[llvm] [NVPTX] Support copysign PTX instruction (PR #107800)

Alex MacLean via llvm-commits llvm-commits at lists.llvm.org
Mon Sep 9 15:51:35 PDT 2024


AlexMaclean wrote:

> LGTM.
> 
> For my own education -- is there hardware support for this instruction in SASS? Or would the copysign get expanded to the bit twiddling LLVM does?

Here is a quick comparison of the SASS before/after this change: https://godbolt.org/z/snWTGEG81. copysign is still getting expanded but it looks like `ptxas` is doing a better job then the SelectionDAG (2 vs. 3 instructions). 

https://github.com/llvm/llvm-project/pull/107800


More information about the llvm-commits mailing list