MPI collective communication operations on large shared memory systems

Author

Bernaschi, Massimo ; Richelli, Giorgio

Author_Institution

IAC-CNR, Rome, Italy

fYear

2001

fDate

2001

Firstpage

159

Lastpage

164

Abstract

Collective communication performance is critical in a number of MPI applications yet relatively few results are available to assess the performance of MPI implementations specially for shared memory multiprocessors. In this paper we focus on the most widely used primitive, broadcast, and present experimental results for the Sun Enterprise 10000. We compare the performance of the Sun MPI primitives with our implementation based on a quasi-optimal algorithm. Our tests highlight advantages and drawbacks of vendors´ implementations of collective communication primitives and suggest that the choice of the best algorithm may depend on exogenous factors like load balancing among tasks

Keywords

performance evaluation; resource allocation; shared memory systems; MPI collective communication operations; Sun Enterprise 10000; Sun MPI primitives; collective communication performance; large shared memory systems; load balancing; quasi-optimal algorithm; shared memory multiprocessors; Algorithm design and analysis; Code standards; Context modeling; Delay; Large-scale systems; Libraries; Message passing; Peer to peer computing; Sun; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Parallel and Distributed Processing, 2001. Proceedings. Ninth Euromicro Workshop on

Conference_Location

Mantova

Print_ISBN

0-7695-0987-8

Type

conf

DOI

10.1109/EMPDP.2001.905038

Filename

905038