[all-commits] [llvm/llvm-project] 4b3eae: [mlir][AMDGPU] Define wrappers for WMMA matrix ops

[Giuseppe Rossini via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 4b3eaee2701afa2549c9cd0f78692598e9bfd44e
      https://github.com/llvm/llvm-project/commit/4b3eaee2701afa2549c9cd0f78692598e9bfd44e
  Author: Giuseppe Rossini <giuseppe.rossini at amd.com>
  Date:   2023-07-20 (Thu, 20 Jul 2023)

  Changed paths:
    M mlir/include/mlir/Dialect/AMDGPU/IR/AMDGPU.td
    M mlir/include/mlir/Dialect/LLVMIR/ROCDLOps.td
    M mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
    M mlir/lib/Dialect/AMDGPU/IR/AMDGPUDialect.cpp
    A mlir/test/Conversion/AMDGPUToROCDL/wmma.mlir
    M mlir/test/Dialect/AMDGPU/invalid.mlir
    M mlir/test/Dialect/AMDGPU/ops.mlir
    M mlir/test/Target/LLVMIR/rocdl.mlir

  Log Message:
  -----------
  [mlir][AMDGPU] Define wrappers for WMMA matrix ops

Wave Matrix Multiply Accumulate (WMMA) is the instruction to accelerate
matrix multiplication on RDNA3 architectures.  LLVM already provides a
set of intrinsics to generate wmma instructions. This change uses those
intrinsics to enable the feature in MLIR.

Reviewed By: krzysz00

Differential Revision: https://reviews.llvm.org/D152451