[all-commits] [llvm/llvm-project] 35f073: [AMDGPU] Fold a constant add/sub into the sudot4/s...

[Harrison Hao via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 35f073eb07f641713513fefb1ae4f4a5eb0bc604
      https://github.com/llvm/llvm-project/commit/35f073eb07f641713513fefb1ae4f4a5eb0bc604
  Author: Harrison Hao <57025411+harrisonGPU at users.noreply.github.com>
  Date:   2026-09-25 (Fri, 25 Sep 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp
    M llvm/test/Transforms/InstCombine/AMDGPU/llvm.amdgcn.sudot.ll

  Log Message:
  -----------
  [AMDGPU] Fold a constant add/sub into the sudot4/sudot8 accumulator (#225322)

Fold a constant add into the accumulator operand of sudot4 and sudot8
when
clamping is disabled:
```
  sudot(a, b, C1, false) + C2 -> sudot(a, b, C1 + C2, false)
```
Subtraction by a constant is canonicalized to addition of its negation.



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