[all-commits] [llvm/llvm-project] 114ba7: [mlir][NVGPU] Handle native mma.sync and ldmatrix(...

[Manish Gupta via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 114ba722c1e58d23bafdf3654e4f8e537150c318
      https://github.com/llvm/llvm-project/commit/114ba722c1e58d23bafdf3654e4f8e537150c318
  Author: Manish Gupta <manigupta at google.com>
  Date:   2022-10-19 (Wed, 19 Oct 2022)

  Changed paths:
    M mlir/include/mlir/Dialect/NVGPU/Utils/MMAUtils.h
    M mlir/lib/Conversion/VectorToGPU/VectorToGPU.cpp
    M mlir/lib/Dialect/NVGPU/Utils/MMAUtils.cpp
    M mlir/test/Conversion/VectorToGPU/vector-to-mma-ops-mma-sync.mlir

  Log Message:
  -----------
  [mlir][NVGPU] Handle native mma.sync and ldmatrix(x4) sizes

This patch handles native `mma.sync` sizes and enables issuing `ldmatrix` on
largest possible tiles for matrixB. It requires handling
`vector.extract_strided_slice` from vector to ngpu lowering.

Differential Revision: https://reviews.llvm.org/D135749