projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: e4296ea) | patch
X86-FMA3: Implemented commute transformations FMA*_Int instructions.
authorVyacheslav Klochkov <vyacheslav.n.klochkov@gmail.com>
Fri, 13 Nov 2015 00:07:35 +0000 (00:07 +0000)
committerVyacheslav Klochkov <vyacheslav.n.klochkov@gmail.com>
Fri, 13 Nov 2015 00:07:35 +0000 (00:07 +0000)
commitc81095daa785e47555daf68f36bb32fac3c4a19e
tree8de186037b6adf8340412e50664cd74ca97bb96etree | snapshot
parente4296ea350f7564ee679dadb48851bf33e27aee2commit | diff
X86-FMA3: Implemented commute transformations FMA*_Int instructions.
It made it possible to apply the memory folding optimization for the 2nd
operand of FMA*_Int instructions.

Reviewer: Quentin Colombet
Differential Revision: http://reviews.llvm.org/D14550

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@252973 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/X86InstrFMA.td diff | blob | history
lib/Target/X86/X86InstrInfo.cpp diff | blob | history
test/CodeGen/X86/fma-commute-x86.ll diff | blob | history
test/CodeGen/X86/fma-intrinsics-x86.ll diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors