summaryrefslogtreecommitdiff
path: root/test/CodeGen/X86/vector-shuffle-256-v8.ll
diff options
context:
space:
mode:
Diffstat (limited to 'test/CodeGen/X86/vector-shuffle-256-v8.ll')
-rw-r--r--test/CodeGen/X86/vector-shuffle-256-v8.ll264
1 files changed, 202 insertions, 62 deletions
diff --git a/test/CodeGen/X86/vector-shuffle-256-v8.ll b/test/CodeGen/X86/vector-shuffle-256-v8.ll
index 44d0217f52954..1b6df7dd2d256 100644
--- a/test/CodeGen/X86/vector-shuffle-256-v8.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v8.ll
@@ -1,7 +1,9 @@
; 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=ALL --check-prefix=AVX1OR2 --check-prefix=AVX1
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX1OR2 --check-prefix=AVX2OR512VL --check-prefix=AVX2
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX2OR512VL --check-prefix=AVX512VL
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX1OR2 --check-prefix=AVX2OR512VL --check-prefix=AVX2 --check-prefix=AVX2-SLOW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2,+fast-variable-shuffle | FileCheck %s --check-prefix=ALL --check-prefix=AVX1OR2 --check-prefix=AVX2OR512VL --check-prefix=AVX2 --check-prefix=AVX2-FAST
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX2OR512VL --check-prefix=AVX512VL --check-prefix=AVX512VL-SLOW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+avx512dq,+fast-variable-shuffle | FileCheck %s --check-prefix=ALL --check-prefix=AVX2OR512VL --check-prefix=AVX512VL --check-prefix=AVX512VL-FAST
define <8 x float> @shuffle_v8f32_00000000(<8 x float> %a, <8 x float> %b) {
; AVX1-LABEL: shuffle_v8f32_00000000:
@@ -342,12 +344,26 @@ define <8 x float> @shuffle_v8f32_09ab1def(<8 x float> %a, <8 x float> %b) {
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
; AVX1-NEXT: retq
;
-; AVX2OR512VL-LABEL: shuffle_v8f32_09ab1def:
-; AVX2OR512VL: # %bb.0:
-; AVX2OR512VL-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,1,1,3]
-; AVX2OR512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
-; AVX2OR512VL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
-; AVX2OR512VL-NEXT: retq
+; AVX2-LABEL: shuffle_v8f32_09ab1def:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,1,1,3]
+; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
+; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-NEXT: retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v8f32_09ab1def:
+; AVX512VL-SLOW: # %bb.0:
+; AVX512VL-SLOW-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,1,1,3]
+; AVX512VL-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
+; AVX512VL-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX512VL-SLOW-NEXT: retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8f32_09ab1def:
+; AVX512VL-FAST: # %bb.0:
+; AVX512VL-FAST-NEXT: vpermilps {{.*#+}} xmm2 = xmm0[0,1,1,3]
+; AVX512VL-FAST-NEXT: vmovaps {{.*#+}} ymm0 = [8,1,2,3,10,5,6,7]
+; AVX512VL-FAST-NEXT: vpermi2ps %ymm2, %ymm1, %ymm0
+; AVX512VL-FAST-NEXT: retq
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 1, i32 13, i32 14, i32 15>
ret <8 x float> %shuffle
}
@@ -651,14 +667,23 @@ define <8 x float> @shuffle_v8f32_c348cda0(<8 x float> %a, <8 x float> %b) {
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2],ymm1[3,4,5,6],ymm0[7]
; AVX1-NEXT: retq
;
-; AVX2-LABEL: shuffle_v8f32_c348cda0:
-; AVX2: # %bb.0:
-; AVX2-NEXT: vmovaps {{.*#+}} ymm2 = <4,u,u,0,4,5,2,u>
-; AVX2-NEXT: vpermps %ymm1, %ymm2, %ymm1
-; AVX2-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,3,2,0,4,7,6,4]
-; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,1]
-; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2],ymm1[3,4,5,6],ymm0[7]
-; AVX2-NEXT: retq
+; AVX2-SLOW-LABEL: shuffle_v8f32_c348cda0:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: vmovaps {{.*#+}} ymm2 = <4,u,u,0,4,5,2,u>
+; AVX2-SLOW-NEXT: vpermps %ymm1, %ymm2, %ymm1
