期刊文献+

非结构网格并行计算预处理方法研究 被引量:5

Research on Pre-processing Methods of Unstructured Grids
下载PDF
导出
摘要 非结构网格预处理方法是非结构网格CFD并行计算的关键技术之一。提出基于缓冲数据结构的快速搜索算法来建立全局网格单元邻接关系图,算法复杂度低,能够显著降低非结构网格预处理的存储需求;在提高核心计算访存命中率方面,提出网格单元重排序算法,该算法能够提高核心计算效率,并通用于各种非结构网格问题。实验结果表明,在用于大网格量的复杂计算区域时该非结构网格预处理技术仍能得到较理想的结果。 The pre-processing methods of unstructured grids are one of the important technologies for unstructured grids CFD parallel computing. The paper supplied a new efficient and robust fast search algorithm to build the relationship graph of the global unstructured cells, which is based on buffer data structure and can be easily implemented with low complexity. And the paper brought forward the reordering algorithm to deal with the out-of-order problem brought by unstructured grids,which can improve the computing efficiency and can be used in all kinds of unstructured grids. Experiment results show that even in the case of complicated areas of large grids number, the prerocessing methods can get good performance.
出处 《计算机科学》 CSCD 北大核心 2012年第3期308-311,共4页 Computer Science
基金 863计划(2010AA012301)资助
关键词 非结构网格 CFD并行计算 预处理 快速搜索 重排序 Unstructured grids,CFD parallel computing,Pre-processing,Fast Search Algorithm,Reordering
  • 相关文献

参考文献4

  • 1Monien B, Preis R, Diekmann R. Quality matching and local improvement for multilevel graph-partitioning[J]. Parallel Computing, 2000,26 (12) : 1609-1634.
  • 2Karypis G, Kumar V. METIS: unstructured graph partitioning and sparse matrix ordering system[R]. Department of Computer Science, University of Minnesota, 1995.
  • 3ANSYS FLUENT [OL]. http://www, ansys, com/Products/ Simulation+ Technology/Fluid+ Dynamics/ANSYS+ FLUENT.
  • 4ANSYS CFX [OL]. http://www, ansys, corn/Products/Simula- tion+ Technology/Fluid+ Dynamics/ANSYS+ CFX.

同被引文献29

  • 1张军,谭俊杰,任登凤.二维含动边界流场的并行计算[J].河海大学学报(自然科学版),2007,35(4):460-463. 被引量:1
  • 2王学德,伍贻兆,夏健.动态负载平衡的二维非结构网格DSMC并行算法研究[J].空气动力学学报,2007,25(3):339-344. 被引量:3
  • 3Chen Z,Burke M P,Ju Y.Effects of compression and stretchon the determination of laminar flame speeds using propa-gating spherical flames[J].Combustion Theory and Modelling,2009,13(2):343-364.
  • 4Sun,Takayama.Conservative smoothing on an adaptive quad-rilateral grid[J]Journal of Computational Physics,1999,150(1):143-180.
  • 5Hummel S F,Schonberg E,Flynn L E.Factoring:a methodfor scheduling parallel IoopsfJ].Comm of the ACM,1992,35(8):90-101.
  • 6Kumar V,Grama A Y,RaoVempaty N.Scalable load balancingtechniques for parallel computers[J].J of Parallel and Distr,1994,22(1):60-79.
  • 7Lain M S,Rothberg E E,Wolf M E.The cache performanceand optimizations of block algorithms[C]//4th Int Conf onArchitectural Support for Progr Lang and Operating Systems,1991:63-74.
  • 8Orlando S,Perego R.A template for non-uniform parallelloops based on dynamic scheduling and prefetching tech-niques[C]//Proc of the 1996 ACM Int Conf on Supercom-puting,1996:117-124.
  • 9Orlando S,Perego H.SUPPLE:an efficient run-time supportfor non-uniform parallel loops[J].J of System Architecture,1999,45(15):1323-1343.
  • 10Oliker L,Biswas R.PLUM:Parallel Load balancing for adap-tive Unstructured Meshes[J].Parallel and Distributed Comput-ing,1998,52(2):150-177.

引证文献5

二级引证文献12

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部