projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 21f7b36) | patch
AMDGPU: Use default operands for clamp/omod
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Sat, 7 Sep 2019 22:42:27 +0000 (18:42 -0400)
committerMatt Arsenault <arsenm2@gmail.com>
Tue, 7 Jan 2020 01:22:13 +0000 (20:22 -0500)
commite93b1ffc8490d943690726370a0e9277fd78520d
tree760eae93be55895d48f3c3d5f6344c133f9daa51tree | snapshot
parent21f7b362095f7497aa87ccc334fe18bbc072854ccommit | diff
AMDGPU: Use default operands for clamp/omod

We have a lot of complex pattern variants that just set the source
modifiers that are really handled, and then set the output modifiers
to 0. We're unlikely to ever match output modifiers from the use
instruction side, and we already match clamp/omod in a separate pass.
llvm/lib/Target/AMDGPU/SIInstrInfo.td diff | blob | history
Domain: System / Toolchain;