diff options
| author | Benjamin Kramer <benny.kra@googlemail.com> | 2013-10-23 19:19:04 +0000 |
|---|---|---|
| committer | Benjamin Kramer <benny.kra@googlemail.com> | 2013-10-23 19:19:04 +0000 |
| commit | 7377cff9e7641c75678fd5c80472942fd7ef869a (patch) | |
| tree | 8ff95078dc08a96effbc21e3cb59dc2630005b96 /test/CodeGen/X86/movntdq-no-avx.ll | |
| parent | 531f025361555e7a695eb559ec02645c054ee146 (diff) | |
| download | external_llvm-7377cff9e7641c75678fd5c80472942fd7ef869a.zip external_llvm-7377cff9e7641c75678fd5c80472942fd7ef869a.tar.gz external_llvm-7377cff9e7641c75678fd5c80472942fd7ef869a.tar.bz2 | |
X86: Custom lower zext v16i8 to v16i16.
On sandy bridge (PR17654) we now get
vpxor %xmm1, %xmm1, %xmm1
vpunpckhbw %xmm1, %xmm0, %xmm2
vpunpcklbw %xmm1, %xmm0, %xmm0
vinsertf128 $1, %xmm2, %ymm0, %ymm0
On haswell it's a simple
vpmovzxbw %xmm0, %ymm0
There is a maze of duplicated and dead transforms and patterns in this
area. Remove the dead custom lowering of zext v8i16 to v8i32, that's
already handled by LowerAVXExtend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193262 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/movntdq-no-avx.ll')
0 files changed, 0 insertions, 0 deletions