[llvm-branch-commits] [llvm] [AMDGPU] Mark all instructions in WWM region as convergent (PR #204572)

[Matt Arsenault via llvm-branch-commits]

arsenm wrote:

There certainly is an argument for marking all VALU instructions as convergent, and then emitting noconvergent in all of the typical cases 

https://github.com/llvm/llvm-project/pull/204572