The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Este artigo examina primeiro os problemas relacionados ao agendamento de aplicações de loop em um sistema de memória compartilhada distribuída por software (DSM). Em seguida, um esquema de escalonamento dinâmico é desenvolvido com base nas questões examinadas para melhorar o desempenho das aplicações de loop no DSM. Comparado com trabalhos anteriores, o esquema proposto possui diversas especialidades. A primeira é que a carga de trabalho dos processadores pode ser efetivamente equilibrada mesmo quando as capacidades computacionais dos processadores e as necessidades computacionais dos threads não são idênticas. A segunda é dividir o mapeamento de threads em duas fases, cada uma com uma consideração, ou seja, equilíbrio de carga ou custo de comunicação, e adotar migração e troca de threads nas duas fases, respectivamente. A terceira é a exploração do compartilhamento de dados entre threads para reduzir a comunicação de consistência de dados, e a última é atacar o efeito negativo do compartilhamento desnecessário entre nós causado pelo remapeamento de threads. O esquema proposto foi implementado em um sistema DSM baseado em páginas chamado Cohesion. Nossos experimentos mostram que o esquema proposto é mais eficaz para melhorar o desempenho dos programas de teste do que esquemas relacionados.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Tyng-Yeu LIANG, Ce-Kuen SHIEH, Deh-Cheng LIU, "Scheduling Loop Applications in Software Distributed Shared Memory Systems" in IEICE TRANSACTIONS on Information,
vol. E83-D, no. 9, pp. 1721-1730, September 2000, doi: .
Abstract: This paper first examines the issues related to scheduling loop applications on a software distributed shared memory (DSM) system. Then, a dynamic scheduling scheme is developed based on the examined issues to enhance the performance of loop applications on DSM. Compared with previous works, the proposed scheme has several specialties. The first is that the workload of processors can be effectively balanced even when the computational capabilities of processors and the computational needs of threads are not identical. The second is it divides thread mapping into two phases, each with one consideration, i.e., load balance or communication cost, and adopts thread migration and exchange in the two phases, respectively. The third is the exploitation of data sharing among threads to reduce data-consistency communication, and the last is to attack the negative effect of the unnecessary inter-node sharing caused by thread re-mapping. The proposed scheme has been implemented on a page-based DSM system called Cohesion. Our experiments show that the proposed scheme is more effective to improve the performance of the test programs than related schemes.
URL: https://global.ieice.org/en_transactions/information/10.1587/e83-d_9_1721/_p
Copiar
@ARTICLE{e83-d_9_1721,
author={Tyng-Yeu LIANG, Ce-Kuen SHIEH, Deh-Cheng LIU, },
journal={IEICE TRANSACTIONS on Information},
title={Scheduling Loop Applications in Software Distributed Shared Memory Systems},
year={2000},
volume={E83-D},
number={9},
pages={1721-1730},
abstract={This paper first examines the issues related to scheduling loop applications on a software distributed shared memory (DSM) system. Then, a dynamic scheduling scheme is developed based on the examined issues to enhance the performance of loop applications on DSM. Compared with previous works, the proposed scheme has several specialties. The first is that the workload of processors can be effectively balanced even when the computational capabilities of processors and the computational needs of threads are not identical. The second is it divides thread mapping into two phases, each with one consideration, i.e., load balance or communication cost, and adopts thread migration and exchange in the two phases, respectively. The third is the exploitation of data sharing among threads to reduce data-consistency communication, and the last is to attack the negative effect of the unnecessary inter-node sharing caused by thread re-mapping. The proposed scheme has been implemented on a page-based DSM system called Cohesion. Our experiments show that the proposed scheme is more effective to improve the performance of the test programs than related schemes.},
keywords={},
doi={},
ISSN={},
month={September},}
Copiar
TY - JOUR
TI - Scheduling Loop Applications in Software Distributed Shared Memory Systems
T2 - IEICE TRANSACTIONS on Information
SP - 1721
EP - 1730
AU - Tyng-Yeu LIANG
AU - Ce-Kuen SHIEH
AU - Deh-Cheng LIU
PY - 2000
DO -
JO - IEICE TRANSACTIONS on Information
SN -
VL - E83-D
IS - 9
JA - IEICE TRANSACTIONS on Information
Y1 - September 2000
AB - This paper first examines the issues related to scheduling loop applications on a software distributed shared memory (DSM) system. Then, a dynamic scheduling scheme is developed based on the examined issues to enhance the performance of loop applications on DSM. Compared with previous works, the proposed scheme has several specialties. The first is that the workload of processors can be effectively balanced even when the computational capabilities of processors and the computational needs of threads are not identical. The second is it divides thread mapping into two phases, each with one consideration, i.e., load balance or communication cost, and adopts thread migration and exchange in the two phases, respectively. The third is the exploitation of data sharing among threads to reduce data-consistency communication, and the last is to attack the negative effect of the unnecessary inter-node sharing caused by thread re-mapping. The proposed scheme has been implemented on a page-based DSM system called Cohesion. Our experiments show that the proposed scheme is more effective to improve the performance of the test programs than related schemes.
ER -