Passes.h

Go to the documentation of this file.

//===- Passes.h - Sparse tensor pipeline entry points -----------*- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This header file defines prototypes of all sparse tensor pipelines.
//
//===----------------------------------------------------------------------===//
 
#ifndef MLIR_DIALECT_SPARSETENSOR_PIPELINES_PASSES_H_
#define MLIR_DIALECT_SPARSETENSOR_PIPELINES_PASSES_H_
 
#include "mlir/Conversion/VectorToLLVM/ConvertVectorToLLVMPass.h"
#include "mlir/Dialect/SparseTensor/Transforms/Passes.h"
#include "mlir/Pass/PassOptions.h"
 
using namespace mlir::detail;
using namespace llvm::cl;
 
namespace mlir {
namespace sparse_tensor {
 
/// Options for the "sparsifier" pipeline.  So far this only contains
/// a subset of the options that can be set for the underlying passes,
/// because it must be manually kept in sync with the tablegen files
/// for those passes.
struct SparsifierOptions : public PassPipelineOptions<SparsifierOptions> {
  // These options must be kept in sync with `SparsificationBase`.
  // TODO(57514): These options are duplicated in Passes.td.
  PassOptions::Option<mlir::SparseParallelizationStrategy> parallelization{
      *this, "parallelization-strategy",
      ::llvm::cl::desc("Set the parallelization strategy"),
      ::llvm::cl::init(mlir::SparseParallelizationStrategy::kNone),
      llvm::cl::values(
          clEnumValN(mlir::SparseParallelizationStrategy::kNone, "none",
                     "Turn off sparse parallelization."),
          clEnumValN(mlir::SparseParallelizationStrategy::kDenseOuterLoop,
                     "dense-outer-loop",
                     "Enable dense outer loop sparse parallelization."),
          clEnumValN(mlir::SparseParallelizationStrategy::kAnyStorageOuterLoop,
                     "any-storage-outer-loop",
                     "Enable sparse parallelization regardless of storage for "
                     "the outer loop."),
          clEnumValN(mlir::SparseParallelizationStrategy::kDenseAnyLoop,
                     "dense-any-loop",
                     "Enable dense parallelization for any loop."),
          clEnumValN(
              mlir::SparseParallelizationStrategy::kAnyStorageAnyLoop,
              "any-storage-any-loop",
              "Enable sparse parallelization for any storage and loop."))};
 
  PassOptions::Option<bool> enableRuntimeLibrary{
      *this, "enable-runtime-library",
      desc("Enable runtime library for manipulating sparse tensors"),
      init(true)};
 
  PassOptions::Option<bool> testBufferizationAnalysisOnly{
      *this, "test-bufferization-analysis-only",
      desc("Run only the inplacability analysis"), init(false)};
 
  PassOptions::Option<bool> enableBufferInitialization{
      *this, "enable-buffer-initialization",
      desc("Enable zero-initialization of memory buffers"), init(false)};
 
  // TODO: Delete the option, it should also be false after switching to
  // buffer-deallocation-pass
  PassOptions::Option<bool> createSparseDeallocs{
      *this, "create-sparse-deallocs",
      desc("Specify if the temporary buffers created by the sparse "
           "compiler should be deallocated. For compatibility with core "
           "bufferization passes. "
           "This option is only used when enable-runtime-library=false."),
      init(true)};
 
  PassOptions::Option<int32_t> vectorLength{
      *this, "vl", desc("Set the vector length (0 disables vectorization)"),
      init(0)};
 
