shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity

[X86] Clearout unused FALLBACK check prefixes from vector extension/concatenation tests. NFC. (#183527)

Browse Source 
These were used before update_llc_test_checks reported when we had lost check coverage - we've cleaned up a lot of check prefixes since then
This commit is contained in:
Simon Pilgrim 2026-02-26 14:36:47 +00:00 committed by GitHub
parent 9ec7b025cf
commit 7516bc1a57
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 2887 additions and 3063 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1434
llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll
View File

File diff suppressed because it is too large Load Diff

552
llvm/test/CodeGen/X86/vector-shuffle-concatenation.ll
View File

@ -1,14 +1,14 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefixes=SSE,SSE2,FALLBACK0
; RUN: llc < %s -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s --check-prefixes=SSE,SSE42,FALLBACK1
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx | FileCheck %s --check-prefixes=AVX,AVX1-ONLY,FALLBACK2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX2,AVX2-SLOW,FALLBACK3
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX2,AVX2-FAST-PERLANE,FALLBACK4
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX2,AVX2-FAST,FALLBACK5
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX512F,AVX512F-SLOW,FALLBACK6
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX512F,AVX512F-FAST,FALLBACK7
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX512BW,AVX512BW-SLOW,FALLBACK8
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX512BW,AVX512BW-FAST,FALLBACK9
; RUN: llc < %s -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefixes=SSE,SSE2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s --check-prefixes=SSE,SSE42
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx | FileCheck %s --check-prefixes=AVX1
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=AVX2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=AVX512
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX512
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX512
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512bw,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle | FileCheck %s --check-prefixes=AVX512
define void @concat_a_to_shuf_of_a(ptr %a.ptr, ptr %dst) {
; SSE-LABEL: concat_a_to_shuf_of_a:
@ -19,14 +19,14 @@ define void @concat_a_to_shuf_of_a(ptr %a.ptr, ptr %dst) {
; SSE-NEXT: movdqa %xmm1, (%rsi)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_a_to_shuf_of_a:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rdi), %xmm0
; AVX-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX-NEXT: vmovaps %ymm0, (%rsi)
; AVX-NEXT: vzeroupper
; AVX-NEXT: retq
; AVX1-LABEL: concat_a_to_shuf_of_a:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rdi), %xmm0
; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT: vmovaps %ymm0, (%rsi)
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_a_to_shuf_of_a:
; AVX2: # %bb.0:
@ -36,21 +36,13 @@ define void @concat_a_to_shuf_of_a(ptr %a.ptr, ptr %dst) {
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_a_to_shuf_of_a:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rdi), %xmm0
; AVX512F-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,0,1]
; AVX512F-NEXT: vmovaps %ymm0, (%rsi)
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_a_to_shuf_of_a:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rdi), %xmm0
; AVX512BW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,0,1]
; AVX512BW-NEXT: vmovaps %ymm0, (%rsi)
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_a_to_shuf_of_a:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rdi), %xmm0
; AVX512-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,0,1]
; AVX512-NEXT: vmovaps %ymm0, (%rsi)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
%concat = shufflevector <2 x i64> %shuffle, <2 x i64> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
@ -66,14 +58,14 @@ define void @concat_shuf_of_a_to_a(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; SSE-NEXT: movdqa %xmm1, 16(%rdx)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_shuf_of_a_to_a:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rdi), %xmm0
; AVX-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX-NEXT: vmovaps %ymm0, (%rdx)
; AVX-NEXT: vzeroupper
; AVX-NEXT: retq
; AVX1-LABEL: concat_shuf_of_a_to_a:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rdi), %xmm0
; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: vmovaps %ymm0, (%rdx)
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_shuf_of_a_to_a:
; AVX2: # %bb.0:
