[all-commits] [llvm/llvm-project] df8525: [mlir] Split up VectorToLLVM pass

[Krzysztof Drewniak via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: df852599f3a1c9faa3c20a1ac245e610a8ed3943
      https://github.com/llvm/llvm-project/commit/df852599f3a1c9faa3c20a1ac245e610a8ed3943
  Author: Krzysztof Drewniak <Krzysztof.Drewniak at amd.com>
  Date:   2023-09-13 (Wed, 13 Sep 2023)

  Changed paths:
    M mlir/include/mlir/Conversion/Passes.h
    M mlir/include/mlir/Conversion/VectorToLLVM/ConvertVectorToLLVM.h
    A mlir/include/mlir/Conversion/VectorToLLVM/ConvertVectorToLLVMPass.h
    M mlir/include/mlir/Dialect/SparseTensor/Pipelines/Passes.h
    M mlir/lib/Conversion/VectorToLLVM/CMakeLists.txt
    M mlir/lib/Conversion/VectorToLLVM/ConvertVectorToLLVMPass.cpp
    M mlir/lib/Dialect/SparseTensor/Pipelines/CMakeLists.txt
    M mlir/test/lib/Dialect/GPU/CMakeLists.txt
    M mlir/test/lib/Dialect/GPU/TestLowerToNVVM.cpp
    M mlir/test/lib/Dialect/LLVM/CMakeLists.txt
    M mlir/test/lib/Dialect/LLVM/TestLowerToLLVM.cpp
    M mlir/tools/mlir-vulkan-runner/CMakeLists.txt
    M mlir/tools/mlir-vulkan-runner/mlir-vulkan-runner.cpp

  Log Message:
  -----------
  [mlir] Split up VectorToLLVM pass

Currently, the VectorToLLVM patterns are built into a library along
with the corresponding pass, which also pulls in all the
platform-specific vector dialects (like AMXDialect) to apply all the
vector to LLVM conversions.

This causes dependency bloat when writing libraries - for example the
GPU to LLVM passes, which use the vector to LLVM patterns, don't need
the X86Vector dialect to be present at all.

This commit partitions the library into VectorToLLVM and
VectorToLLVMPass, where the latter pulls in all the other vector
transformations.

Reviewed By: nicolasvasilache, mehdi_amini

Differential Revision: https://reviews.llvm.org/D158287