[llvm-branch-commits] [llvm] [AMDGPU] Optimize i64 uitofp for unsigned byte values (PR #222899)

[Matt Arsenault via llvm-branch-commits]

https://github.com/arsenm commented:

It seems like this should have been a separate optimization somewhere else to reduce the source type of the conversion to start with

https://github.com/llvm/llvm-project/pull/222899