[llvm] r262134 - [X86][AVX] vpermilvar.pd mask element indices only use bit1
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Sat Feb 27 04:51:46 PST 2016
Author: rksimon
Date: Sat Feb 27 06:51:46 2016
New Revision: 262134
URL: http://llvm.org/viewvc/llvm-project?rev=262134&view=rev
Log:
[X86][AVX] vpermilvar.pd mask element indices only use bit1
Modified:
llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx.ll
Modified: llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx.ll?rev=262134&r1=262133&r2=262134&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx.ll (original)
+++ llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx.ll Sat Feb 27 06:51:46 2016
@@ -46,24 +46,24 @@ define <8 x float> @combine_vpermilvar_8
define <2 x double> @combine_vpermilvar_2f64(<2 x double> %a0) {
; ALL-LABEL: combine_vpermilvar_2f64:
; ALL: # BB#0:
-; ALL-NEXT: movl $1, %eax
+; ALL-NEXT: movl $2, %eax
; ALL-NEXT: vmovq %rax, %xmm1
; ALL-NEXT: vpermilpd %xmm1, %xmm0, %xmm0
; ALL-NEXT: vpermilpd %xmm1, %xmm0, %xmm0
; ALL-NEXT: retq
- %1 = tail call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> %a0, <2 x i64> <i64 1, i64 0>)
- %2 = tail call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> %1, <2 x i64> <i64 1, i64 0>)
+ %1 = tail call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> %a0, <2 x i64> <i64 2, i64 0>)
+ %2 = tail call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> %1, <2 x i64> <i64 2, i64 0>)
ret <2 x double> %2
}
define <4 x double> @combine_vpermilvar_4f64(<4 x double> %a0) {
; ALL-LABEL: combine_vpermilvar_4f64:
; ALL: # BB#0:
-; ALL-NEXT: vmovapd {{.*#+}} ymm1 = [1,0,1,0]
+; ALL-NEXT: vmovapd {{.*#+}} ymm1 = [2,0,2,0]
; ALL-NEXT: vpermilpd %ymm1, %ymm0, %ymm0
; ALL-NEXT: vpermilpd %ymm1, %ymm0, %ymm0
; ALL-NEXT: retq
- %1 = tail call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> %a0, <4 x i64> <i64 1, i64 0, i64 1, i64 0>)
- %2 = tail call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> %1, <4 x i64> <i64 1, i64 0, i64 1, i64 0>)
+ %1 = tail call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> %a0, <4 x i64> <i64 2, i64 0, i64 2, i64 0>)
+ %2 = tail call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> %1, <4 x i64> <i64 2, i64 0, i64 2, i64 0>)
ret <4 x double> %2
}
More information about the llvm-commits
mailing list