[all-commits] [llvm/llvm-project] 243199: [mlir][NVVM] Reject f16 WGMMA accumulators for BF1...

[Mehdi Amini via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 24319927cb0c4acbe9fb3f2bd042b07dad2a4f9f
      https://github.com/llvm/llvm-project/commit/24319927cb0c4acbe9fb3f2bd042b07dad2a4f9f
  Author: Mehdi Amini <joker.eph at gmail.com>
  Date:   2026-09-07 (Mon, 07 Sep 2026)

  Changed paths:
    M mlir/lib/Dialect/LLVMIR/IR/NVVMDialect.cpp
    M mlir/test/Dialect/LLVMIR/nvvm.mlir

  Log Message:
  -----------
  [mlir][NVVM] Reject f16 WGMMA accumulators for BF16 inputs (#220869)

WGMMA supports an f32 accumulator for BF16 multiplicands. The verifier
also accepted an f16 accumulator, allowing an unsupported operation
through dialect verification.

Restrict the BF16 case to f32 and add verifier coverage to the existing
NVVM dialect test.

Found by Coverity.

Assisted-by: Codex



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