@ -83,21 +75,13 @@ define void @concat_shuf_of_a_to_a(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_shuf_of_a_to_a:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rdi), %xmm0
; AVX512F-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,0]
; AVX512F-NEXT: vmovaps %ymm0, (%rdx)
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_shuf_of_a_to_a:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rdi), %xmm0
; AVX512BW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,0]
; AVX512BW-NEXT: vmovaps %ymm0, (%rdx)
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_shuf_of_a_to_a:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rdi), %xmm0
; AVX512-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,0]
; AVX512-NEXT: vmovaps %ymm0, (%rdx)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%b = load <2 x i64>, ptr %b.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
@ -116,14 +100,14 @@ define void @concat_a_to_shuf_of_a_extrause_of_shuf(ptr %a.ptr, ptr %dst, ptr %s
; SSE-NEXT: movdqa %xmm1, (%rsi)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_a_to_shuf_of_a_extrause_of_shuf:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rdi), %xmm0
; AVX-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX-NEXT: vmovaps %xmm1, (%rdx)
; AVX-NEXT: vmovaps %xmm0, 16(%rsi)
; AVX-NEXT: vmovaps %xmm1, (%rsi)
; AVX-NEXT: retq
; AVX1-LABEL: concat_a_to_shuf_of_a_extrause_of_shuf:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rdi), %xmm0
; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX1-NEXT: vmovaps %xmm1, (%rdx)
; AVX1-NEXT: vmovaps %xmm0, 16(%rsi)
; AVX1-NEXT: vmovaps %xmm1, (%rsi)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_a_to_shuf_of_a_extrause_of_shuf:
; AVX2: # %bb.0:
@ -134,23 +118,14 @@ define void @concat_a_to_shuf_of_a_extrause_of_shuf(ptr %a.ptr, ptr %dst, ptr %s
; AVX2-NEXT: vmovaps %xmm1, (%rsi)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_a_to_shuf_of_a_extrause_of_shuf:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rdi), %xmm0
; AVX512F-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512F-NEXT: vmovaps %xmm1, (%rdx)
; AVX512F-NEXT: vmovaps %xmm0, 16(%rsi)
; AVX512F-NEXT: vmovaps %xmm1, (%rsi)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_a_to_shuf_of_a_extrause_of_shuf:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rdi), %xmm0
; AVX512BW-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512BW-NEXT: vmovaps %xmm1, (%rdx)
; AVX512BW-NEXT: vmovaps %xmm0, 16(%rsi)
; AVX512BW-NEXT: vmovaps %xmm1, (%rsi)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_a_to_shuf_of_a_extrause_of_shuf:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rdi), %xmm0
; AVX512-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512-NEXT: vmovaps %xmm1, (%rdx)
; AVX512-NEXT: vmovaps %xmm0, 16(%rsi)
; AVX512-NEXT: vmovaps %xmm1, (%rsi)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
store <2 x i64> %shuffle, ptr %shuf.escape.ptr, align 64
@ -178,13 +153,13 @@ define void @concat_a_to_shuf_of_ab(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; SSE42-NEXT: movaps %xmm1, (%rdx)
; SSE42-NEXT: retq
;
; AVX-LABEL: concat_a_to_shuf_of_ab:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rdi), %xmm0
; AVX-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX-NEXT: vmovaps %xmm1, (%rdx)
; AVX-NEXT: retq
; AVX1-LABEL: concat_a_to_shuf_of_ab:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rdi), %xmm0
; AVX1-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX1-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX1-NEXT: vmovaps %xmm1, (%rdx)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_a_to_shuf_of_ab:
; AVX2: # %bb.0:
@ -194,21 +169,13 @@ define void @concat_a_to_shuf_of_ab(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; AVX2-NEXT: vmovaps %xmm1, (%rdx)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_a_to_shuf_of_ab:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rdi), %xmm0
; AVX512F-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX512F-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512F-NEXT: vmovaps %xmm1, (%rdx)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_a_to_shuf_of_ab:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rdi), %xmm0
; AVX512BW-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX512BW-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512BW-NEXT: vmovaps %xmm1, (%rdx)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_a_to_shuf_of_ab:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rdi), %xmm0
