一种基于大同步并行编程模式的N体问题的优化实现

An Optimized Implementation of N-Body Problem Based on Maximal Synchronous Parallel Programming Model

文章基于多核机群系统对并行编程模型进行了深入研究,实现了多层次并行体系结构的Open MP/MPI混合编程模型的设计。在以SMP机群系统为背景的情况下,实现其节点间和节点内的分层,运用多层次的并行编程模型进行实验与分析。同时对多层次并行编程模型的性能进行深入的研究,提出了一种大同步混合设计新思路。设计了N-Body问题的大同步优化并行算法,并在曙光TC 5000A机群上与传统的并行算法作了性能方面的比较。通过理论研究并结合大量的实验分析统计,得到了多核机群的混合并行编程模型的性能优化的诸多结论。

This paper discusses hybrid programming paradigm and different implementation for the multi-core cluster, and designs an Open MP/MPI hybrid programming model. Considering the SMP cluster system, we design a hierarchical inter-nodes and hierarchical intra-nodes, and conduct a lot of experiments and analysis. The performance of parallel programming model for multi levels is studied, and a new hybrid design idea of maximal synchronous method is proposed. We design a maximal synchronous optimization parallel algorithm on the N-Body problem, and compare its performance with traditional hybrid parallel algorithms on the Dawning 5000 A cluster. Through theoretical research combined with statistical analysis on the basis of large number of experiments, we obtain many conclusions of performance optimization on multi-core cluster hierarchical parallel programming model.