首页> 外文会议>International Symposium on Flood Forecasting and Water Resources Assessment for IAHS-PUB >The Design and Realization of the Parallel Computing Platform for Digital Watershed Management
【24h】

The Design and Realization of the Parallel Computing Platform for Digital Watershed Management

机译:数字流域管理并行计算平台的设计与实现

获取原文

摘要

Large-scale watershed systems are demanding in computing speed, so we adopt parallel computing method into it and build up a Parallel Operation Unified Platform (POUP) to control the computing process. POUP has a hierarchical structure with four layers: model layer, computing layer, control layer, and application layer from bottom to top. For the convenience of parallel programming, we also develop a Binary Tree River Encoding Method (BTREM), which transfers all streams in a watershed into a binary tree and use non-fixed-length binary code to indicate them, POUP is implemented with MPICH2 in the master-slave pattern while the massive data are stored in an Oracle database on the server. The result of our experiments on the simulation of sediment-laden and coarse sand producing region of the Yellow River, shows that POUP facilitates the control of parallel computing process, raises the computing speed and gives good results that coincides with the actual data. However, mainly because of the restriction in the read-write capability of our database, the parallel efficiency of POUP decays when the number of computing nodes adds up to a certain threshold.
机译:大规模的流域系统对计算速度要求苛刻,因此我们采用并行计算方法进入它并建立并行操作统一平台(POUP)来控制计算过程。 POUP具有具有四层的层次结构:模型层,计算层,控制层和从底部到顶部的应用层。为了便于并行编程,我们还开发了一个二进制树河河编码方法(BTREM),它将流域中的所有流传输到二叉树中,并使用非固定长度的二进制代码来指示它们,POUP使用MPICH2实现大规模数据存储在服务器上的Oracle数据库中时主从模式。我们对黄河的沉积物和粗砂区域模拟的实验结果表明,POOP促进了并行计算过程的控制,提高了计算速度并提供了与实际数据一致的良好结果。然而,主要是因为在数据库的读写能力中的限制,当计算节点的数量增加到特定阈值时,POUP衰减的并行效率。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号