projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: d372e92) | patch
[AMDGPU] move hasAGPRs/hasVGPRs into header
authorStanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com>
Wed, 29 Sep 2021 19:43:12 +0000 (12:43 -0700)
committerStanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com>
Thu, 30 Sep 2021 17:02:02 +0000 (10:02 -0700)
commit244aa7f7358da791a231ffd4efd849f166050c4b
tree437ed0afa94b00692b509271f16446bf86f52536tree | snapshot
parentd372e92baf96cb0913aa361ecbc53a38a5f29ae1commit | diff
[AMDGPU] move hasAGPRs/hasVGPRs into header

It is now very simple and can go right into the header
allowing optimizer to combine callers, such as isVGPRClass
and similar.

It does not need anything from the TRI itself anymore, so
make it static class member along with the callers.

Differential Revision: https://reviews.llvm.org/D110762
llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp diff | blob | history
llvm/lib/Target/AMDGPU/SIRegisterInfo.h diff | blob | history
Domain: System / Toolchain;