projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: c20323a) | patch
Remove intrinsic specific instructions for SSE/SSE2/AVX floating point max/min instru...
authorCraig Topper <craig.topper@gmail.com>
Sat, 29 Dec 2012 16:44:25 +0000 (16:44 +0000)
committerCraig Topper <craig.topper@gmail.com>
Sat, 29 Dec 2012 16:44:25 +0000 (16:44 +0000)
commit6d183e400720b703dc9dbe6c8a28b615441601a2
tree26fa06238c49006592f44eb6893a7b2ed94c8d6ftree | snapshot
parentc20323a7588bc5ad327155fe09c2b51435c3ada2commit | diff
Remove intrinsic specific instructions for SSE/SSE2/AVX floating point max/min instructions. Lower them to target specific nodes and use those patterns instead. This also allows them to be commuted if UnsafeFPMath is enabled.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171227 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/X86ISelLowering.cpp diff | blob | history
lib/Target/X86/X86InstrInfo.cpp diff | blob | history
lib/Target/X86/X86InstrSSE.td diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors