diff options
| author | Jakob Stoklund Olesen <stoklund@2pi.dk> | 2011-11-07 23:08:21 +0000 |
|---|---|---|
| committer | Jakob Stoklund Olesen <stoklund@2pi.dk> | 2011-11-07 23:08:21 +0000 |
| commit | b26c7727c9a45613d9bae69995cfd719c57c5614 (patch) | |
| tree | fc4faf0805970b46a9cea0314099f74ee0e17897 /test/CodeGen/X86/sse2-blend.ll | |
| parent | a29fc806fe02cea76f7896b7e344bb919dd7ac25 (diff) | |
| download | llvm-b26c7727c9a45613d9bae69995cfd719c57c5614.tar.gz llvm-b26c7727c9a45613d9bae69995cfd719c57c5614.tar.bz2 llvm-b26c7727c9a45613d9bae69995cfd719c57c5614.tar.xz | |
Kill and collapse outstanding DomainValues.
DomainValues that are only used by "don't care" instructions are now
collapsed to the first possible execution domain after all basic blocks
have been processed. This typically means the PS domain on x86.
For example, the vsel_i64 and vsel_double functions in sse2-blend.ll are
completely collapsed to the PS domain instead of containing a mix of
execution domains created by isel.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144037 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/sse2-blend.ll')
| -rw-r--r-- | test/CodeGen/X86/sse2-blend.ll | 11 |
1 files changed, 4 insertions, 7 deletions
diff --git a/test/CodeGen/X86/sse2-blend.ll b/test/CodeGen/X86/sse2-blend.ll index 4ff1d035e4..2f4317bf29 100644 --- a/test/CodeGen/X86/sse2-blend.ll +++ b/test/CodeGen/X86/sse2-blend.ll @@ -26,11 +26,10 @@ define void@vsel_i32(<4 x i32>* %v1, <4 x i32>* %v2) { ret void } -; FIXME: The -mattr=+sse2,-sse41 disable the ExecutionDepsFix pass causing the -; mixed domains here. +; Without forcing instructions, fall back to the preferred PS domain. ; CHECK: vsel_i64 ; CHECK: xorps -; CHECK: pand +; CHECK: andps ; CHECK: andnps ; CHECK: orps ; CHECK: ret @@ -43,16 +42,14 @@ define void@vsel_i64(<4 x i64>* %v1, <4 x i64>* %v2) { ret void } -; FIXME: The -mattr=+sse2,-sse41 disable the ExecutionDepsFix pass causing the -; mixed domains here. +; Without forcing instructions, fall back to the preferred PS domain. ; CHECK: vsel_double ; CHECK: xorps -; CHECK: pand +; CHECK: andps ; CHECK: andnps ; CHECK: orps ; CHECK: ret - define void@vsel_double(<4 x double>* %v1, <4 x double>* %v2) { %A = load <4 x double>* %v1 %B = load <4 x double>* %v2 |