[PATCH] D82194: [AMDGPU] Enable compare operations to be selected by divergence

[Jay Foad via Phabricator via llvm-commits]

foad added a comment.

In D82194#2111508 <https://reviews.llvm.org/D82194#2111508>, @michel.daenzer wrote:

> This change broke thousands of piglit gpu profile tests with Mesa radeonsi on Navi 14.


It also broke Vulkan CTS testing with LLPC on Navi 10, and as well as the failures it caused spurious debug output like:

  Test case 'dEQP-VK.subgroups.arithmetic.framebuffer.subgroupxor_u16vec4_tess_eval'..
  S_CMP_LG_U32 killed $sgpr2_sgpr3, 0, implicit-def $scc
  S_CMP_LG_U32 killed $sgpr0_sgpr1, 0, implicit-def $scc
    Pass (OK)
  
  [...]
  
  Test case 'dEQP-VK.subgroups.arithmetic.framebuffer.subgroupmax_int_tess_eval'..
  S_CMP_LG_U32 killed $sgpr2_sgpr3, 0, implicit-def $scc
  S_CMP_LG_U32 killed $sgpr0_sgpr1, 0, implicit-def $scc
    Fail (Failed!)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D82194/new/

https://reviews.llvm.org/D82194