EUVIMEDEUVIMED entdecken
Crossref

Collective Algorithms for Multiported Torus Networks

Paul Sack, William Gropp

ACM Transactions on Parallel Computing · 2015

DOI: 10.1145/2686882 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ACM Transactions on Parallel Computing, 2015. Verfasst von Paul Sack und 1 weiteren Autor:innen. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 20 Literaturangaben. Gefördert von: Department of Energy, National Science Foundation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Modern supercomputers with torus networks allow each node to simultaneously pass messages on all of its links. However, most collective algorithms are designed to only use one link at a time. In this work, we present novel multiported algorithms for the scatter, gather, all-gather, and reduce-scatter operations. Our algorithms can be combined to create multiported reduce, all-reduce, and broadcast algorithms. Several of these algorithms involve a new technique where we relax the MPI message-ordering constraints to achieve high performance and restore the correctordering using an additional stage of redundant communication. According to our models, on an n -dimensional torus, our algorithms should allow for nearly a 2 n -fold improvement in communication performance compared to known, single-ported torus algorithms. In practice, we have achieved nearly 6x better performance on a 32k-node 3-dimensional torus.

Förderung

  • Department of Energy
  • National Science Foundation

Zitieren

APA

Sack, P., & Gropp, W. (2015). Collective Algorithms for Multiported Torus Networks. ACM Transactions on Parallel Computing. https://doi.org/10.1145/2686882

Vancouver

Sack P, Gropp W. Collective Algorithms for Multiported Torus Networks. ACM Transactions on Parallel Computing. 2015. doi:10.1145/2686882

BibTeX
@article{Sack2015,
  author = {Sack, Paul and Gropp, William},
  title = {Collective Algorithms for Multiported Torus Networks},
  journal = {ACM Transactions on Parallel Computing},
  year = {2015},
  doi = {10.1145/2686882}
}
Kostenlos starten →