; AVX512-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX512-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512-NEXT: vmovaps %xmm1, (%rdx)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%b = load <2 x i64>, ptr %b.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 3>
@ -235,13 +202,13 @@ define void @concat_b_to_shuf_of_ab(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; SSE42-NEXT: movaps %xmm1, (%rdx)
; SSE42-NEXT: retq
;
; AVX-LABEL: concat_b_to_shuf_of_ab:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rsi), %xmm0
; AVX-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX-NEXT: vmovaps %xmm1, (%rdx)
; AVX-NEXT: retq
; AVX1-LABEL: concat_b_to_shuf_of_ab:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rsi), %xmm0
; AVX1-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX1-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX1-NEXT: vmovaps %xmm1, (%rdx)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_b_to_shuf_of_ab:
; AVX2: # %bb.0:
@ -251,21 +218,13 @@ define void @concat_b_to_shuf_of_ab(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; AVX2-NEXT: vmovaps %xmm1, (%rdx)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_b_to_shuf_of_ab:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rsi), %xmm0
; AVX512F-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX512F-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512F-NEXT: vmovaps %xmm1, (%rdx)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_b_to_shuf_of_ab:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rsi), %xmm0
; AVX512BW-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX512BW-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512BW-NEXT: vmovaps %xmm1, (%rdx)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_b_to_shuf_of_ab:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rsi), %xmm0
; AVX512-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX512-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512-NEXT: vmovaps %xmm1, (%rdx)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%b = load <2 x i64>, ptr %b.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 3>
@ -293,13 +252,13 @@ define void @concat_shuf_of_ab_to_a(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; SSE42-NEXT: movaps %xmm0, (%rdx)
; SSE42-NEXT: retq
;
; AVX-LABEL: concat_shuf_of_ab_to_a:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rdi), %xmm0
; AVX-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX-NEXT: vmovaps %xmm0, (%rdx)
; AVX-NEXT: retq
; AVX1-LABEL: concat_shuf_of_ab_to_a:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rdi), %xmm0
; AVX1-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX1-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX1-NEXT: vmovaps %xmm0, (%rdx)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_shuf_of_ab_to_a:
; AVX2: # %bb.0:
@ -309,21 +268,13 @@ define void @concat_shuf_of_ab_to_a(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; AVX2-NEXT: vmovaps %xmm0, (%rdx)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_shuf_of_ab_to_a:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rdi), %xmm0
; AVX512F-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX512F-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512F-NEXT: vmovaps %xmm0, (%rdx)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_shuf_of_ab_to_a:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rdi), %xmm0
; AVX512BW-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX512BW-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512BW-NEXT: vmovaps %xmm0, (%rdx)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_shuf_of_ab_to_a:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rdi), %xmm0
; AVX512-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],mem[2,3]
; AVX512-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512-NEXT: vmovaps %xmm0, (%rdx)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%b = load <2 x i64>, ptr %b.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 3>
@ -350,13 +301,13 @@ define void @concat_shuf_of_ab_to_b(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; SSE42-NEXT: movaps %xmm0, (%rdx)
; SSE42-NEXT: retq
;
; AVX-LABEL: concat_shuf_of_ab_to_b:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rsi), %xmm0
; AVX-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX-NEXT: vmovaps %xmm0, (%rdx)
; AVX-NEXT: retq
; AVX1-LABEL: concat_shuf_of_ab_to_b:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rsi), %xmm0
; AVX1-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX1-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX1-NEXT: vmovaps %xmm0, (%rdx)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_shuf_of_ab_to_b:
; AVX2: # %bb.0:
