[llvm] [AMDGPU] Fix OpenCL conformance test failures for ctlz. (PR #83170)
Leon Clark via llvm-commits
llvm-commits at lists.llvm.org
Tue Apr 9 08:25:37 PDT 2024
PeddleSpam wrote:
> > The version with the shift works fine for ctlz_zero_undef (and it would still be good to generate it since it's usually shorter) but not ctlz.
>
> See #83039 which optimizes generic codegen for ctlz_zero_undef to use the version with the shift.
Thanks, I'll take a look.
https://github.com/llvm/llvm-project/pull/83170
More information about the llvm-commits
mailing list