首页 | 官方网站   微博 | 高级检索  
     

基于MPI的FDTD并行算法及其优化策略
引用本文:徐藻,王毅,李琳,曹群生.基于MPI的FDTD并行算法及其优化策略[J].计算机仿真,2009,26(3).
作者姓名:徐藻  王毅  李琳  曹群生
作者单位:南京航空航天大学,江苏,南京,210016
摘    要:由高性能PC机通过网络互联构成的集群(COW)并行计算系统上应用基于消息传递(Message Passing)的方式实现FDTD的并行算法,获得了足够的加速比,有效地解决了传统的FDTD方法计算电大尺寸目标电磁散射问题时的不足.通过区域分割,各个子区域在边界处与其相邻的子区域进行场值的数据传递,从而实现了FDTD算法的并行化.利用并行FDTD方法研究了电磁波的介质层的散射,结果表明并行算法和串行计算结果的一致性,并有效提高计算效率.最后还给出了对算法进行通信隐藏的优化方法,进一步提高了并行计算的效率.

关 键 词:时域有限差分  区域分割  集群  并行计算

A FDTD Parallel Algorithm Based on MPI and Its Optimization Strategy
XU Zao,WANG Yi,LI Lin,CAO Qun-sheng.A FDTD Parallel Algorithm Based on MPI and Its Optimization Strategy[J].Computer Simulation,2009,26(3).
Authors:XU Zao  WANG Yi  LI Lin  CAO Qun-sheng
Affiliation:Nanjing University of Aeronautics and Astronautics;Nanjing Jiangsu 210016;China
Abstract:The high-performance PC constitute a cluster(COW) by application of parallel computing systems based on message passing through the network to achieve the FDTD parallel algorithms,which can gain sufficient speedup,overcome the deficiencies with the traditional FDTD method in resolving the problems of the goal's electromagnetic scattering in the electric-large size effectively.Divided by sub-regions,in all sub-regions of the borders,the value of the data of the field information has been communicated to achi...
Keywords:FDTD  Domain decomposition  Cluster  Parallel computing  
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号