[llvm] [NVPTX] designate fabs and fneg as free (PR #121513)
Artem Belevich via llvm-commits
llvm-commits at lists.llvm.org
Tue Jan 7 14:10:49 PST 2025
================
@@ -261,6 +261,9 @@ class NVPTXTargetLowering : public TargetLowering {
return true;
}
+ bool isFAbsFree(EVT VT) const override { return true; }
+ bool isFNegFree(EVT VT) const override { return true; }
----------------
Artem-B wrote:
What do we get in SASS for those? I do not see any FABS/FNEG instructions in https://docs.nvidia.com/cuda/cuda-binary-utilities/index.html#hopper-instruction-set
If ptxas ends up lowering fabs/fneg to the logical ops, we may as well let LLVM do it which may let it optimize some of that away better than ptxas.
https://github.com/llvm/llvm-project/pull/121513
More information about the llvm-commits
mailing list