@ -366,21 +317,13 @@ define void @concat_shuf_of_ab_to_b(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; AVX2-NEXT: vmovaps %xmm0, (%rdx)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_shuf_of_ab_to_b:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rsi), %xmm0
; AVX512F-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX512F-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512F-NEXT: vmovaps %xmm0, (%rdx)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_shuf_of_ab_to_b:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rsi), %xmm0
; AVX512BW-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX512BW-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512BW-NEXT: vmovaps %xmm0, (%rdx)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_shuf_of_ab_to_b:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rsi), %xmm0
; AVX512-NEXT: vblendps {{.*#+}} xmm1 = mem[0,1],xmm0[2,3]
; AVX512-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512-NEXT: vmovaps %xmm0, (%rdx)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%b = load <2 x i64>, ptr %b.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 3>
@ -398,13 +341,13 @@ define void @concat_b_to_shuf_of_a(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; SSE-NEXT: movdqa %xmm1, (%rdx)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_b_to_shuf_of_a:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rsi), %xmm0
; AVX-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX-NEXT: vmovaps %xmm1, (%rdx)
; AVX-NEXT: retq
; AVX1-LABEL: concat_b_to_shuf_of_a:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rsi), %xmm0
; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX1-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX1-NEXT: vmovaps %xmm1, (%rdx)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_b_to_shuf_of_a:
; AVX2: # %bb.0:
@ -414,21 +357,13 @@ define void @concat_b_to_shuf_of_a(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; AVX2-NEXT: vmovaps %xmm1, (%rdx)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_b_to_shuf_of_a:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rsi), %xmm0
; AVX512F-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX512F-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512F-NEXT: vmovaps %xmm1, (%rdx)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_b_to_shuf_of_a:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rsi), %xmm0
; AVX512BW-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX512BW-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512BW-NEXT: vmovaps %xmm1, (%rdx)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_b_to_shuf_of_a:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rsi), %xmm0
; AVX512-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX512-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512-NEXT: vmovaps %xmm1, (%rdx)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%b = load <2 x i64>, ptr %b.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
@ -445,13 +380,13 @@ define void @concat_shuf_of_a_to_b(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; SSE-NEXT: movaps %xmm0, (%rdx)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_shuf_of_a_to_b:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps (%rsi), %xmm0
; AVX-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX-NEXT: vmovaps %xmm0, (%rdx)
; AVX-NEXT: retq
; AVX1-LABEL: concat_shuf_of_a_to_b:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovaps (%rsi), %xmm0
; AVX1-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX1-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX1-NEXT: vmovaps %xmm0, (%rdx)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_shuf_of_a_to_b:
; AVX2: # %bb.0:
@ -461,21 +396,13 @@ define void @concat_shuf_of_a_to_b(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; AVX2-NEXT: vmovaps %xmm0, (%rdx)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_shuf_of_a_to_b:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rsi), %xmm0
; AVX512F-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX512F-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512F-NEXT: vmovaps %xmm0, (%rdx)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_shuf_of_a_to_b:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rsi), %xmm0
; AVX512BW-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX512BW-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512BW-NEXT: vmovaps %xmm0, (%rdx)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_shuf_of_a_to_b:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rsi), %xmm0
; AVX512-NEXT: vpermilps {{.*#+}} xmm1 = mem[2,3,0,1]
; AVX512-NEXT: vmovaps %xmm1, 16(%rdx)
