[PATCH] D157265: [AMDGPU] Reorder atomic optimizer to avoid CAS loop.

[Jay Foad via Phabricator via llvm-commits]

foad added a comment.

> Expand-Atomic pass emits the CAS loop for FP operations
> which limits the optimizations offered by atomic optimizer.
>
> Moving atomic optimizer before expand-atomics allows
> better codegen.

So the intention is that you still get a CAS loop, but the whole loop is executed by a single lane, instead of switching into and out of single-lane mode each time around the loop?

Makes sense to me.



================
Comment at: llvm/test/CodeGen/AMDGPU/atomic_optimizations_pixelshader.ll:44
 ; GFX7-NEXT:    s_wqm_b64 s[4:5], -1
+; GFX7-NEXT:    s_and_b64 s[4:5], s[4:5], s[4:5]
 ; GFX7-NEXT:    s_andn2_b64 vcc, exec, s[4:5]
----------------
It would be nice to at least understand where these weird redundant instructions are coming from.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D157265/new/

https://reviews.llvm.org/D157265