  // These options must be kept in sync with the `ConvertVectorToLLVM`
  // (defined in include/mlir/Dialect/SparseTensor/Pipelines/Passes.h).
  PassOptions::Option<bool> reassociateFPReductions{
      *this, "reassociate-fp-reductions",
      desc("Allows llvm to reassociate floating-point reductions for speed"),
      init(false)};
  PassOptions::Option<bool> force32BitVectorIndices{
      *this, "enable-index-optimizations",
      desc("Allows compiler to assume indices fit in 32-bit if that yields "
           "faster code"),
      init(true)};
  PassOptions::Option<bool> amx{
      *this, "enable-amx",
      desc("Enables the use of AMX dialect while lowering the vector dialect"),
      init(false)};
  PassOptions::Option<bool> armNeon{
      *this, "enable-arm-neon",
      desc("Enables the use of ArmNeon dialect while lowering the vector "
           "dialect"),
      init(false)};
  PassOptions::Option<bool> armSVE{
      *this, "enable-arm-sve",
      desc("Enables the use of ArmSVE dialect while lowering the vector "
           "dialect"),
      init(false)};
  PassOptions::Option<bool> x86Vector{
      *this, "enable-x86vector",
      desc("Enables the use of X86Vector dialect while lowering the vector "
           "dialect"),
      init(false)};
 
  /// These options are used to enable GPU code generation.
  PassOptions::Option<std::string> gpuTriple{*this, "gpu-triple",
                                             desc("GPU target triple")};
  PassOptions::Option<std::string> gpuChip{*this, "gpu-chip",
                                           desc("GPU target architecture")};
  PassOptions::Option<std::string> gpuFeatures{*this, "gpu-features",
                                               desc("GPU target features")};
  /// For NVIDIA GPUs there are 3 compilation format options:
  /// 1. `isa`: the compiler generates PTX and the driver JITs the PTX.
  /// 2. `bin`: generates a CUBIN object for `chip=gpuChip`.
  /// 3. `fatbin`: generates a fat binary with a CUBIN object for `gpuChip` and
  /// also embeds the PTX in the fat binary.
  /// Notes:
  /// Option 1 adds a significant runtime performance hit, however, tests are
  /// more likely to pass with this option.
  /// Option 2 is better for execution time as there is no JIT; however, the
  /// program will fail if there's an architecture mismatch between `gpuChip`
  /// and the GPU running the program.
  /// Option 3 is the best compromise between options 1 and 2 as it can JIT in
  /// case of an architecture mismatch between `gpuChip` and the running
  /// architecture. However, it's only possible to JIT to a higher CC than
  /// `gpuChip`.
  PassOptions::Option<std::string> gpuFormat{
      *this, "gpu-format", desc("GPU compilation format"), init("fatbin")};
 
  /// This option is used to enable GPU library generation.
  PassOptions::Option<bool> enableGPULibgen{
      *this, "enable-gpu-libgen",
      desc("Enables GPU acceleration by means of direct library calls (like "
           "cuSPARSE)")};
 
  /// Projects out the options for `createSparsificationPass`.
  SparsificationOptions sparsificationOptions() const {
    return SparsificationOptions(parallelization, enableRuntimeLibrary);
  }
 
  /// Projects out the options for `createConvertVectorToLLVMPass`.
  ConvertVectorToLLVMPassOptions lowerVectorToLLVMOptions() const {
    ConvertVectorToLLVMPassOptions opts{};
    opts.reassociateFPReductions = reassociateFPReductions;
    opts.force32BitVectorIndices = force32BitVectorIndices;
    opts.armNeon = armNeon;
    opts.armSVE = armSVE;
    opts.amx = amx;
    opts.x86Vector = x86Vector;
    return opts;
  }
};
 
//===----------------------------------------------------------------------===//
// Building and Registering.
//===----------------------------------------------------------------------===//
 
/// Adds the "sparsifier" pipeline to the `OpPassManager`.  This
/// is the standard pipeline for taking sparsity-agnostic IR using
/// the sparse-tensor type and lowering it to LLVM IR with concrete
/// representations and algorithms for sparse tensors.
void buildSparsifier(OpPassManager &pm, const SparsifierOptions &options);
 
/// Registers all pipelines for the `sparse_tensor` dialect.  At present,
/// this includes only "sparsifier".
void registerSparseTensorPipelines();
 
} // namespace sparse_tensor
} // namespace mlir
 
#endif // MLIR_DIALECT_SPARSETENSOR_PIPELINES_PASSES_H_