Multi <scp>GPU</scp> Sparse Matrix by Sparse Matrix Multiplication
ABSTRACT The paper focuses on the improvement of the existing nsparse Nagasaka et al. algorithm and its extension to the multi‐GPU setting for the application of real engineering problems. In this work, we propose a distributed multi‐GPU framework for SpGEMM that is designed specifically for the nsparse like algorithms. The results show ∼2 times speed‐up for nsparse and close to ideal scalability of the multi‐GPU extension with the number of GPUs. Finally, we test the proposed algorithm in the AMG setting by computing the double SpGEMM product.
Authors
- Alessandro Celestini (ORCID: https://orcid.org/0000-0002-8493-4550)
- Massimo Bernaschi (ORCID: https://orcid.org/0000-0003-3661-9836)
- Giovanni Isotton (ORCID: https://orcid.org/0000-0002-1548-2378)
- Carlo Janna (ORCID: https://orcid.org/0000-0001-7286-8122)
- Artem Mavliutov
Institutions
- University of Padua (IT)
- Institute for High Performance Computing and Networking (IT)
- Confederazione Nazionale dell'Artigianato e Della Piccola e Media Impresa (IT)
Publication Details
- Journal
- Concurrency and Computation Practice and Experience
- Published
- 2025-09-30
- DOI
- https://doi.org/10.1002/cpe.70313
- Primary Topic
- Parallel Computing and Optimization Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00