[PATCH] D157683: [AMDGPU] W/a for gfx940 byte0 fp8 conversion bug

[Jay Foad via Phabricator via llvm-commits]

foad accepted this revision.
foad added inline comments.
This revision is now accepted and ready to land.


================
Comment at: llvm/lib/Target/AMDGPU/GCNSubtarget.h:1175
 
+  // \returns true is FP8/BF8 VOP1 form of conversion to F32 in unreliable.
+  bool hasCvtFP8VOP1Bug() const { return true; }
----------------
"true **if** ... **is** unreliable"


================
Comment at: llvm/lib/Target/AMDGPU/VOP1Instructions.td:589
     (f32 (node i32:$src, index)),
     !if (index,
          (inst_sdwa 0, $src, 0, 0, index),
----------------
Could simplify this since index is never 0 now.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D157683/new/

https://reviews.llvm.org/D157683