[all-commits] [llvm/llvm-project] b15daa: [AMDGPU] Add scalar_to_vector pattern for v2bf16 (...

[Changpeng Fang via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: b15daae148e77f1ad2e20990094ddbeb8cc598ac
      https://github.com/llvm/llvm-project/commit/b15daae148e77f1ad2e20990094ddbeb8cc598ac
  Author: Changpeng Fang <changpeng.fang at amd.com>
  Date:   2026-09-24 (Thu, 24 Sep 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInstructions.td
    A llvm/test/CodeGen/AMDGPU/scalar_to_vector_v2bf16.ll

  Log Message:
  -----------
  [AMDGPU] Add scalar_to_vector pattern for v2bf16 (#225303)

A target DAGCombine rewrites `v2bf16 = scalar_to_vector` into bitcast(any_extend),
but it only runs at -O1 and above, so at -O0 the node reaches ISel and fails to select.
Add the missing pattern, mirroring the v2f16 one, for both fake-true16 and real-true16.

Fixes: ROCM-31415



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