Otimização das Operações Coletivas para um Aglomerado de 8 PCs usando uma Rede Ethernet 10 Mbs baseada em Hub
Resumo
This paper presents the optimization done on collective communication operations of message passing library MPICH version 1.1. The optimization efficiency was demonstrated evaluating the execution time of collective operations and measuring their impact on the total execution time of several application programs. We also compared our results with literature dates. We concluded implementing efficient collective communication operations on cluster based of Ethernet hub provides high performance gain.
Palavras-chave:
Collective communication operations, MPI, performance evaluation
Referências
BALA, V. et al. CCL: a portable and tunable collective communication library for scalable parallel computers. Proceedings of 8th International Parallel Processing Symposium IEEE 1994, p.835-44.
BROOKS, E. D. The butterfly barrier. International Journal of Parallel Programming, v. 15, n.4, p.295-307, 1986.
GROPP, W. et. al. A high-performance implementation of the MPI message passing interface standard. Parallel Computing. v. 22, p. 789-828, 1996.
HUANG. Ch. McKINLEY, P.K. Design and implementations of global reduction operations across ATM networks. Proceedings of Third international symposium of high-performance distributed computing. San francisco, california, 1994, p.43-50.
HUANG, Y. McKINLEY, P. K. Efficient Collective Operations with ATM network interface supon. Procedings of the 1996 International conference on parallel processing. Bloomingdale, Illinois, USA. August 1996, Vol. I, pp. 34-43.
HENGSEN, D. et al. Two algorithms for barrier synchronization. International Journal of Parallel Programming, v.17, n.1, p. 1-17, 1988.
LAURIA, M., CHIEN, A. MPI-FM: high performance MPI on workstation clusters. Journal of Parallel and Distributed Computing, v.40, n.S, p. 431-52, Jan. 1997.
SNIR, M. et al. MPI: the complete reference. London: The MIT Press, 1996
SUNDERAM, V.S. et al. The PVM concurrent computing systems: Evolution, experiences, and trends. Parallel Computing. v.20, n.4, p. 531-545, april 1994.
TANAKA, Y. A comparison of collective communication performance on parallel algorithms. 1996. Real World Computing Partnership parallel and distributed system performance laboratory.
WANG, X. BLUM E. K. Parallel execution of iterative computations on workstation clusters. Journal of Parallel and Distributed Computing, v.34, p.218-226, 1996.
WANG, X. et al. The dance party problem and its application to collective communication in computer networks. Parallel Computing, 23, p. 1141 - 1156, 1997.
BROOKS, E. D. The butterfly barrier. International Journal of Parallel Programming, v. 15, n.4, p.295-307, 1986.
GROPP, W. et. al. A high-performance implementation of the MPI message passing interface standard. Parallel Computing. v. 22, p. 789-828, 1996.
HUANG. Ch. McKINLEY, P.K. Design and implementations of global reduction operations across ATM networks. Proceedings of Third international symposium of high-performance distributed computing. San francisco, california, 1994, p.43-50.
HUANG, Y. McKINLEY, P. K. Efficient Collective Operations with ATM network interface supon. Procedings of the 1996 International conference on parallel processing. Bloomingdale, Illinois, USA. August 1996, Vol. I, pp. 34-43.
HENGSEN, D. et al. Two algorithms for barrier synchronization. International Journal of Parallel Programming, v.17, n.1, p. 1-17, 1988.
LAURIA, M., CHIEN, A. MPI-FM: high performance MPI on workstation clusters. Journal of Parallel and Distributed Computing, v.40, n.S, p. 431-52, Jan. 1997.
SNIR, M. et al. MPI: the complete reference. London: The MIT Press, 1996
SUNDERAM, V.S. et al. The PVM concurrent computing systems: Evolution, experiences, and trends. Parallel Computing. v.20, n.4, p. 531-545, april 1994.
TANAKA, Y. A comparison of collective communication performance on parallel algorithms. 1996. Real World Computing Partnership parallel and distributed system performance laboratory.
WANG, X. BLUM E. K. Parallel execution of iterative computations on workstation clusters. Journal of Parallel and Distributed Computing, v.34, p.218-226, 1996.
WANG, X. et al. The dance party problem and its application to collective communication in computer networks. Parallel Computing, 23, p. 1141 - 1156, 1997.
Publicado
29/09/1999
Como Citar
TORRES, Martha; KOFUJI, Sergio T..
Otimização das Operações Coletivas para um Aglomerado de 8 PCs usando uma Rede Ethernet 10 Mbs baseada em Hub. In: INTERNATIONAL SYMPOSIUM ON COMPUTER ARCHITECTURE AND HIGH PERFORMANCE COMPUTING (SBAC-PAD), 11. , 1999, Natal.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
1999
.
p. 101-106.
DOI: https://doi.org/10.5753/sbac-pad.1999.19777.
