检索规则说明:AND代表“并且”;OR代表“或者”;NOT代表“不包含”;(注意必须大写,运算符两边需空一格)
检 索 范 例 :范例一: (K=图书馆学 OR K=情报学) AND A=范并思 范例二:J=计算机应用与软件 AND (U=C++ OR U=Basic) NOT M=Visual
作 者:张亚英 吴乘胜[1] 王建春[1] 顾寒锋 ZHANG Ya-ying;WU Cheng-sheng;WANG Jian-chun;GU Han-feng(China Ship Scientific Research Center,Wuxi 214082,China;National Supercomputing Center in Wuxi,Wuxi 214072,China)
机构地区:[1]中国船舶科学研究中心,江苏无锡214082 [2]国家超级计算无锡中心,江苏无锡214072
出 处:《船舶力学》2022年第9期1255-1267,共13页Journal of Ship Mechanics
基 金:水动力预研项目(50907070201)。
摘 要:异构众核已成为超级计算机架构的发展趋势,而如何有效利用异构超算系统的能力,是当前CFD界面临的一项重大挑战。本文基于神威异构超算平台,采用自主开发代码,开展三维非定常不可压流动的大规模并行数值模拟研究。论文的数值计算采用MPI+Athread多级并行方式,其中在Athread众核并行时,针对神威众核处理器的特点,采用循环融合的方式对SIMPLE算法求解流程进行优化,提升加速效果。首先以Re=10的三维方柱准定常绕流为例,开展并行计算测试,包括MPI并行和MPI+Athread多级并行,结果显示2025万网格125进程MPI并行效率为70%以上,Athread众核并行加速7.9倍。随后,针对Re=250的三维方柱非定常绕流,开展网格数量从384万到2.46亿的大规模并行计算,结果与相关文献符合良好,同时表明网格数量增加能够更好地捕捉流场细节。本文的研究工作展现了神威异构超算平台在非定常不可压缩流动CFD大规模并行计算方面的应用能力。Many-core heterogeneous architecture is increasingly adopted in supercomputers nowadays.Effec⁃tive utilization of the computing capability of heterogeneous architecture supercomputers is a tremendous challenge to CFD.Some research work on massive parallel numerical computation for 3D unsteady incom⁃pressible flow based on Sunway supercomputer was presented in this paper.MPI+Athread was adopted in par⁃allel computation.The SIMPLE algorithm was optimized by way of loop consolidation to enhance the speedup rate in light of the characteristics of Sunway many-core processor.Firstly,numerical simulations of quasisteady flow of Re=10 past 3D square cylinder were carried out for parallel computation testing,including MPI and MPI+Athread parallel computation.The results show that over 70%of MPI parallel efficiency and 7.9 times of many-core speedup are achieved in the case of 20.25 million grid cells on 125 processes.Then,mas⁃sive parallel numerical computations for unsteady flow of Re=250 past 3D square cylinder were carried out with grid cell numbers varying from 3.84 million to 246 million.The computational results agree well with those of relevant references and show that the refined grids can capture more details in the flow field.The re⁃search work in this paper visualizes the capability of utilizing Sunway heterogeneous architecture supercom⁃puter in massive parallel CFD computation for unsteady incompressible flows.
关 键 词:异构超算平台 三维方柱绕流 数值模拟 SIMPLE算法 MPI+Athread并行计算
分 类 号:U661.1[交通运输工程—船舶及航道工程]
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在链接到云南高校图书馆文献保障联盟下载...
云南高校图书馆联盟文献共享服务平台 版权所有©
您的IP:18.117.158.174