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

[Carl Ritson via llvm-branch-commits]

perlfu wrote:

What the motivation here?
At the moment early-machinelicm runs before WWM region formation, so the test is speculative on the basis that it *might* run after WQM pass? Or is this a proxy for machinelicm that runs after RA?

My mid-term plan is to separate WWM and WQM formation.  Locking down stuff in WWM region as this patch does is probably sensible.  However, I haven't seen an argument for moving WWM region formation earlier -- exactly because it blocks various optimizations which are AFAIK safe until marking and region formation has been done.

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