Publications of Torsten Hoefler

Lukas Gianinazzi, Alexandros Nikolaos Ziogas, Piotr Luczynski, Langwen Huang, Saleh Ashkboos, Florian Scheidl, Armon Carigiet, Chio Ge, Nabil Abubaker, Maciej Besta, Tal Ben-Nun, Torsten Hoefler:

Arrow Matrix Decomposition: A Novel Approach for Communication-Efficient Sparse Matrix Multiplication

(In The Proceedings of the 2024 USENIX Annual Technical Conference, presented in Edinburgh, United Kingdom, pages 404-416, Association for Computing Machinery, Mar. 2024)

Publisher Reference

Abstract

We propose a novel approach to iterated sparse matrix dense matrix multiplication, a fundamental computational kernel in scientific computing and graph neural network training. In cases where matrix sizes exceed the memory of a single compute node, data transfer becomes a bottleneck. An approach based on dense matrix multiplication algorithms leads to suboptimal scalability and fails to exploit the sparsity in the problem. To address these challenges, we propose decomposing the sparse matrix into a small number of highly structured matrices called arrow matrices, which are connected by permutations. Our approach enables communication-avoiding multiplications, achieving a polynomial reduction in communication volume per iteration for matrices corresponding to planar graphs and other minor-excluded families of graphs. Our evaluation demonstrates that our approach outperforms a state-of-the-art method for sparse matrix multiplication on matrices with hundreds of millions of rows, offering near-linear strong and weak scaling.

Documents

Publisher URL: https://dl.acm.org/doi/10.1145/3627535.3638496download article:

download slides:

BibTeX

@inproceedings{gianinazzi2024arrow,
  author={Lukas Gianinazzi and Alexandros Nikolaos Ziogas and Piotr Luczynski and Langwen Huang and Saleh Ashkboos and Florian Scheidl and Armon Carigiet and Chio Ge and Nabil Abubaker and Maciej Besta and Tal Ben-Nun and Torsten Hoefler},
  title={{Arrow Matrix Decomposition: A Novel Approach for Communication-Efficient Sparse Matrix Multiplication}},
  year={2024},
  month={Mar.},
  pages={404-416},
  booktitle={The Proceedings of the 2024 USENIX Annual Technical Conference},
  location={Edinburgh, United Kingdom},
  publisher={Association for Computing Machinery},
  source={http://www.unixer.de/~htor/publications/},
}


serving: 3.137.162.21:45740	© Torsten Hoefler