AVX512BW-NEXT: vmovd %xmm0, %eax
+; AVX512BW-NEXT: vpinsrw $6, %eax, %xmm1, %xmm1
+; AVX512BW-NEXT: vpextrw $4, %xmm0, %eax
+; AVX512BW-NEXT: vpinsrw $7, %eax, %xmm1, %xmm0
+; AVX512BW-NEXT: vmovdqa %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v32i16_to_v8i16:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu16 (%rdi), %zmm0
+; AVX512BWVL-NEXT: vpextrw $4, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrw $1, %eax, %xmm0, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $1, %zmm0, %xmm2
+; AVX512BWVL-NEXT: vmovd %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrw $2, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrw $4, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrw $3, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $2, %zmm0, %xmm2
+; AVX512BWVL-NEXT: vmovd %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrw $4, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrw $4, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrw $5, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $3, %zmm0, %xmm0
+; AVX512BWVL-NEXT: vmovd %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrw $6, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrw $4, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrw $7, %eax, %xmm1, %xmm0
+; AVX512BWVL-NEXT: vmovdqu %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <32 x i16>, <32 x i16>* %L
+ %strided.vec = shufflevector <32 x i16> %vec, <32 x i16> undef, <8 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28>
+ store <8 x i16> %strided.vec, <8 x i16>* %S
+ ret void
+}
+
+define void @trunc_v8i64_to_v8i16(<32 x i16>* %L, <8 x i16>* %S) nounwind {
+; AVX512-LABEL: trunc_v8i64_to_v8i16:
+; AVX512: # BB#0:
+; AVX512-NEXT: vmovdqa64 (%rdi), %zmm0
+; AVX512-NEXT: vpmovqw %zmm0, (%rsi)
+; AVX512-NEXT: retq
+ %vec = load <32 x i16>, <32 x i16>* %L
+ %bc = bitcast <32 x i16> %vec to <8 x i64>
+ %strided.vec = trunc <8 x i64> %bc to <8 x i16>
+ store <8 x i16> %strided.vec, <8 x i16>* %S
+ ret void
+}
+
+define void @shuffle_v64i8_to_v8i8(<64 x i8>* %L, <8 x i8>* %S) nounwind {
+; AVX512F-LABEL: shuffle_v64i8_to_v8i8:
+; AVX512F: # BB#0:
+; AVX512F-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512F-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512F-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX512F-NEXT: vmovdqa {{.*#+}} xmm3 = <u,u,0,8,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512F-NEXT: vpshufb %xmm3, %xmm2, %xmm2
+; AVX512F-NEXT: vpshufb %xmm3, %xmm1, %xmm1
+; AVX512F-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm2
+; AVX512F-NEXT: vmovdqa {{.*#+}} xmm3 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512F-NEXT: vpshufb %xmm3, %xmm2, %xmm2
+; AVX512F-NEXT: vpshufb %xmm3, %xmm0, %xmm0
+; AVX512F-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; AVX512F-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX512F-NEXT: vmovq %xmm0, (%rsi)
+; AVX512F-NEXT: retq
+;
+; AVX512VL-LABEL: shuffle_v64i8_to_v8i8:
+; AVX512VL: # BB#0:
+; AVX512VL-NEXT: vmovdqa (%rdi), %ymm0
+; AVX512VL-NEXT: vmovdqa 32(%rdi), %ymm1
+; AVX512VL-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm3 = <u,u,0,8,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512VL-NEXT: vpshufb %xmm3, %xmm2, %xmm2
+; AVX512VL-NEXT: vpshufb %xmm3, %xmm1, %xmm1
+; AVX512VL-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; AVX512VL-NEXT: vextracti128 $1, %ymm0, %xmm2
+; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm3 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
+; AVX512VL-NEXT: vpshufb %xmm3, %xmm2, %xmm2
+; AVX512VL-NEXT: vpshufb %xmm3, %xmm0, %xmm0
+; AVX512VL-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; AVX512VL-NEXT: vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX512VL-NEXT: vmovq %xmm0, (%rsi)
+; AVX512VL-NEXT: retq
+;
+; AVX512BW-LABEL: shuffle_v64i8_to_v8i8:
+; AVX512BW: # BB#0:
+; AVX512BW-NEXT: vmovdqu8 (%rdi), %zmm0
+; AVX512BW-NEXT: vextracti32x4 $3, %zmm0, %xmm1
+; AVX512BW-NEXT: vpextrb $8, %xmm1, %r8d
+; AVX512BW-NEXT: vpextrb $0, %xmm1, %r9d
+; AVX512BW-NEXT: vextracti32x4 $2, %zmm0, %xmm1
+; AVX512BW-NEXT: vpextrb $8, %xmm1, %r10d
+; AVX512BW-NEXT: vpextrb $0, %xmm1, %r11d
+; AVX512BW-NEXT: vextracti32x4 $1, %zmm0, %xmm1
+; AVX512BW-NEXT: vpextrb $8, %xmm1, %eax
+; AVX512BW-NEXT: vpextrb $0, %xmm1, %ecx
+; AVX512BW-NEXT: vpextrb $8, %xmm0, %edx
+; AVX512BW-NEXT: vpextrb $0, %xmm0, %edi
+; AVX512BW-NEXT: vpinsrb $0, %edi, %xmm0, %xmm0
+; AVX512BW-NEXT: vpinsrb $1, %edx, %xmm0, %xmm0
+; AVX512BW-NEXT: vpinsrb $2, %ecx, %xmm0, %xmm0
+; AVX512BW-NEXT: vpinsrb $3, %eax, %xmm0, %xmm0
+; AVX512BW-NEXT: vpinsrb $4, %r11d, %xmm0, %xmm0
+; AVX512BW-NEXT: vpinsrb $5, %r10d, %xmm0, %xmm0
+; AVX512BW-NEXT: vpinsrb $6, %r9d, %xmm0, %xmm0
+; AVX512BW-NEXT: vpinsrb $7, %r8d, %xmm0, %xmm0
+; AVX512BW-NEXT: vmovq %xmm0, (%rsi)
+; AVX512BW-NEXT: retq
+;
+; AVX512BWVL-LABEL: shuffle_v64i8_to_v8i8:
+; AVX512BWVL: # BB#0:
+; AVX512BWVL-NEXT: vmovdqu8 (%rdi), %zmm0
+; AVX512BWVL-NEXT: vpextrb $0, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $0, %eax, %xmm0, %xmm1
+; AVX512BWVL-NEXT: vpextrb $8, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $2, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $1, %zmm0, %xmm2
+; AVX512BWVL-NEXT: vpextrb $0, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $4, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $8, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $6, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $2, %zmm0, %xmm2
+; AVX512BWVL-NEXT: vpextrb $0, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $8, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $8, %xmm2, %eax
+; AVX512BWVL-NEXT: vpinsrb $10, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vextracti32x4 $3, %zmm0, %xmm0
+; AVX512BWVL-NEXT: vpextrb $0, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $12, %eax, %xmm1, %xmm1
+; AVX512BWVL-NEXT: vpextrb $8, %xmm0, %eax
+; AVX512BWVL-NEXT: vpinsrb $14, %eax, %xmm1, %xmm0
+; AVX512BWVL-NEXT: vpmovwb %xmm0, (%rsi)
+; AVX512BWVL-NEXT: retq
+ %vec = load <64 x i8>, <64 x i8>* %L
+ %strided.vec = shufflevector <64 x i8> %vec, <64 x i8> undef, <8 x i32> <i32 0, i32 8, i32 16, i32 24, i32 32, i32 40, i32 48, i32 56>
+ store <8 x i8> %strided.vec, <8 x i8>* %S
+ ret void
+}
+
+define void @trunc_v8i64_to_v8i8(<64 x i8>* %L, <8 x i8>* %S) nounwind {
+; AVX512-LABEL: trunc_v8i64_to_v8i8:
+; AVX512: # BB#0:
+; AVX512-NEXT: vmovdqa64 (%rdi), %zmm0
+; AVX512-NEXT: vpmovqb %zmm0, (%rsi)
+; AVX512-NEXT: retq
+ %vec = load <64 x i8>, <64 x i8>* %L
+ %bc = bitcast <64 x i8> %vec to <8 x i64>
+ %strided.vec = trunc <8 x i64> %bc to <8 x i8>
+ store <8 x i8> %strided.vec, <8 x i8>* %S
+ ret void
+}
OpenPOWER on IntegriCloud