论文部分内容阅读
多细胞分子动力学算法是分子动力学模拟中普遍使用的并行算法.因为利用不同作用路径的多细胞分子动力学算法的消息传递次数彼此不同,利用作用路径能够优化消息传递次数.优化消息传递次数是一种设计高效并行算法的方法.因此,本文研究十六个不同的作用路径,推导出十六个表示消息传递次数的数学模型,并在高性能通信和负载平衡方面决定十六个模型中最好的.实验结果表明模型的平均正确率是99.1391%,它的一个时间步骤的并行效率比HS算法平均提高了5.16%,处理器数的增加和截断半径的优化提高其并行效率.
The multicellular molecular dynamics algorithm is a commonly used parallel algorithm in molecular dynamics simulation.Multiple-cell molecular dynamics algorithms using different pathways differ from each other in the number of messages being transmitted, and the number of messages can be optimized using the path of action.Optimization of the number of messages transmitted Is a way to design an efficient parallel algorithm.Thus, this paper investigates sixteen different pathways of action, derives sixteen mathematical models that represent the number of messages delivered, and determines sixteen models in terms of high performance communications and load balancing The experimental results show that the average correct rate of the model is 99.1391%, the parallel efficiency of one time step is 5.16% higher than that of HS algorithm, the increase of the number of processors and the optimization of truncation radius improve the parallel efficiency.