projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 7e8cebf) | patch
ARM cost model: Unaligned vectorized double stores are expensive
authorArnold Schwaighofer <aschwaighofer@apple.com>
Tue, 29 Oct 2013 01:33:57 +0000 (01:33 +0000)
committerArnold Schwaighofer <aschwaighofer@apple.com>
Tue, 29 Oct 2013 01:33:57 +0000 (01:33 +0000)
commitc04d241d13820b33224b5cbd89a427fc08e5d1d9
treedbf40b8b49ab3278f68615ad0af8db89758d9f07tree | snapshot
parent7e8cebf22d170769b0bf0c2a69309faa0e36ac4ccommit | diff
ARM cost model: Unaligned vectorized double stores are expensive

Updated a test case that assumed that <2 x double> would vectorize to use
<4 x float>.

radar://15338229

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193574 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/ARM/ARMTargetTransformInfo.cpp diff | blob | history
test/Transforms/LoopVectorize/ARM/width-detect.ll diff | blob | history
test/Transforms/SLPVectorizer/ARM/memory.ll [new file with mode: 0644] blob
C/C++ LLVM-based compilers that forbids OOTA behaviors