[all-commits] [llvm/llvm-project] a20fea: [AMDGPU] Use WriteSALUDummy for v_div_scale* (#191...

[Jeffrey Byrnes via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: a20fea899e2adb34f604ef7e81e408701cba109b
      https://github.com/llvm/llvm-project/commit/a20fea899e2adb34f604ef7e81e408701cba109b
  Author: Jeffrey Byrnes <jeffrey.byrnes at amd.com>
  Date:   2026-04-13 (Mon, 13 Apr 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/VOP3Instructions.td
    M llvm/test/CodeGen/AMDGPU/GlobalISel/fdiv.f16.ll
    M llvm/test/CodeGen/AMDGPU/GlobalISel/fdiv.f32.ll
    M llvm/test/CodeGen/AMDGPU/GlobalISel/fdiv.f64.ll
    M llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.div.scale.ll
    M llvm/test/CodeGen/AMDGPU/bf16.ll
    M llvm/test/CodeGen/AMDGPU/dagcombine-select.ll
    M llvm/test/CodeGen/AMDGPU/fdiv.f16.ll
    M llvm/test/CodeGen/AMDGPU/fdiv.ll
    M llvm/test/CodeGen/AMDGPU/fdiv_flags.f32.ll
    M llvm/test/CodeGen/AMDGPU/freeze-binary.ll
    M llvm/test/CodeGen/AMDGPU/repeated-divisor.ll
    M llvm/test/CodeGen/AMDGPU/rsq.f32-safe.ll
    M llvm/test/CodeGen/AMDGPU/rsq.f64.ll
    A llvm/test/CodeGen/AMDGPU/schedmodel-dummywrite.mir
    M llvm/test/tools/llvm-mca/AMDGPU/gfx10-double.s
    M llvm/test/tools/llvm-mca/AMDGPU/gfx11-double.s

  Log Message:
  -----------
  [AMDGPU] Use WriteSALUDummy for v_div_scale* (#191670)

This uses the new HWWriteRes for v_div_scale*

For an explanation of why we want to do this , see
https://github.com/llvm/llvm-project/pull/190095 . In short, the
scheduler will not try to cover the full latency of the instructions
without this new modelling.

For a clear example of this, see the changes to
llvm/test/CodeGen/AMDGPU/schedmodel-dummywrite.mir in the git log of
this PR.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications