projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: f4646d9) | patch
The memory alignment requirement on some of the mov{h|l}p{d|s} patterns are 16-byte...
authorEvan Cheng <evan.cheng@apple.com>
Wed, 28 Jan 2009 08:35:02 +0000 (08:35 +0000)
committerEvan Cheng <evan.cheng@apple.com>
Wed, 28 Jan 2009 08:35:02 +0000 (08:35 +0000)
commit1632782fe9a845e545f0ebd05fb4dc304adb5bd2
treec0fef539fd12b677df7b6e0f5cdb16e6b411ce4atree | snapshot
parentf4646d9902fcdcb121dbfbabbc2ee00bb41d7ca1commit | diff
The memory alignment requirement on some of the mov{h|l}p{d|s} patterns are 16-byte. That is overly strict. These instructions read / write f64 memory locations without alignment requirement.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@63195 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/README-SSE.txt diff | blob | history
lib/Target/X86/X86InstrSSE.td diff | blob | history
test/CodeGen/X86/swizzle.ll [new file with mode: 0644] blob
C/C++ LLVM-based compilers that forbids OOTA behaviors