[PATCH] D153953: Revert "[AMDGPU] Mark mbcnt as convergent"

[Yaxun Liu via Phabricator via cfe-commits]

yaxunl accepted this revision.
yaxunl added a comment.
This revision is now accepted and ready to land.

I checked the ISA manual about mbcnt and I agree that its value should only depend on thread position in warp and not on CFG.

It is possible that the side effect of preventing it from merging fixed something by accident, but marking mbcnt as convergent seems not right. Therefore I agree to revert that patch.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D153953/new/

https://reviews.llvm.org/D153953