[all-commits] [llvm/llvm-project] 20daed: 2d Arm Neon sdot op, and lowering to the intrinsic.

bjacob via All-commits all-commits at lists.llvm.org
Thu Jun 10 14:37:05 PDT 2021


  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 20daedacca803b81db6d8773b705345702bf0fc3
      https://github.com/llvm/llvm-project/commit/20daedacca803b81db6d8773b705345702bf0fc3
  Author: Benoit Jacob <benoitjacob at google.com>
  Date:   2021-06-10 (Thu, 10 Jun 2021)

  Changed paths:
    A mlir/include/mlir/Conversion/ArmNeon2dToIntr/ArmNeon2dToIntr.h
    M mlir/include/mlir/Conversion/Passes.h
    M mlir/include/mlir/Conversion/Passes.td
    M mlir/include/mlir/Dialect/ArmNeon/ArmNeon.td
    A mlir/lib/Conversion/ArmNeon2dToIntr/ArmNeon2dToIntr.cpp
    A mlir/lib/Conversion/ArmNeon2dToIntr/CMakeLists.txt
    M mlir/lib/Conversion/CMakeLists.txt
    M mlir/lib/Conversion/PassDetail.h
    A mlir/test/Dialect/ArmNeon/invalid.mlir
    A mlir/test/Target/LLVMIR/arm-neon-2d.mlir

  Log Message:
  -----------
  2d Arm Neon sdot op, and lowering to the intrinsic.

This adds Sdot2d op, which is similar to the usual Neon
intrinsic except that it takes 2d vector operands, reflecting the
structure of the arithmetic that it's performing: 4 separate
4-dimensional dot products, whence the vector<4x4xi8> shape.

This also adds a new pass, arm-neon-2d-to-intr, lowering
this new 2d op to the 1d intrinsic.

Reviewed By: nicolasvasilache

Differential Revision: https://reviews.llvm.org/D102504




More information about the All-commits mailing list