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,3,2,0,4,7,6,4]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,1]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2],ymm1[3,4,5,6],ymm0[7]
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-LABEL: shuffle_v8f32_c348cda0:
+; AVX2-FAST: # %bb.0:
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm2 = [0,3,4,7,4,7,2,0]
+; AVX2-FAST-NEXT: vpermps %ymm0, %ymm2, %ymm0
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm2 = <4,u,u,0,4,5,2,u>
+; AVX2-FAST-NEXT: vpermps %ymm1, %ymm2, %ymm1
+; AVX2-FAST-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2],ymm1[3,4,5,6],ymm0[7]
+; AVX2-FAST-NEXT: retq
;
; AVX512VL-LABEL: shuffle_v8f32_c348cda0:
; AVX512VL: # %bb.0:
@@ -681,14 +706,23 @@ define <8 x float> @shuffle_v8f32_f511235a(<8 x float> %a, <8 x float> %b) {
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3,4,5,6],ymm1[7]
; AVX1-NEXT: retq
;
-; AVX2-LABEL: shuffle_v8f32_f511235a:
-; AVX2: # %bb.0:
-; AVX2-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[3,2,2,3,7,6,6,7]
-; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[2,1,2,0]
-; AVX2-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[1,1,2,3,5,5,6,7]
-; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,0,1,2]
-; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3,4,5,6],ymm1[7]
-; AVX2-NEXT: retq
+; AVX2-SLOW-LABEL: shuffle_v8f32_f511235a:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[3,2,2,3,7,6,6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[2,1,2,0]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[1,1,2,3,5,5,6,7]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,0,1,2]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3,4,5,6],ymm1[7]
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-LABEL: shuffle_v8f32_f511235a:
+; AVX2-FAST: # %bb.0:
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm2 = [7,6,2,3,7,6,3,2]
+; AVX2-FAST-NEXT: vpermps %ymm1, %ymm2, %ymm1
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm2 = [5,5,1,1,2,3,5,5]
+; AVX2-FAST-NEXT: vpermps %ymm0, %ymm2, %ymm0
+; AVX2-FAST-NEXT: vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3,4,5,6],ymm1[7]
+; AVX2-FAST-NEXT: retq
;
; AVX512VL-LABEL: shuffle_v8f32_f511235a:
; AVX512VL: # %bb.0:
@@ -722,11 +756,29 @@ define <8 x float> @shuffle_v8f32_76547654(<8 x float> %a, <8 x float> %b) {
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: retq
;
-; AVX2OR512VL-LABEL: shuffle_v8f32_76547654:
-; AVX2OR512VL: # %bb.0:
-; AVX2OR512VL-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2OR512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; AVX2OR512VL-NEXT: retq
+; AVX2-SLOW-LABEL: shuffle_v8f32_76547654:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-LABEL: shuffle_v8f32_76547654:
+; AVX2-FAST: # %bb.0:
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm1 = [7,6,5,4,7,6,5,4]
+; AVX2-FAST-NEXT: vpermps %ymm0, %ymm1, %ymm0
+; AVX2-FAST-NEXT: retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v8f32_76547654:
+; AVX512VL-SLOW: # %bb.0:
+; AVX512VL-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX512VL-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX512VL-SLOW-NEXT: retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8f32_76547654:
+; AVX512VL-FAST: # %bb.0:
+; AVX512VL-FAST-NEXT: vmovaps {{.*#+}} ymm1 = [7,6,5,4,7,6,5,4]
+; AVX512VL-FAST-NEXT: vpermps %ymm0, %ymm1, %ymm0
+; AVX512VL-FAST-NEXT: retq
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 7, i32 6, i32 5, i32 4>
ret <8 x float> %shuffle
}
@@ -738,11 +790,29 @@ define <8 x float> @shuffle_v8f32_76543210(<8 x float> %a, <8 x float> %b) {
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
; AVX1-NEXT: retq
;
-; AVX2OR512VL-LABEL: shuffle_v8f32_76543210:
-; AVX2OR512VL: # %bb.0:
