地下水流动三维有限差分模型并行计算  被引量:3

Parallel Computation of Ground Water Flow for 3D Finite Difference Analysis

在线阅读下载全文

作  者:杨多兴[1,2] 李国敏[2] 董艳辉[2] 黎明[2] 

机构地区:[1]中国地震局地壳应力研究所,北京100085 [2]中国科学院地质与地球物理研究所,北京100029

出  处:《西南大学学报(自然科学版)》2009年第10期149-153,共5页Journal of Southwest University(Natural Science Edition)

基  金:国家自然科学基金资助项目(40672170);中国地震局地壳应力研究所基本科研业务专项资助项目(ZDJ2007-06)

摘  要:基于高速内联网落InfiniBand构建(IBA),采用分布式虚拟共亨内存(DVSM),利用Linux PC机群系统,根据地下水流动的有限差分模型MODFLOW,通过并行计算来模拟三维复杂几何形状的非均质各向异性地下孔隙介质含水系统的水流问题.对于给定的边值问题,计算出不同计算节点的加速比和效率.模拟结果显示,IBA可以通过远程直接内存访问(RDMA)和在硬件中自动运行的方式支持共亨内存进程,在DVSM系统上使计算节点与管理节点之间的通讯时间消耗最小.通过IBA,DVSM能够在很大程度上提高MODFLOW的并行计算效率.结果表明:对于1 000万有限差分计算单元,4,8,16,32,64个计算节点分别参与计算,并行效率分别为32.75%,59.63%,63.63%,58.88%,55.23%.实验证实了基于Linux PC机群系统的MODFLOW并行计算的有效性.In this paper, the authors present the distributed virtual shared-memory (DVSM) on the interconnection technique, the InfiniBand Architecture (IBA). On the Linux PC cluster system, by using the finite difference code called MODLOW, parallel computation is applied to simulate the problem of ground water flow in heterogeneous and anisotropic porous media restricted by the complex geometries. For solving the specific boundary value problem, the speed-up ratios of various nodes are computed and their efficiency is analyzed. The results show that the IBA supports shared-memory programming semantics by means of remote direct-memory access (RDMA) and automatic operations in the hardware. This can minimize the communication overhead for memory consistency on the DVSM system. The IBA-hased DVSM can significantly improve the performance of MODFLOW parallel computation. For 10 000 thousands of finite difference elemems, from 4, 8, 16, 32 and 64 processors on Linux PC cluster, the parallel efficiency is 32.75%, 59. 63%, 63.63%, 58.88% and 55.23%, respectively. Our Linux PC cluster experiments prove the effectiveness of MODFLOW parallel computation.

关 键 词:并行计算 INFINIBAND 分布式虚拟共享内存 加速比 MODFLOW Linux PC机群 

分 类 号:X169[环境科学与工程—环境科学]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象