[llvm] [AMDGPU] Vectorize i8 Shuffles (PR #95840)

Matt Arsenault via llvm-commits llvm-commits at lists.llvm.org
Thu Aug 22 06:32:10 PDT 2024


https://github.com/arsenm commented:

Can you rebase this and point me to the code in the vectorizer where this is making a difference? The SLP vectorizer's logic about what is vectorizable based on operation legality never really made sense to me.

https://github.com/llvm/llvm-project/pull/95840


More information about the llvm-commits mailing list