南京大学学报（自然科学版）2017，Vol.53Issue(6)：1161-1170,10.DOI:10.13232/j.cnki.jnju.2017.06.018
共享内存式并行离散元程序的设计与测试
Design and test of parallel discrete element method program of shared memory type
摘要
Abstract
With the help of such date structure and changing the neighbor search algorithm based on grid,the lock-free parallel computing of discrete element method was achieved,and an efficient program of discrete element method was written in the C language.As a test case,a rock specimen was simulated under uniaxial compression by using the solid lattice model,and the entire process of shear failure was observed.The parallel test results showed that the speedup ratio basically remained unchanged when the number of particles reached tens of thousands or more.It reflected the phenomenon that the speedup of parallel has nothing to do with the scale of the problems.The parallel efficiency was about 0.85 and the time costs linear increases with the increase of number of particles,when we used four threads.The parallel computing of discrete element method was realized based on procedure-oriented language,which could make full use of existing multi-core resources and improve computing speed greatly and be useful to simulate the complex geotechnical engineering and geological model required more particles.关键词
DEM/OpenMP/数据结构/邻居搜索/固体晶格模型/单轴试验Key words
DEM/OpenMP/data structure/neighbor search/solid lattice model/uniaxial tests分类
信息技术与安全科学引用本文复制引用
李长圣,尹宏伟,刘春,蔡申阳..共享内存式并行离散元程序的设计与测试[J].南京大学学报（自然科学版）,2017,53(6):1161-1170,10.基金项目
国家重点基础研究发展计划(973)项目(2012CB214703),国家自然科学基金(41272227,41572187) （973）
