{"title":"Radiation-magnetohydrodynamics of plasmas on parallel supercomputers","authors":"O. Yasar, G. Moses, T. Tautges","doi":"10.1109/FMPC.1992.234915","DOIUrl":null,"url":null,"abstract":"Presents a parallel computational model to simulate plasmas in the radiation-magnetohydrodynamics (R-MHD) framework. The solution of the radiation field usually dominates the R-MHD computation. The authors solve the linear Boltzmann equation for the radiation field intensity, using the deterministic S/sub N/ discrete ordinates method. Choosing an energy-domain decomposition the authors have implemented the S/sub N/ method on a parallel processor, the Intel iPSC/860, and the speedups are very favorable. Increasing almost linearly with the number of processors, the speedup reaches 14 on 16 processors. A comparison of timing measurements between a single processor CRAY Y-MP and a 16 processor iPSC/860 implementation strongly favors parallelism by a factor of 3.7.<<ETX>>","PeriodicalId":117789,"journal":{"name":"[Proceedings 1992] The Fourth Symposium on the Frontiers of Massively Parallel Computation","volume":"41 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1992-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"[Proceedings 1992] The Fourth Symposium on the Frontiers of Massively Parallel Computation","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/FMPC.1992.234915","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
Presents a parallel computational model to simulate plasmas in the radiation-magnetohydrodynamics (R-MHD) framework. The solution of the radiation field usually dominates the R-MHD computation. The authors solve the linear Boltzmann equation for the radiation field intensity, using the deterministic S/sub N/ discrete ordinates method. Choosing an energy-domain decomposition the authors have implemented the S/sub N/ method on a parallel processor, the Intel iPSC/860, and the speedups are very favorable. Increasing almost linearly with the number of processors, the speedup reaches 14 on 16 processors. A comparison of timing measurements between a single processor CRAY Y-MP and a 16 processor iPSC/860 implementation strongly favors parallelism by a factor of 3.7.<>