数理化科学

多层网格LBM并行算法的静态负载平衡方法

  • 逄宗扬 ,
  • 丁虎 ,
  • 张武
展开
  • 1. 上海大学 力学与工程科学学院, 上海 200444;
    2. 上海大学 上海市应用数学与力学研究所, 上海 200072

收稿日期: 2020-05-25

  网络出版日期: 2020-09-16

基金资助

国家自然科学基金重大研究计划重点项目(91630206);上海市教委创新重大项目(2019-01-07-00-09-E00018)

Static load balance method for lattice Boltzmann method of multi-layer grid

  • PANG Zongyang ,
  • DING Hu ,
  • ZHANG Wu
Expand
  • 1. School of Mechanics and Engineering Science, Shanghai University, Shanghai 200444, China;
    2. Shanghai Institute of Applied Mathematics and Mechanics, Shanghai University, Shanghai 200072, China

Received date: 2020-05-25

  Online published: 2020-09-16

摘要

负载平衡是提升大规模并行计算性能的重要因素.针对格子Boltzmann方法提出一种一维分割下的多层网格静态负载平衡方法,该方法包括基于时间负载的计算域分配优化以及通信隐藏的通信优化.理论分析和实验结果表明,本方法对格子Boltzmann方法并行计算的负载平衡有明显改善,可应用于飞机气动噪声流场和声场的大规模精确计算.

本文引用格式

逄宗扬 , 丁虎 , 张武 . 多层网格LBM并行算法的静态负载平衡方法[J]. 上海大学学报(自然科学版), 2025 , 31(6) : 1076 -1086 . DOI: 10.12066/j.issn.1007-2861.2256

Abstract

Load balancing is an important factor to improve the performance of large-scale parallel computing. A static load balancing method of multi-layer grid with one-dimensional partition is proposed for the lattice Boltzmann method. This hybrid method includes the optimization of computing domain allocation based on time load and the optimization of communication hiding. Theoretical analysis and experimental results show that this method can improve the load balance problem of parallel calculation of lattice Boltzmann method, and can be applied to large-scale accurate calculation of aircraft aerodynamic noise flow field and sound field.

参考文献

[1] 何雅玲, 王勇, 李庆. 格子Boltzmann方法的理论及应用[M]. 北京: 科学出版社, 2009.
[2] Karypis G, Kumar V. Parallel multilevel k-way partitioning scheme for irregular graphs [C]// Proceedings of the 1996 ACM/IEEE conference on Supercomputing. 1996: 35.
[3] Lasalle D, Karypis G. Multi-threaded graph partitioning [C]// IEEE 27th International Symposium on Parallel and Distributed Processing. 2013: 225-236.
[4] Devine K D, Boman E G, Heaphy R T, et al. Parallel hypergraph partitioning for scientiflc computing [C]// Proceedings 20th IEEE International Parallel & Distributed Processing Symposium. 2006: 10.
[5] Chevalier C, Pellegrini F. PT-Scotch: a tool for efficient parallel graph ordering [J]. Parallel Computing, 2008, 34(6/7/8): 318-331.
[6] 张娟, 陆林生. 基于多区域多代码问题的自动分块算法[J]. 计算机工程, 2010, 36(9): 73-76.
[7] 李桂波, 杨国伟. 基于多块结构网格的并行计算及负载平衡研究[J]. 宇航学报, 2011, 32(6): 1224-1230.
[8] Berger, Bokhari. A partitioning strategy for nonuniform problems on multiprocessors [J]. IEEE Transactions on Computers, 1987, 36(5): 570-580.
[9] Farhat C, Lesoinne M. Automatic partitioning of unstructured meshes for the parallel solution of problems in computational mechanics [J]. International Journal for Numerical Methods in Engineering, 1993, 36(5): 745-764.
[10] Lockard D, Choudhari M. The variation of slat noise with Mach and Reynolds numbers [C]// 17th AIAA/CEAS Aeroacoustics Conference. 2011: 2910.
[11] 王良军. 基于国产神威超级计算机的格子Boltzmann方法及应用研究[D]. 上海: 上海大学, 2019.
文章导航

/