{"title":"A more efficient mechanism of FDTD method based on message passing interface","authors":"Xu Zhang, Xiaoming Liu, Shaohua Liu, Junsheng Yu, Ran Ding, Bingran Liu, Rui Xiao","doi":"10.1109/ICMMT.2008.4540489","DOIUrl":null,"url":null,"abstract":"Based on the message passing interface (MPI) implement of parallel FDTD, this paper puts forward a more efficient mechanism, which decreases the time of communication and synchronization so as to enhance the efficiency of parallel FDTD. The efficient mechanism mainly consists of overlapping model and logic layer model. In addition, this paper also analyzes the performance gain in various network environments. Meanwhile the mechanism reduces the interdependence between processes so that expanding parallel FDTD into grid computing (Foster et al., 2003) environment is feasible.","PeriodicalId":315133,"journal":{"name":"2008 International Conference on Microwave and Millimeter Wave Technology","volume":"74 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2008-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2008 International Conference on Microwave and Millimeter Wave Technology","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICMMT.2008.4540489","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 2
Abstract
Based on the message passing interface (MPI) implement of parallel FDTD, this paper puts forward a more efficient mechanism, which decreases the time of communication and synchronization so as to enhance the efficiency of parallel FDTD. The efficient mechanism mainly consists of overlapping model and logic layer model. In addition, this paper also analyzes the performance gain in various network environments. Meanwhile the mechanism reduces the interdependence between processes so that expanding parallel FDTD into grid computing (Foster et al., 2003) environment is feasible.
在并行时域有限差分的消息传递接口(MPI)实现的基础上,提出了一种更有效的机制,减少了通信和同步的时间,从而提高了并行时域有限差分的效率。有效机制主要包括重叠模型和逻辑层模型。此外,本文还分析了各种网络环境下的性能增益。同时,该机制减少了进程之间的相互依赖性,从而将并行FDTD扩展到网格计算(Foster et al., 2003)环境中是可行的。