projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 19ca80e) | patch
[AMDGPU] Allow tied operand subreg folding
authorStanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com>
Tue, 22 Oct 2019 18:07:15 +0000 (11:07 -0700)
committerStanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com>
Tue, 22 Oct 2019 18:27:36 +0000 (11:27 -0700)
commit48f57138be55a939afc64d357f7b26220206127a
tree2ca3176f861ebfc0ae3f34aa6552471db6dd0e2btree | snapshot
parent19ca80ef0575b376e135271c7d44799803407941commit | diff
[AMDGPU] Allow tied operand subreg folding

Turns out it makes sense, contrarily to what comment said.

Differential Revision: https://reviews.llvm.org/D69287
llvm/lib/Target/AMDGPU/SIFoldOperands.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/fold-vgpr-copy.mir diff | blob | history
Domain: System / Toolchain;