diff options
| author | Evan Cheng <evan.cheng@apple.com> | 2008-04-05 00:30:36 +0000 |
|---|---|---|
| committer | Evan Cheng <evan.cheng@apple.com> | 2008-04-05 00:30:36 +0000 |
| commit | 0c0f83ff5d214a7f42e86ae62814526ba40a28cf (patch) | |
| tree | ffb95313874d040042271f886170c24b80ebd693 /test/CodeGen/X86/vec_splat.ll | |
| parent | a4091d34f3c05717eb5bb66a1257a0c1005e6dfa (diff) | |
| download | external_llvm-0c0f83ff5d214a7f42e86ae62814526ba40a28cf.zip external_llvm-0c0f83ff5d214a7f42e86ae62814526ba40a28cf.tar.gz external_llvm-0c0f83ff5d214a7f42e86ae62814526ba40a28cf.tar.bz2 | |
Favors pshufd over shufps when shuffling elements from one vector. pshufd is faster than shufps.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@49244 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/vec_splat.ll')
| -rw-r--r-- | test/CodeGen/X86/vec_splat.ll | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/test/CodeGen/X86/vec_splat.ll b/test/CodeGen/X86/vec_splat.ll index c6100ec..64222e4 100644 --- a/test/CodeGen/X86/vec_splat.ll +++ b/test/CodeGen/X86/vec_splat.ll @@ -1,7 +1,7 @@ -; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | grep shufps +; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | grep pshufd ; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse3 | grep movddup -define void @test_v4sf(<4 x float>* %P, <4 x float>* %Q, float %X) { +define void @test_v4sf(<4 x float>* %P, <4 x float>* %Q, float %X) nounwind { %tmp = insertelement <4 x float> zeroinitializer, float %X, i32 0 ; <<4 x float>> [#uses=1] %tmp2 = insertelement <4 x float> %tmp, float %X, i32 1 ; <<4 x float>> [#uses=1] %tmp4 = insertelement <4 x float> %tmp2, float %X, i32 2 ; <<4 x float>> [#uses=1] @@ -12,7 +12,7 @@ define void @test_v4sf(<4 x float>* %P, <4 x float>* %Q, float %X) { ret void } -define void @test_v2sd(<2 x double>* %P, <2 x double>* %Q, double %X) { +define void @test_v2sd(<2 x double>* %P, <2 x double>* %Q, double %X) nounwind { %tmp = insertelement <2 x double> zeroinitializer, double %X, i32 0 ; <<2 x double>> [#uses=1] %tmp2 = insertelement <2 x double> %tmp, double %X, i32 1 ; <<2 x double>> [#uses=1] %tmp4 = load <2 x double>* %Q ; <<2 x double>> [#uses=1] |