projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 143619a) | patch
AMDGPU: Don't create intermediate SALU instructions
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Wed, 26 Aug 2015 20:47:50 +0000 (20:47 +0000)
committerMatt Arsenault <Matthew.Arsenault@amd.com>
Wed, 26 Aug 2015 20:47:50 +0000 (20:47 +0000)
commit0017a91afc2fe470846fd188d5af6dd121f1d062
tree19eaca7f9dc4e28935d3279a0897a24cd3c0b58etree | snapshot
parent143619a66d07b6d381083828c1810ab5809cf7cdcommit | diff
AMDGPU: Don't create intermediate SALU instructions

When splitting 64-bit operations, create the correct
VALU instructions immediately.

This was splitting things like s_or_b64 into the two
s_or_b32s and then pushing the new instructions
onto the worklist. There's no reason we need
to do this intermediate step.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@246077 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/AMDGPU/SIInstrInfo.cpp diff | blob | history
lib/Target/AMDGPU/SIInstrInfo.h diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors