文章摘要

文章信息/Info

Title:: Performance analysis of parallel computing under different numbers of grid based on OpenFOAM

作者:: 王冠石1; 陈宇1; 蒋文涛1; 田晓宝1△; 王清晟2; 周志宏1; 1. 四川大学力学科学与工程系，四川省生物力学工程实验室,成都 610065;2. 北京市顺义区体育运动学校,北京 101300

Author(s):: WANG Guanshi1; CHEN Yu1; JIANG Wentao1; TIAN Xiaobao1; WANG Qingsheng2; ZHOU Zhihong1; 1. Department of Mechanical Science and Engineering, College of Architecture and Environment Sichuan Province Biomechanical Engineering Laboratory, Sichuan University,Chengdu 610065,China;2. Beijing Shunyi District Sports School,Beijing 101300, China

Keywords:: High performance computing; Parallel computing; Mesh numbers; OpenFOAM; Performance analysis

摘要:: 当前，高性能计算极大地促进了众多领域的发展，因此，探究高性能计算的潜在的运算能力尤为重要。本研究基于自由、开源的计算流体力学（computational fluid dynamics, CFD）平台OpenFOAM，测试了一个多核计算集群在不同网格数和计算节点数下的并行计算时间，通过分析加速性能和计算效率来探究其潜在的运算能力。结果表明，多核并行计算很难达到理想的线性加速性能，甚至当计算能力达到饱和时，加速性能反而下降。同时，随着网格数量的增加，即计算规模的增大，计算效率也越来越高。本研究对于实现基于OpenFOAM的最佳多核并行计算策略具有一定参考价值。

Abstract:: At present, high-performance computing (HPC) has greatly promoted the development of many fields, so it is very important to explore the computing potential of HPC. Using the free and open source comptutational fluidynamics(CFD) platform, OpenFOAM, we tested the parallel computation time of a multi-core computing cluster under different mesh numbers and computing nodes, and explored its computing potential by analyzing the acceleration performance and efficiency. The results showed that multi-core parallel computing was difficult to achieve the ideal linear acceleration performance, and even when reaching the computing bottleneck, on the contrary, the acceleration performance decreased. At the same time, with the increase of the number of grids, i.e., the increase in the scale of computation, the computational efficiency became higher too. It is helpful for realizing the best strategy of multi-core parallel computing using OpenFOAM.