左一鸣, 崔广柏. 二维水动力模型的并行计算研究[J]. 水科学进展, 2008, 19(6): 846-850.
引用本文: 左一鸣, 崔广柏. 二维水动力模型的并行计算研究[J]. 水科学进展, 2008, 19(6): 846-850.
ZUO Yi-ming, CUI Guang-bai. Parallel computation for 2D flow model[J]. Advances in Water Science, 2008, 19(6): 846-850.
Citation: ZUO Yi-ming, CUI Guang-bai. Parallel computation for 2D flow model[J]. Advances in Water Science, 2008, 19(6): 846-850.

二维水动力模型的并行计算研究

Parallel computation for 2D flow model

  • 摘要: 建立了二维水动力并行模型;针对MPI不能实现进程迁移现状,自主开发了并行通讯平台,平台机群负载平衡采用基于蚁群算法的人工智能算法,并根据模型需要制定了相关通讯协议;对长江内江段进行数值模拟,结果表明当网格数一定时,存在一最优客户端数,当客户端数小于最优客户端数时,并行算法所需时间小于串行算法时间,并随着客户端数增加,所需时间也逐渐减少;反之,所需时间则逐渐增大。

     

    Abstract: This article develops a parallel depth-averaged 2D flow model.Since the message passing interface has not implemented the process transference function,the article researches message parallel platform for the model.The platform adopts ant algorithm to schedule tasks and constitutes message protocol to pass message.After its applying it to changjiang river of neijiang,the results indicate that if machines count is less than the oriental count,the consumed time increases along with the adding machine count and vice versa.

     

/

返回文章
返回