雷霆, 姚振汉, 王海涛. 快速多极与常规边界元法机群并行计算的比较[J]. 工程力学, 2006, 23(11): 28-32,5.
引用本文: 雷霆, 姚振汉, 王海涛. 快速多极与常规边界元法机群并行计算的比较[J]. 工程力学, 2006, 23(11): 28-32,5.
LEI Ting, YAO Zhen-han, WANG Hai-tao. THE COMPARISON OF PARALLEL COMPUTATION BETWEEN FAST MULTIPOLE AND CONVENTIONAL BEM ON PC CLUSTER[J]. Engineering Mechanics, 2006, 23(11): 28-32,5.
Citation: LEI Ting, YAO Zhen-han, WANG Hai-tao. THE COMPARISON OF PARALLEL COMPUTATION BETWEEN FAST MULTIPOLE AND CONVENTIONAL BEM ON PC CLUSTER[J]. Engineering Mechanics, 2006, 23(11): 28-32,5.

快速多极与常规边界元法机群并行计算的比较

THE COMPARISON OF PARALLEL COMPUTATION BETWEEN FAST MULTIPOLE AND CONVENTIONAL BEM ON PC CLUSTER

  • 摘要: 以三维弹性力学问题为例,对快速多极与常规边界元法机群并行计算进行了比较。其中常规边界元法求解方程采用高斯消去法,通过调用标准并行求解函数库ScaLAPACK实现;快速多极边界元法并行计算程序采用ANSI C++语言、调用MPI并行通信库自行编写。两种程序均运行于同一机群并行环境。数值算例表明,在同样的机群条件下,采用快速多极边界元法可使解题规模有数量级的提高,计算速度明显高于常规边界元法,并行效率也优于常规边界元法。

     

    Abstract: For 3D elasticity problems,the parallel computations based on the fast multipole and the conventional boundary element method(BEM)on PC cluster are compared.The parallel computation of conventional BEM applies Gauss elimination by calling the ScaLAPACK library,while the fast multipole version is implemented by programming with ANSI C++ and MPI(Message Passing Interface).Both approaches are tested on the same PC cluster.Numerical examples show that the scale of solvable problems can be increased significantly for the fast multipole BEM,and that the computational speed is much faster than conventional BEM.Besides,the parallel speedup of the fast multipole BEM is also higher than that of the conventional one.

     

/

返回文章
返回