The Community for Technology Leaders
Parallel and Distributed Processing Symposium, International (2005)
Denver, Colorado
Apr. 4, 2005 to Apr. 8, 2005
ISSN: 1530-2075
ISBN: 0-7695-2312-9
pp: 272a
Jelena Pjesivac-Grbovic , University of Tennessee Computer Science Department, Knoxville
George Bosilca , University of Tennessee Computer Science Department, Knoxville
Jack J. Dongarra , University of Tennessee Computer Science Department, Knoxville
Thara Angskun , University of Tennessee Computer Science Department, Knoxville
Edgar Gabriel , University of Tennessee Computer Science Department, Knoxville
Graham E. Fagg , University of Tennessee Computer Science Department, Knoxville
ABSTRACT
Previous studies of application usage show that the performance of collective communications are critical for high-performance computing and are often overlooked when compared to the point-to-point performance. In this paper, we analyze and attempt to improve intra-cluster collective communication in the context of the widely deployed MPI programming paradigm by extending accepted models of point-to-point communication, such as Hockney, LogP/LogGP, and PLogP. The predictions from the models were compared to the experimentally gathered data and our findings were used to optimize the implementation of collective operations in the FT-MPI library.
INDEX TERMS
null
CITATION
Jelena Pjesivac-Grbovic, George Bosilca, Jack J. Dongarra, Thara Angskun, Edgar Gabriel, Graham E. Fagg, "Performance Analysis of MPI Collective Operations", Parallel and Distributed Processing Symposium, International, vol. 16, no. , pp. 272a, 2005, doi:10.1109/IPDPS.2005.335
94 ms
(Ver 3.1 (10032016))