-; AVX2OR512VL-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2OR512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; AVX2OR512VL-NEXT: retq
+; AVX2-SLOW-LABEL: shuffle_v8f32_76543210:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-LABEL: shuffle_v8f32_76543210:
+; AVX2-FAST: # %bb.0:
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm1 = [7,6,5,4,3,2,1,0]
+; AVX2-FAST-NEXT: vpermps %ymm0, %ymm1, %ymm0
+; AVX2-FAST-NEXT: retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v8f32_76543210:
+; AVX512VL-SLOW: # %bb.0:
+; AVX512VL-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX512VL-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX512VL-SLOW-NEXT: retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8f32_76543210:
+; AVX512VL-FAST: # %bb.0:
+; AVX512VL-FAST-NEXT: vmovaps {{.*#+}} ymm1 = [7,6,5,4,3,2,1,0]
+; AVX512VL-FAST-NEXT: vpermps %ymm0, %ymm1, %ymm0
+; AVX512VL-FAST-NEXT: retq
%shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
ret <8 x float> %shuffle
}
@@ -798,11 +868,23 @@ define <8 x float> @PR21138(<8 x float> %truc, <8 x float> %tchose) {
; AVX1-NEXT: vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
; AVX1-NEXT: retq
;
-; AVX2OR512VL-LABEL: PR21138:
-; AVX2OR512VL: # %bb.0:
-; AVX2OR512VL-NEXT: vshufps {{.*#+}} ymm0 = ymm0[1,3],ymm1[1,3],ymm0[5,7],ymm1[5,7]
-; AVX2OR512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,1,3]
-; AVX2OR512VL-NEXT: retq
+; AVX2-LABEL: PR21138:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vshufps {{.*#+}} ymm0 = ymm0[1,3],ymm1[1,3],ymm0[5,7],ymm1[5,7]
+; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX2-NEXT: retq
+;
+; AVX512VL-SLOW-LABEL: PR21138:
+; AVX512VL-SLOW: # %bb.0:
+; AVX512VL-SLOW-NEXT: vshufps {{.*#+}} ymm0 = ymm0[1,3],ymm1[1,3],ymm0[5,7],ymm1[5,7]
+; AVX512VL-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,1,3]
+; AVX512VL-SLOW-NEXT: retq
+;
+; AVX512VL-FAST-LABEL: PR21138:
+; AVX512VL-FAST: # %bb.0:
+; AVX512VL-FAST-NEXT: vmovaps {{.*#+}} ymm2 = [1,3,5,7,9,11,13,15]
+; AVX512VL-FAST-NEXT: vpermt2ps %ymm1, %ymm2, %ymm0
+; AVX512VL-FAST-NEXT: retq
%shuffle = shufflevector <8 x float> %truc, <8 x float> %tchose, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
ret <8 x float> %shuffle
}
@@ -1264,12 +1346,26 @@ define <8 x i32> @shuffle_v8i32_09ab1def(<8 x i32> %a, <8 x i32> %b) {
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
; AVX1-NEXT: retq
;
-; AVX2OR512VL-LABEL: shuffle_v8i32_09ab1def:
-; AVX2OR512VL: # %bb.0:
-; AVX2OR512VL-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; AVX2OR512VL-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
-; AVX2OR512VL-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
-; AVX2OR512VL-NEXT: retq
+; AVX2-LABEL: shuffle_v8i32_09ab1def:
+; AVX2: # %bb.0:
+; AVX2-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
+; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
+; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-NEXT: retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v8i32_09ab1def:
+; AVX512VL-SLOW: # %bb.0:
+; AVX512VL-SLOW-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
+; AVX512VL-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
+; AVX512VL-SLOW-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX512VL-SLOW-NEXT: retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8i32_09ab1def:
+; AVX512VL-FAST: # %bb.0:
+; AVX512VL-FAST-NEXT: vpmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero
+; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} ymm0 = [8,1,2,3,10,5,6,7]
+; AVX512VL-FAST-NEXT: vpermi2d %ymm2, %ymm1, %ymm0
+; AVX512VL-FAST-NEXT: retq
%shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 1, i32 13, i32 14, i32 15>
ret <8 x i32> %shuffle
}
@@ -1696,13 +1792,21 @@ define <8 x i32> @shuffle_v8i32_6caa87e5(<8 x i32> %a, <8 x i32> %b) {
; AVX1-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4],ymm0[5],ymm1[6],ymm0[7]
