projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 14ceb45) | patch
AMDGPU/SI: Don't move copies of immediates to the VALU
authorTom Stellard <thomas.stellard@amd.com>
Tue, 6 Dec 2016 21:13:30 +0000 (21:13 +0000)
committerTom Stellard <thomas.stellard@amd.com>
Tue, 6 Dec 2016 21:13:30 +0000 (21:13 +0000)
commit00cfa747156c8a7bb25be3e6a85272a61b0b15ca
tree3f9aaf7d2c08c06a3c88cf8b2bf7d81e7cb1287dtree | snapshot
parent14ceb45fb491ea54a61efb9e32c7b1dc902fa009commit | diff
AMDGPU/SI: Don't move copies of immediates to the VALU

Summary:
If we write an immediate to a VGPR and then copy the VGPR to an
SGPR, we can replace the copy with a S_MOV_B32 sgpr, imm, rather than
moving the copy to the SALU.

Reviewers: arsenm

Subscribers: kzhuravl, wdng, nhaehnle, yaxunl, llvm-commits, tony-tye

Differential Revision: https://reviews.llvm.org/D27272

llvm-svn: 288849
llvm/lib/Target/AMDGPU/SIFixSGPRCopies.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/salu-to-valu.ll diff | blob | history
Domain: System / Toolchain;