O. Terzo, P. Ruiu, L. Mossucca, G. Caragnano, M. Francavilla, F. Vipiana
{"title":"Low-Cost High Performance Infrastructure for Domain-Decomposition Methods in Computational Electromagnetics","authors":"O. Terzo, P. Ruiu, L. Mossucca, G. Caragnano, M. Francavilla, F. Vipiana","doi":"10.1109/3PGCIC.2011.24","DOIUrl":null,"url":null,"abstract":"Computational Electromagnetics (CEM) is a technique used to simulate the interaction of electromagnetic fields with structures in a particular environment. The geometries to analyze are typically very complexes and they could be simulated with a CAD model, that can include a fine level of details. This is reflected in a higher computational effort that must be faced by high performances infrastructures. The paper addresses the potential benefit of adopting a grid computing infrastructure to solve CEM problems. In particular here we focus on a Domain Decomposition (DD) technique to reduce the complexity of a MoM analysis of large and complex structures. It was build a low-cost grid infrastructure, composed of heterogeneous recovery machines, both physical and virtual, on which to run scientific applications without the need to modify the source codes of the used algorithms. Virtualization enables the optimization of resources and simplify infrastructure management. Were carried out some performance tests to verify the time gain and the overhead introduced by the grid infrastructure.","PeriodicalId":251730,"journal":{"name":"2011 International Conference on P2P, Parallel, Grid, Cloud and Internet Computing","volume":"9 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2011-10-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2011 International Conference on P2P, Parallel, Grid, Cloud and Internet Computing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/3PGCIC.2011.24","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 3
Abstract
Computational Electromagnetics (CEM) is a technique used to simulate the interaction of electromagnetic fields with structures in a particular environment. The geometries to analyze are typically very complexes and they could be simulated with a CAD model, that can include a fine level of details. This is reflected in a higher computational effort that must be faced by high performances infrastructures. The paper addresses the potential benefit of adopting a grid computing infrastructure to solve CEM problems. In particular here we focus on a Domain Decomposition (DD) technique to reduce the complexity of a MoM analysis of large and complex structures. It was build a low-cost grid infrastructure, composed of heterogeneous recovery machines, both physical and virtual, on which to run scientific applications without the need to modify the source codes of the used algorithms. Virtualization enables the optimization of resources and simplify infrastructure management. Were carried out some performance tests to verify the time gain and the overhead introduced by the grid infrastructure.