Abstract
We present experimental results for parallelizing two breadth-first search-based applications on Thinking Machines CM-5 by using two different message-passing paradigms, one based onsend/receive and the other based onactive messages. The parallelization of these applications requires fine-grained communication. Our results show that the active messages-based implementation gives significant improvement over the send/receive-based implementation. The improvements can largely be attributed to the lower latency of the active messages implementation.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
C. Aykanat and T. Kurc. Efficient parallel maze routing algorithm on a hypercube multicomputer. In Proc., 1991 Internat. Conf. on Parallel Processing, vol. 3, pages 224–227, 1991.
Seungjo Bae, Sung-Hoon Ko, and Paul Coddington. Parallel Wolff cluster algorithm.International Journal of Modern Physics C, 6(2): 197–210, 1995.
Seungjo Bae and Sanjay Ranka. A comparison of different message-passing paradigms for the parallelization of two breadth-first search-based applications. Technical Report CRPC-TR94505, The Center for Research on Parallel Computation, 1994.
Clive F. Baillie and Paul D. Coddington. Cluster identification algorithms for spin models—Sequential and parallel. Technical ReportC 3P-855, California Institute of Technology, June 1990.
G. Brassard and P. Bratley.Algorithmics: Theory and Practice. Prentice=Hall, 1987.
E. A. Brewer and R. D. Blumofe. Strata: A multi-layer communication library. Technical Report (to appear), MIT Laboratory for Computer Science, February 1994.
E. A. Brewer and B. C. Kuszmaul. How to get good performance from the CM-5 data network. In Proc. of Internat. Parallel Processing Symposium, pages 858–867, 1994.
F. T. Chong, S. D. Sharma, E. A. Brewer, and J. Saltz. Multiprocessor runtime support for fine-grained, irregular DAGs. To appear in Toward Teraflop Computing and New Grand Challenge Applications. Eds. R. K. Kalia and P. Vashishta, Nova Science Publishers, Inc., New York, 1995.
Paul D. Coddington. Lecture note. Lecture notes for the course CPS713 at Syracuse University, Spring 1993.
Thinking Machines Corporation.CM-5 Technical Summary, November 1993.
Thinking Machines Corporation.CMMD version 3.0 Reference Manual, May 1993.
Y. Fang, I. Yen, and R. Dubash. Improving the performance of Lee's maze routing algorithm on parallel computers via semi-dynamic mapping strategies. Technical Report CPS-93-35, Michigan State University, December 1993.
D. W. Heermann and A. N. Burki.Parallel Algorithms in Computational Science. Springer-Verlag, 1990.
C. Y. Lee. An algorithm for path connections and its applications.IRE Trans., Electronic Computers, EC-10:346–365, September 1961.
R. H. Swendsen and J. S. Wang. Nonuniversal critical dynamics in Monte Carlo simulation.Phys. Rev. Lett., 58(2):86, 1987.
P. Tamayo, R. C. Brower, and W. Klein. Single-cluster Monte Carlo dynamics for the Ising model.J. Statis. Phys., 58(5): 1083, 1990.
T. von Eicken, D. E. Culler, S. C. Goldstein, and K. E. Schauser. Active messages: A mechanism for integrated communication and computation. In Proc. of ISCA 1992, pages 256–266, 1992.
J. S. Wang and R. H. Swendsen. Cluster Monte Carlo algorithm.Physica A, 167:565, 1990.
U. Wolff. Lattice field theory as a percolation process.Phys. Rev. Lett., 62(15):361, 1988.
Y. Won and S. Sahni. Maze routing on a hypercube multiprocessors computers. In Proc. of Internat. Conf. on Parallel Processing, St. Charles, pages 630–637, August 1987.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bae, S., Ranka, S. A comparison of different message-passing paradigms for the parallelization of two irregular applications. J Supercomput 10, 55–85 (1996). https://doi.org/10.1007/BF00128099
Issue Date:
DOI: https://doi.org/10.1007/BF00128099