AMDGPU/SI: Avoid moving PHIs to VALU when phi values are defined in scalar branches
authorTom Stellard <thomas.stellard@amd.com>
Tue, 29 Nov 2016 00:46:46 +0000 (00:46 +0000)
committerTom Stellard <thomas.stellard@amd.com>
Tue, 29 Nov 2016 00:46:46 +0000 (00:46 +0000)
commit0bc688116c95d74a1b72a650607eab1f8fdc2a33
tree241502224e7feda1d8e6409af6202a7cd76884a4
parent3666e9c54b61a5b90387213d586c6a0751462178
AMDGPU/SI: Avoid moving PHIs to VALU when phi values are defined in scalar branches

Reviewers: arsenm

Subscribers: arsenm, llvm-commits, kzhuravl

Differential Revision: https://reviews.llvm.org/D23417

llvm-svn: 288095
llvm/lib/Target/AMDGPU/SIFixSGPRCopies.cpp
llvm/test/CodeGen/AMDGPU/branch-relaxation.ll
llvm/test/CodeGen/AMDGPU/cf-loop-on-constant.ll
llvm/test/CodeGen/AMDGPU/coalescer_remat.ll
llvm/test/CodeGen/AMDGPU/hoist-cond.ll
llvm/test/CodeGen/AMDGPU/lds-m0-init-in-loop.ll
llvm/test/CodeGen/AMDGPU/loop_break.ll
llvm/test/CodeGen/AMDGPU/uniform-cfg.ll
llvm/test/CodeGen/AMDGPU/uniform-loop-inside-nonuniform.ll
llvm/test/CodeGen/AMDGPU/valu-i1.ll