mlir-graphblas

  • Installation
  • GraphBLAS Dialect
  • Tools

Contents:

  • GraphBLAS Dialect Op Reference
  • Supported GraphBLAS Spec Operations
  • GraphBLAS Passes and graphblas-opt
  • GraphBLAS Dialect Tutorials
    • GraphBLAS Lowering Pass
      • Python Utilities for MLIR’s Sparse Tensors
      • Working with Sparse Layouts in the GraphBLAS Dialect
      • Vector Ops
      • Matrix Ops
      • Debugging Ops
    • GraphBLAS Structuralizing Pass
      • Lowering graphblas.matrix_multiply to Generic Form
      • Lowering graphblas.apply to Generic Form
      • Lowering graphblas.reduce_to_scalar to Generic Form
    • GraphBLAS Optimizing Pass
      • Fusion of graphblas.matrix_select Ops
      • Fusing graphblas.matrix_multiply with graphblas.reduce_to_scalar_generic
      • Fusing graphblas.matrix_multiply with graphblas.apply

GraphBLAS Dialect¶

The graphblas dialect is designed to make it possible to express GraphBLAS algorithms in MLIR in a compact way. The dialect does not define any new types, but rather operates on MLIR sparse tensors.

Contents:

  • GraphBLAS Dialect Op Reference
    • Assumptions
    • Operation Definitions
  • Supported GraphBLAS Spec Operations
  • GraphBLAS Passes and graphblas-opt
    • --graphblas-structuralize Pass
    • --graphblas-optimize Pass
    • --graphblas-lower Pass
  • GraphBLAS Dialect Tutorials
    • GraphBLAS Lowering Pass
    • GraphBLAS Structuralizing Pass
    • GraphBLAS Optimizing Pass

previous

Installation

next

GraphBLAS Dialect Op Reference

© Copyright 2021, Anaconda, Inc.

Created using Sphinx 3.0.4.