Hierarchical load balancing for Charm++ applications on large supercomputers

Gengbin Zheng, Esteban Meneses, Abhinav Bhatelé, Laxmikant V Kale

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Large parallel machines with hundreds of thousands of processors are being built. Recent studies have shown that ensuring good load balance is critical for scaling certain classes of parallel applications on even thousands of processors. Centralized load balancing algorithms suffer from scalability problems, especially on machines with relatively small amount of memory. Fully distributed load balancing algorithms, on the other hand, tend to yield poor load balance on very large machines. In this paper, we present an automatic dynamic hierarchical load balancing method that overcomes the scalability challenges of centralized schemes and poor solutions of traditional distributed schemes. This is done by creating multiple levels of aggressive load balancing domains which form a tree. This hierarchical method is demonstrated within a measurementbased load balancing framework in Charm++. We present techniques to deal with scalability challenges of load balancing at very large scale. We show performance data of the hierarchical load balancing method on up to 16,384 cores of Ranger (at TACC) for a synthetic benchmark. We also demonstrate the successful deployment of the method in a scientific application, NAMD with results on the Blue Gene/P machine at ANL.

Original languageEnglish (US)
Title of host publicationProceedings - 2010 39th International Conference on Parallel Processing Workshops, ICPPW 2010
Pages436-444
Number of pages9
DOIs
StatePublished - Dec 13 2010
Event2010 39th International Conference on Parallel Processing Workshops, ICPPW 2010 - San Diego, CA, United States
Duration: Sep 13 2010Sep 16 2010

Publication series

NameProceedings of the International Conference on Parallel Processing Workshops
ISSN (Print)1530-2016

Other

Other2010 39th International Conference on Parallel Processing Workshops, ICPPW 2010
CountryUnited States
CitySan Diego, CA
Period9/13/109/16/10

    Fingerprint

ASJC Scopus subject areas

  • Software
  • Mathematics(all)
  • Hardware and Architecture

Cite this

Zheng, G., Meneses, E., Bhatelé, A., & Kale, L. V. (2010). Hierarchical load balancing for Charm++ applications on large supercomputers. In Proceedings - 2010 39th International Conference on Parallel Processing Workshops, ICPPW 2010 (pp. 436-444). [5599103] (Proceedings of the International Conference on Parallel Processing Workshops). https://doi.org/10.1109/ICPPW.2010.65