; AVX512-NEXT: vmovaps %xmm0, (%rdx)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%b = load <2 x i64>, ptr %b.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
@ -491,11 +418,11 @@ define void @concat_poison_to_shuf_of_a(ptr %a.ptr, ptr %dst) {
; SSE-NEXT: movdqa %xmm0, (%rsi)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_poison_to_shuf_of_a:
; AVX: # %bb.0:
; AVX-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX-NEXT: vmovaps %xmm0, (%rsi)
; AVX-NEXT: retq
; AVX1-LABEL: concat_poison_to_shuf_of_a:
; AVX1: # %bb.0:
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX1-NEXT: vmovaps %xmm0, (%rsi)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_poison_to_shuf_of_a:
; AVX2: # %bb.0:
@ -503,17 +430,11 @@ define void @concat_poison_to_shuf_of_a(ptr %a.ptr, ptr %dst) {
; AVX2-NEXT: vmovaps %xmm0, (%rsi)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_poison_to_shuf_of_a:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX512F-NEXT: vmovaps %xmm0, (%rsi)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_poison_to_shuf_of_a:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX512BW-NEXT: vmovaps %xmm0, (%rsi)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_poison_to_shuf_of_a:
; AVX512: # %bb.0:
; AVX512-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX512-NEXT: vmovaps %xmm0, (%rsi)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
%concat = shufflevector <2 x i64> %shuffle, <2 x i64> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
@ -527,11 +448,11 @@ define void @concat_shuf_of_a_to_poison(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; SSE-NEXT: movdqa %xmm0, 16(%rdx)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_shuf_of_a_to_poison:
; AVX: # %bb.0:
; AVX-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX-NEXT: retq
; AVX1-LABEL: concat_shuf_of_a_to_poison:
; AVX1: # %bb.0:
; AVX1-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX1-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_shuf_of_a_to_poison:
; AVX2: # %bb.0:
@ -539,17 +460,11 @@ define void @concat_shuf_of_a_to_poison(ptr %a.ptr, ptr %b.ptr, ptr %dst) {
; AVX2-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_shuf_of_a_to_poison:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX512F-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_shuf_of_a_to_poison:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX512BW-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_shuf_of_a_to_poison:
; AVX512: # %bb.0:
; AVX512-NEXT: vpermilps {{.*#+}} xmm0 = mem[2,3,0,1]
; AVX512-NEXT: vmovaps %xmm0, 16(%rdx)
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%b = load <2 x i64>, ptr %b.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
@ -566,13 +481,13 @@ define void @concat_shuf_of_a_to_itself(ptr %a.ptr, ptr %dst) {
; SSE-NEXT: movdqa %xmm0, (%rsi)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_shuf_of_a_to_itself:
; AVX: # %bb.0:
; AVX-NEXT: vpermilpd {{.*#+}} xmm0 = mem[1,0]
; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX-NEXT: vmovaps %ymm0, (%rsi)
; AVX-NEXT: vzeroupper
; AVX-NEXT: retq
; AVX1-LABEL: concat_shuf_of_a_to_itself:
; AVX1: # %bb.0:
; AVX1-NEXT: vpermilpd {{.*#+}} xmm0 = mem[1,0]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX1-NEXT: vmovaps %ymm0, (%rsi)
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_shuf_of_a_to_itself:
; AVX2: # %bb.0:
@ -582,21 +497,13 @@ define void @concat_shuf_of_a_to_itself(ptr %a.ptr, ptr %dst) {
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_shuf_of_a_to_itself:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vmovaps (%rdi), %xmm0
; AVX512F-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,1,0]
; AVX512F-NEXT: vmovaps %ymm0, (%rsi)
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_shuf_of_a_to_itself:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovaps (%rdi), %xmm0
; AVX512BW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,1,0]
; AVX512BW-NEXT: vmovaps %ymm0, (%rsi)
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_shuf_of_a_to_itself:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps (%rdi), %xmm0
; AVX512-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[1,0,1,0]
; AVX512-NEXT: vmovaps %ymm0, (%rsi)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
%concat = shufflevector <2 x i64> %shuffle, <2 x i64> %shuffle, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
@ -615,15 +522,15 @@ define void @concat_aaa_to_shuf_of_a(ptr %a.ptr, ptr %dst) {
; SSE-NEXT: movdqa %xmm1, (%rsi)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_aaa_to_shuf_of_a:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm1
; AVX-NEXT: vmovaps %ymm0, 32(%rsi)
; AVX-NEXT: vmovaps %ymm1, (%rsi)
