{"title":"Subdomain Mapping Approach to Enhance the Coupling in Earth System Modeling","authors":"Yingsheng Ji, Guangwen Yang, Li Liu, Shu Wang","doi":"10.1109/IPDPSW.2013.62","DOIUrl":null,"url":null,"abstract":"To alleviate the technical barrier for different scientific communities to join in Earth System Modeling, the coupler is widely used to link two or more climate simulation applications called models. With the advent of advanced models, the bigger data volume for transfer and transformation will incur great performance overhead to the coupler. However, the current independent modular design cannot obtain the optimal coupling performance. In this paper, we propose a method called sub domain mapping approach to improve coupling. Our method can merge all the communications during one coupling execution and thus reduce the dispensable cost. The evaluation results show that the sub domain mapping scheme can achieve considerable speedup ranged from 1.14 to 4.81 fold on a cluster interconnected via high-speed Infiniband. The results also show that our approach can effectively enhance the coupler's performance and scalability in most cases.","PeriodicalId":234552,"journal":{"name":"2013 IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum","volume":"EC-1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2013-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2013 IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IPDPSW.2013.62","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
To alleviate the technical barrier for different scientific communities to join in Earth System Modeling, the coupler is widely used to link two or more climate simulation applications called models. With the advent of advanced models, the bigger data volume for transfer and transformation will incur great performance overhead to the coupler. However, the current independent modular design cannot obtain the optimal coupling performance. In this paper, we propose a method called sub domain mapping approach to improve coupling. Our method can merge all the communications during one coupling execution and thus reduce the dispensable cost. The evaluation results show that the sub domain mapping scheme can achieve considerable speedup ranged from 1.14 to 4.81 fold on a cluster interconnected via high-speed Infiniband. The results also show that our approach can effectively enhance the coupler's performance and scalability in most cases.