Michele Martone
Leibniz Supercomputing Centre (LRZ), Garching near Munich, Germany

Simone Bacchio
CaSToRC, The Cyprus Institute, Nicosia, Cyprus

Abstract

This article introduces PyRSB, a Python interface to the LIBRSB library. LIBRSB is a portable performance library offering so called Sparse BLAS (Sparse Basic Linear Algebra Subprograms) operations for modern multicore CPUs. It is based on the Recursive Sparse Blocks (RSB) format, which is particularly well suited for matrices of large dimensions. PyRSB allows LIBRSB usage with an interface styled after that of SciPy's sparse matrix classes, and offers the extra benefit of exploiting multicore parallelism. This article introduces the concepts behind the RSB format and LIBRSB, and illustrates usage of PyRSB. It concludes with a user-oriented overview of speedup advantage of rsb\_matrix over scipy.sparse.csr\_matrix running general sparse matrix-matrix multiplication on a modern shared-memory computer.

Keywords

DOI

10.25080/majora-1b6fd038-00e

Bibtex entry

Full text PDF

Copyright © 2021 Michele Martone et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.