; AVX1-NEXT: retq
;
-; AVX2-LABEL: shuffle_v8i32_6caa87e5:
-; AVX2: # %bb.0:
-; AVX2-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,1,3,2]
-; AVX2-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,0,2,2,4,4,6,6]
-; AVX2-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[2,1,0,3]
-; AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4],ymm0[5],ymm1[6],ymm0[7]
-; AVX2-NEXT: retq
+; AVX2-SLOW-LABEL: shuffle_v8i32_6caa87e5:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,1,3,2]
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,0,2,2,4,4,6,6]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[2,1,0,3]
+; AVX2-SLOW-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4],ymm0[5],ymm1[6],ymm0[7]
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-LABEL: shuffle_v8i32_6caa87e5:
+; AVX2-FAST: # %bb.0:
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm2 = [4,4,2,2,0,0,6,6]
+; AVX2-FAST-NEXT: vpermps %ymm1, %ymm2, %ymm1
+; AVX2-FAST-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,1,3,2]
+; AVX2-FAST-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4],ymm0[5],ymm1[6],ymm0[7]
+; AVX2-FAST-NEXT: retq
;
; AVX512VL-LABEL: shuffle_v8i32_6caa87e5:
; AVX512VL: # %bb.0:
@@ -1737,11 +1841,29 @@ define <8 x i32> @shuffle_v8i32_76547654(<8 x i32> %a, <8 x i32> %b) {
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
; AVX1-NEXT: retq
;
-; AVX2OR512VL-LABEL: shuffle_v8i32_76547654:
-; AVX2OR512VL: # %bb.0:
-; AVX2OR512VL-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2OR512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; AVX2OR512VL-NEXT: retq
+; AVX2-SLOW-LABEL: shuffle_v8i32_76547654:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-LABEL: shuffle_v8i32_76547654:
+; AVX2-FAST: # %bb.0:
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm1 = [7,6,5,4,7,6,5,4]
+; AVX2-FAST-NEXT: vpermps %ymm0, %ymm1, %ymm0
+; AVX2-FAST-NEXT: retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v8i32_76547654:
+; AVX512VL-SLOW: # %bb.0:
+; AVX512VL-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX512VL-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX512VL-SLOW-NEXT: retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8i32_76547654:
+; AVX512VL-FAST: # %bb.0:
+; AVX512VL-FAST-NEXT: vmovaps {{.*#+}} ymm1 = [7,6,5,4,7,6,5,4]
+; AVX512VL-FAST-NEXT: vpermps %ymm0, %ymm1, %ymm0
+; AVX512VL-FAST-NEXT: retq
%shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 7, i32 6, i32 5, i32 4>
ret <8 x i32> %shuffle
}
@@ -1753,11 +1875,29 @@ define <8 x i32> @shuffle_v8i32_76543210(<8 x i32> %a, <8 x i32> %b) {
; AVX1-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
; AVX1-NEXT: retq
;
-; AVX2OR512VL-LABEL: shuffle_v8i32_76543210:
-; AVX2OR512VL: # %bb.0:
-; AVX2OR512VL-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2OR512VL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; AVX2OR512VL-NEXT: retq
+; AVX2-SLOW-LABEL: shuffle_v8i32_76543210:
+; AVX2-SLOW: # %bb.0:
+; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2-SLOW-NEXT: retq
+;
+; AVX2-FAST-LABEL: shuffle_v8i32_76543210:
+; AVX2-FAST: # %bb.0:
+; AVX2-FAST-NEXT: vmovaps {{.*#+}} ymm1 = [7,6,5,4,3,2,1,0]
+; AVX2-FAST-NEXT: vpermps %ymm0, %ymm1, %ymm0
+; AVX2-FAST-NEXT: retq
+;
+; AVX512VL-SLOW-LABEL: shuffle_v8i32_76543210:
+; AVX512VL-SLOW: # %bb.0:
+; AVX512VL-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX512VL-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX512VL-SLOW-NEXT: retq
+;
+; AVX512VL-FAST-LABEL: shuffle_v8i32_76543210:
+; AVX512VL-FAST: # %bb.0:
+; AVX512VL-FAST-NEXT: vmovaps {{.*#+}} ymm1 = [7,6,5,4,3,2,1,0]
+; AVX512VL-FAST-NEXT: vpermps %ymm0, %ymm1, %ymm0
+; AVX512VL-FAST-NEXT: retq
%shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
ret <8 x i32> %shuffle
}
generated by cgit v1.2.3 (git 2.25.1) at 2025-08-02 14:41:36 +0000