; AVX-NEXT: vzeroupper
; AVX-NEXT: retq
; AVX1-LABEL: concat_aaa_to_shuf_of_a:
; AVX1: # %bb.0:
; AVX1-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm1
; AVX1-NEXT: vmovaps %ymm0, 32(%rsi)
; AVX1-NEXT: vmovaps %ymm1, (%rsi)
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_aaa_to_shuf_of_a:
; AVX2: # %bb.0:
@ -634,25 +541,15 @@ define void @concat_aaa_to_shuf_of_a(ptr %a.ptr, ptr %dst) {
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_aaa_to_shuf_of_a:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vbroadcasti128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX512F-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512F-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm1
; AVX512F-NEXT: vinserti64x4 $1, %ymm0, %zmm1, %zmm0
; AVX512F-NEXT: vmovdqa64 %zmm0, (%rsi)
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_aaa_to_shuf_of_a:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vbroadcasti128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX512BW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512BW-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm1
; AVX512BW-NEXT: vinserti64x4 $1, %ymm0, %zmm1, %zmm0
; AVX512BW-NEXT: vmovdqa64 %zmm0, (%rsi)
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_aaa_to_shuf_of_a:
; AVX512: # %bb.0:
; AVX512-NEXT: vbroadcasti128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm1
; AVX512-NEXT: vinserti64x4 $1, %ymm0, %zmm1, %zmm0
; AVX512-NEXT: vmovdqa64 %zmm0, (%rsi)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
%concat01 = shufflevector <2 x i64> %shuffle, <2 x i64> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
@ -672,15 +569,15 @@ define void @concat_shuf_of_a_to_aaa(ptr %a.ptr, ptr %dst) {
; SSE-NEXT: movdqa %xmm1, 48(%rsi)
; SSE-NEXT: retq
;
; AVX-LABEL: concat_shuf_of_a_to_aaa:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
; AVX-NEXT: vmovaps %ymm0, (%rsi)
; AVX-NEXT: vmovaps %ymm1, 32(%rsi)
; AVX-NEXT: vzeroupper
; AVX-NEXT: retq
; AVX1-LABEL: concat_shuf_of_a_to_aaa:
; AVX1: # %bb.0:
; AVX1-NEXT: vbroadcastf128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
; AVX1-NEXT: vmovaps %ymm0, (%rsi)
; AVX1-NEXT: vmovaps %ymm1, 32(%rsi)
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_shuf_of_a_to_aaa:
; AVX2: # %bb.0:
@ -691,25 +588,15 @@ define void @concat_shuf_of_a_to_aaa(ptr %a.ptr, ptr %dst) {
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512F-LABEL: concat_shuf_of_a_to_aaa:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vbroadcasti128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX512F-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512F-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX512F-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512F-NEXT: vmovdqa64 %zmm0, (%rsi)
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_shuf_of_a_to_aaa:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vbroadcasti128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX512BW-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512BW-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX512BW-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512BW-NEXT: vmovdqa64 %zmm0, (%rsi)
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
; AVX512-LABEL: concat_shuf_of_a_to_aaa:
; AVX512: # %bb.0:
; AVX512-NEXT: vbroadcasti128 {{.*#+}} ymm0 = mem[0,1,0,1]
; AVX512-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX512-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm1
; AVX512-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-NEXT: vmovdqa64 %zmm0, (%rsi)
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%a = load <2 x i64>, ptr %a.ptr, align 64
%shuffle = shufflevector <2 x i64> %a, <2 x i64> poison, <2 x i32> <i32 1, i32 0>
%concat01 = shufflevector <2 x i64> %a, <2 x i64> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
@ -718,22 +605,3 @@ define void @concat_shuf_of_a_to_aaa(ptr %a.ptr, ptr %dst) {
store <8 x i64> %concat, ptr %dst, align 64
ret void
}
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; AVX1-ONLY: {{.*}}
; AVX2-FAST: {{.*}}
; AVX2-FAST-PERLANE: {{.*}}
; AVX2-SLOW: {{.*}}
; AVX512BW-FAST: {{.*}}
; AVX512BW-SLOW: {{.*}}
; AVX512F-FAST: {{.*}}
; AVX512F-SLOW: {{.*}}
; FALLBACK0: {{.*}}
; FALLBACK1: {{.*}}
; FALLBACK2: {{.*}}
; FALLBACK3: {{.*}}
; FALLBACK4: {{.*}}
; FALLBACK5: {{.*}}
; FALLBACK6: {{.*}}
; FALLBACK7: {{.*}}
; FALLBACK8: {{.*}}
; FALLBACK9: {{.*}}

2200
llvm/test/CodeGen/X86/zero_extend_vector_inreg.ll
View File

File diff suppressed because it is too large Load Diff

1764
llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll
View File

File diff suppressed because it is too large Load Diff