期刊文献+

提升PD效率的粒子对方法及信息传递接口并行方法

Particle pair method for improving peridynamics computation efficiency and its MPI parallelism
下载PDF
导出
摘要 为了避免近场动力方法中粒子间物理信息重复求解消耗的计算成本和占据的额外内存,克服近场动力学低效率计算过程对其工程应用的制约,本文提出一种提升PD计算效率的粒子对方法并建立了该方法的信息传递接口并行方法。利用粒子对方法的信息传递接口并行策略并分析其计算效率的优越性,通过大型的工程应用实例给出了本文方法的工程实用性。研究结果表明:粒子对方法的应用不仅提升了计算效率,也优化了邻域粒子信息储存数组占据的内存和计算时间,粒子对方法可与信息传递接口并行编程相结合,能够进一步提升其计算效率,并且在较多线程并行时,计算效率高于OpenMP并行方法,而在实际计算中,加速效果也会受到不同线程间的通信任务影响。 To avoid the increased computational cost and extra occupied memory caused by repeated calculations of the physical information between particles in the peridynamics method and overcome the constraints of the inefficient calculation process of peridynamics on its engineering application,this paper proposes a particle pair method(PPM)to improve the computational efficiency of PD and establishes its message passing interface(MPI)parallel strategy.First,the MPI parallel strategy on PPM is used to analyze the superiority of its calculation efficiency,and the engineering practicability of the method is illustrated by a large-scale engineering application example.The research results show that the method can improve computational efficiency and optimize the memory and computing time occupied by the neighbored particle information-storing array.PPM can be combined with MPI parallel programming to further improve its computing efficiency.In addition,in the case of more parallel threads,the calculation efficiency is higher than that of the traditional OpenMP parallel method.In actual calculations,the speedup effect is also affected by the communication tasks between different threads.
作者 韩康 张媛 王超 叶礼裕 郭春雨 HAN Kang;ZHANG Yuan;WANG Chao;YE Liyu;GUO Chunyu(College of Shipbuilding Engineering,Harbin Engineering University,Harbin 150001,China;National Key Laboratory on Ship Vibration and Noise,China Ship Scientific Research Center,Wuxi 214082,China)
出处 《哈尔滨工程大学学报》 EI CAS CSCD 北大核心 2024年第8期1451-1459,共9页 Journal of Harbin Engineering University
基金 国家自然科学基金项目(51809055).
关键词 粒子对方法 近场动力学 计算效率 信息传递接口并行 工程应用 数值计算 并行计算 粒子搜索 particle pair method peridynamics computation efficiency MPI parallel engineering application numerical calculation parallel calculation particle search
  • 相关文献

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