diff options
| author | Craig Topper <craig.topper@intel.com> | 2017-09-18 04:40:58 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2017-09-18 04:40:58 +0000 |
| commit | a6054328e8b31217bec6e6d5dcbb28ca304d48d1 (patch) | |
| tree | 5d1cb5f095fe831f3345ebbd1ce4fb5d72bd64c0 /llvm/test/CodeGen/X86/build-vector-256.ll | |
| parent | 87f7381edf47c104f3067827cd93df5d0009e103 (diff) | |
| download | bcm5719-llvm-a6054328e8b31217bec6e6d5dcbb28ca304d48d1.tar.gz bcm5719-llvm-a6054328e8b31217bec6e6d5dcbb28ca304d48d1.zip | |
[X86] Teach the execution domain fixing tables to use movlhps inplace of unpcklpd for the packed single domain.
MOVLHPS has a smaller encoding than UNPCKLPD in the legacy encodings. With VEX and EVEX encodings it doesn't matter.
llvm-svn: 313509
Diffstat (limited to 'llvm/test/CodeGen/X86/build-vector-256.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/build-vector-256.ll | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/llvm/test/CodeGen/X86/build-vector-256.ll b/llvm/test/CodeGen/X86/build-vector-256.ll index 942b7779abe..4b077cc2469 100644 --- a/llvm/test/CodeGen/X86/build-vector-256.ll +++ b/llvm/test/CodeGen/X86/build-vector-256.ll @@ -12,8 +12,8 @@ define <4 x double> @test_buildvector_v4f64(double %a0, double %a1, double %a2, ; ; AVX-64-LABEL: test_buildvector_v4f64: ; AVX-64: # BB#0: -; AVX-64-NEXT: vunpcklpd {{.*#+}} xmm2 = xmm2[0],xmm3[0] -; AVX-64-NEXT: vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; AVX-64-NEXT: vmovlhps {{.*#+}} xmm2 = xmm2[0],xmm3[0] +; AVX-64-NEXT: vmovlhps {{.*#+}} xmm0 = xmm0[0],xmm1[0] ; AVX-64-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0 ; AVX-64-NEXT: retq %ins0 = insertelement <4 x double> undef, double %a0, i32 0 |
