[all-commits] [llvm/llvm-project] 171a5a: [mlir][Linalg] Add a greedy transform to map copie...

[Nicolas Vasilache via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 171a5a761d24664786f65c861d083df306532a35
      https://github.com/llvm/llvm-project/commit/171a5a761d24664786f65c861d083df306532a35
  Author: Nicolas Vasilache <nicolasvasilache at users.noreply.github.com>
  Date:   2023-07-10 (Mon, 10 Jul 2023)

  Changed paths:
    A mlir/include/mlir/Dialect/Linalg/TransformOps/GPUHeuristics.h
    M mlir/include/mlir/Dialect/Linalg/TransformOps/LinalgTransformOps.h
    M mlir/include/mlir/Dialect/Linalg/TransformOps/LinalgTransformOps.td
    M mlir/lib/Dialect/Linalg/TransformOps/CMakeLists.txt
    A mlir/lib/Dialect/Linalg/TransformOps/GPUHeuristics.cpp
    M mlir/lib/Dialect/Linalg/TransformOps/LinalgTransformOps.cpp
    A mlir/test/Dialect/Linalg/transform-op-gpu-map-copy-to-threads.mlir

  Log Message:
  -----------
  [mlir][Linalg] Add a greedy transform to map copies to threads efficiently.

This revision adds a new transformation to map a copy operation to a gpu grid of threads.
It implements a first heuristic that allows trading off coalesced accesses vs predication and occupancy.

Differential Revision: https://reviews.llvm.org/D154836