diff options
| author | Jay Foad <jay.foad@amd.com> | 2023-11-16 12:50:26 +0000 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-11-16 12:50:26 +0000 |
| commit | be2388c0d9923ed02fa4723fe9a06f07a0491fe5 (patch) | |
| tree | fe5f94514837052ccba4f23607a60a8def3dfd85 /llvm/lib/Bitcode/Reader/BitcodeAnalyzer.cpp | |
| parent | 1b82cc1186b8bf716205c9b86b851db667792a64 (diff) | |
| download | llvm-be2388c0d9923ed02fa4723fe9a06f07a0491fe5.zip llvm-be2388c0d9923ed02fa4723fe9a06f07a0491fe5.tar.gz llvm-be2388c0d9923ed02fa4723fe9a06f07a0491fe5.tar.bz2 | |
[AMDGPU] Prefer v_madak_f32 over v_madmk_f32 to reduce vgpr pressure (#72506)
As explained in the comment in SIInstrInfo::FoldImmediate, if we have a
choice between v_madak_f32 and v_madmk_f32 we should choose the former
so that the literal that is not folded into the instruction can be
materialized in an sgpr instead of a vgpr.
Diffstat (limited to 'llvm/lib/Bitcode/Reader/BitcodeAnalyzer.cpp')
0 files changed, 0 insertions, 0 deletions