[all-commits] [llvm/llvm-project] 87782b: [mlir][x86vector] AVX512-BF16 Dot op (#124800)

[Adam Siemieniuk via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 87782b216fd3e7a8f8b2de04d4af467b390e9a34
      https://github.com/llvm/llvm-project/commit/87782b216fd3e7a8f8b2de04d4af467b390e9a34
  Author: Adam Siemieniuk <adam.siemieniuk at intel.com>
  Date:   2025-01-29 (Wed, 29 Jan 2025)

  Changed paths:
    M mlir/include/mlir/Dialect/X86Vector/X86Vector.td
    M mlir/lib/Dialect/X86Vector/Transforms/LegalizeForLLVMExport.cpp
    A mlir/test/Dialect/X86Vector/dot-bf16.mlir
    M mlir/test/Dialect/X86Vector/legalize-for-llvm.mlir
    M mlir/test/Dialect/X86Vector/roundtrip.mlir
    M mlir/test/Target/LLVMIR/x86vector.mlir

  Log Message:
  -----------
  [mlir][x86vector] AVX512-BF16 Dot op (#124800)

Adds AVX512 bf16 dot-product operation and defines lowering to LLVM
intrinsics.

AVX512 intrinsic operation definition is extended with an optional
extension field that allows specifying necessary LLVM mnemonic suffix
e.g., `"bf16"` for `x86_avx512bf16_` intrinsics.



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