S. R. Reddy, J. Sebastian, S. M. Miyyadad, R. Banerjee, N. Sivadasan
{"title":"基于GPU的单、两相流CFD求解","authors":"S. R. Reddy, J. Sebastian, S. M. Miyyadad, R. Banerjee, N. Sivadasan","doi":"10.1109/PARCOMPTECH.2013.6621398","DOIUrl":null,"url":null,"abstract":"We present parallelization of single and two phase flow CFD solvers on a graphics processing unit (GPU) platform. Numerical simulations are done for some standard benchmark test cases for both single and two phase flow solvers. The formulation is based on finite volume method with SMAC algorithm on regular cartesian and collocated grid. Volume of fluid method in used for tracking the interface in multiphase flow solver. Pressure poisson equation is the most time consuming part of the solvers and hence this part is imported on to GPU. Pressure Poisson equation is solved by the conventional Gauss siedel method. Present day modern graphics hardware has several hundred cores which can be effectively used by CFD solvers to parallelize the computation. The results are validated against the reference solutions of the teat cases. A comparison is done between CPU and GPU simulations to estimate the computational acceleration and accuracy obtained.","PeriodicalId":344858,"journal":{"name":"2013 National Conference on Parallel Computing Technologies (PARCOMPTECH)","volume":"82 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2013-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Single and two phase flow CFD solvers using GPU\",\"authors\":\"S. R. Reddy, J. Sebastian, S. M. Miyyadad, R. Banerjee, N. Sivadasan\",\"doi\":\"10.1109/PARCOMPTECH.2013.6621398\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"We present parallelization of single and two phase flow CFD solvers on a graphics processing unit (GPU) platform. Numerical simulations are done for some standard benchmark test cases for both single and two phase flow solvers. The formulation is based on finite volume method with SMAC algorithm on regular cartesian and collocated grid. Volume of fluid method in used for tracking the interface in multiphase flow solver. Pressure poisson equation is the most time consuming part of the solvers and hence this part is imported on to GPU. Pressure Poisson equation is solved by the conventional Gauss siedel method. Present day modern graphics hardware has several hundred cores which can be effectively used by CFD solvers to parallelize the computation. The results are validated against the reference solutions of the teat cases. A comparison is done between CPU and GPU simulations to estimate the computational acceleration and accuracy obtained.\",\"PeriodicalId\":344858,\"journal\":{\"name\":\"2013 National Conference on Parallel Computing Technologies (PARCOMPTECH)\",\"volume\":\"82 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2013-10-08\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2013 National Conference on Parallel Computing Technologies (PARCOMPTECH)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/PARCOMPTECH.2013.6621398\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2013 National Conference on Parallel Computing Technologies (PARCOMPTECH)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/PARCOMPTECH.2013.6621398","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
We present parallelization of single and two phase flow CFD solvers on a graphics processing unit (GPU) platform. Numerical simulations are done for some standard benchmark test cases for both single and two phase flow solvers. The formulation is based on finite volume method with SMAC algorithm on regular cartesian and collocated grid. Volume of fluid method in used for tracking the interface in multiphase flow solver. Pressure poisson equation is the most time consuming part of the solvers and hence this part is imported on to GPU. Pressure Poisson equation is solved by the conventional Gauss siedel method. Present day modern graphics hardware has several hundred cores which can be effectively used by CFD solvers to parallelize the computation. The results are validated against the reference solutions of the teat cases. A comparison is done between CPU and GPU simulations to estimate the computational acceleration and accuracy obtained.