[PATCH] D125567: [AMDGPU] Shrink MAD/FMA to MADAK/MADMK/FMAAK/FMAMK on GFX10

Jay Foad via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon May 16 06:22:01 PDT 2022


foad added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIShrinkInstructions.cpp:809
+
+      // TODO: Also shrink F16 forms.
+      if (MI.getOpcode() == AMDGPU::V_MAD_F32_e64 ||
----------------
arsenm wrote:
> Are you going to do this in a separate change?
Sure, can do, but there are no in-tree tests for it, and no use for it in the graphics shaders I've been looking at either.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D125567/new/

https://reviews.llvm.org/D125567



More information about the llvm-commits mailing list