The evaluation and prediction of parallel programs performance are becoming more and more important, so that they require appropriate techniques to identify the factors which influence the application execution time and also the way they interact. In this paper, we present some contributions of our research in this area by describing PEMPIs, a new methodology applied to the performance analysis and prediction of MPI programs. A new task graph helps us both to understand details of the application and to increase the accuracy of the prediction models. The proposed techniques are detailed and tested through the modeling of a complete application. PEMPIs efficiency has been proved by the results of this application modeling-most tests executed in a cluster of computers showed errors up to 10%.

No abstract

The evaluation and prediction of parallel programs performance are becoming more and more important, so that they require appropriate techniques to identify the factors which influence the application execution time and also the way they interact. In this paper, we present some contributions of our research in this area by describing PEMPIs, a new methodology applied to the performance analysis and prediction of MPI programs. A new task graph helps us both to understand details of the application and to increase the accuracy of the prediction models. The proposed techniques are detailed and tested through the modeling of a complete application. PEMPIs efficiency has been proved by the results of this application modeling-most tests executed in a cluster of computers showed errors up to 10%.

No abstract

Um programa paralelo e distribuído pode ter seu código fonte estruturado de diferentes modos. A organização da divisão e distribuição dos dados é uma atividade crítica para o desempenho final da aplicação. Assim, é importante que exista uma metodologia capaz de auxiliar estudos com o objetivo de comparar diferentes abordagens de soluções e prever qual modelo é o mais adequado para organizar a solução da aplicação. Nesse artigo, demonstramos como a metodologia PEMPIs-Het pode ser utilizada para esse propósito. Os resultados obtidos confirmam a capacidade da metodologia em avaliar e predizer corretamente o desempenho de diferentes estruturas de programas paralelos.