wenju-he wrote: The old code is passing on intel gpu and probably also on amd gpu as the code is ported from amdgpu repo.
> On some target where vector select(cond, ret, 0.0f) is lowered to ret * (cond > ? 1.0f : 0.0f) Could it be a backend lowering bug? https://github.com/llvm/llvm-project/pull/233787 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
