| Issue | 129984 |
|---|---|
| Summary | AMDGPU should try to make use of s_cmov_b32, s_cmov_b64, and s_cmovk_i32 |
| Labels | backend:AMDGPU, missed-optimization |
| Assignees | |
| Reporter | arsenm |
We do not try to make use of these at all. We should form them from uniform selects, and convert them to s_cselect_b32/b64 in convertToThreeAddress
_______________________________________________ llvm-bugs mailing list llvm-bugs@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
