基于GPU的并行非结构网格生成技术研究
详细信息 本馆镜像全文    |  推荐本文 | | 获取馆网全文
摘要
为了解决非结构网格生成在时间和内存上的问题,研究了非结构网格的并行生成方法,提出了一种基于CUDA架构的GPU并行非结构网格生成技术。该技术结合了GPU的高速并行性和并行Delaunay网格生成技术的优点,在CUDA编程框架下,将非结构网格生成的技术应用到GPU并行环境中。通过分析此方法的加速比和效率,对其性能进行了评估。实验结果表明,所提出的方法具备有高效性,与传统方法相比,在保证网格质量的同时,大幅度减少了其时间消耗。
In order to solve the problems of unstructured mesh generation technology in time and memory,the parallel generation method of unstructured grid is researched,and the GPU unstructured mesh generation technology based on the framework of CUDA is put forward.In CUDA programming framework,unstructured mesh generation technology is applied to GPU parallel environment,Combining the high-speed parallel GPU with parallel delaunay generation technology.Its performance is evaluated by the analysis of the speedup rate and efficiency.According to the experimental results,the suggested method is of high efficiency.Compared with traditional methods,it greatly reduces the time consumption in the same mesh quality.
引文
[1]Yasushi I,Alan M.S,Anil K..E,et al.Parallel unstructured mesh generationbyanadvancingfront method[J].MathematicsandComputersinSimulation,2007,75(5-6):200-209.
    [2]Chrisochoides N,NAVE D.Parallel Delaunay mesh generation kerne[l J].International Journal for Numerical Methods in Engineering,2003,58(2):161-176.
    [3]OWEN S J.A survey of unstructured mesh generation technology[C]//Proceedings of the 7th International Meshing Roundtable.Dearborn,Michigan,USA:[s.n.],1998:239-267.
    [4]R.Lhner,J.Camberos.MarshalMerriam.Parallelunstructuredgridgeneration[J].Computer Methods in Applied Mechanics and Engineering,1992,95(3):343-357.
    [5]T.Okusanya,J.Peraire.Parallel unstructuredmeshgeneration[C]//Procee-dings of the 5th Internation Conforence on Numerical Grid Generation inComputational Fluid Dynamics and Related Fields,Mississippi StateUniversity,MS,USA,1996:719-729.
    [6]Harish P,Narayanan P J.Accelerating large graph algrithms on the GPUusing CUDA[J].Computer Science,2007(4873):197-208.
    [7]陈建军.非结构化网格生成及其并行化的若干问题研究[D].杭州:浙江大学,2006.(Chen Jian-jun.Unstructured grid generation and some problems of itsparallelization[D].Hangzhou:Zhejiang University,2006).
    [8]Chrisochoides N.Parallel mesh generation[M]//Bruaset A M,Tveito A.Numerical solution of partial differential equations on parallel computers.Heidelberg:Springer,2006:237-264.
    [9]张舒,褚艳利.GPU高性能运算之CUDA[M].北京:中国水利水电出版社,2009.(Zhang Shu,Chu Yan-li.GPU High Performance Operation CUDA[M].Beijing:China Water Conservancy and Hydropower Press,2009).
    [10]Garland M,Grand S L,Nickolls J,et al.Parallel computing experienceswith CUDA[J].Micro,IEEE,2008,28(4):13-27.
    [11]Hawick K A,Leist A,Playne D P.Parallel graph component labellingwith GPUs and CUDA[J].Parallel Computing,2010,36(12):655-678.
    [12]吴恩华,柳有权.基于图形处理器(GPU)的通用计算[J].计算机辅助设计与图形学学报,2004,16(5):601-612.(Wu En-hua,Liu You-quan.The general calculation based on thegraphics processor(GPU)[J].Computer Aided Design And GraphicsJournal,2004,16(5):601-612).
    [13]吴焰斌.CUDA编程模型[J].科技风,2009(3):63-64.(Wu Yan-bin.CUDA programming model[J].Science and TechnologyWind,2009(3):63-64).
    [14]NVIDIA Corporation,NVIDIA C 3.1,NVIDIA CUDA C ProgrammingGuid Version.1[S].2010.
    [15]张广智,陈雷.基于GPU的地震属性提取[J].物探化探计算技术,2011,33(4):358-364.(ZhangGuang-zhi,ChenLei.Seismicattributeextractionbased on GPU[J].Journal of Geophysical and Geochemical Exploration ComputingTechnology,2011,33(4):358-364.)

版权所有:© 2023 中国地质图书馆 中国地质调查局地学文献中心