[all-commits] [llvm/llvm-project] a61cc1: [AMDGPU][True16][CodeGen] Skip combineDpp with t16...

[Brox Chen via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: a61cc1b99a50d832c650132cc9956320bfe594f5
      https://github.com/llvm/llvm-project/commit/a61cc1b99a50d832c650132cc9956320bfe594f5
  Author: Brox Chen <guochen2 at amd.com>
  Date:   2025-03-31 (Mon, 31 Mar 2025)

  Changed paths:
    M llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp
    A llvm/test/CodeGen/AMDGPU/dpp_combine-true16.mir
    M llvm/test/CodeGen/AMDGPU/dpp_combine.ll
    A llvm/test/CodeGen/AMDGPU/vopc_dpp-true16.mir

  Log Message:
  -----------
  [AMDGPU][True16][CodeGen] Skip combineDpp with t16 instructions (#128918)

We only emits v_mov_b32/64_dpp. Don't combine t16 instructions with mov
dpp. Update the test inputs to be legal.

It is future work to emit v_mov_b16_dpp, and then update GCNDPPCombine
to combine it with the 16-bit instructions.



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