[1]张恩红,尹海燕.高性能计算机在华南气象行业的应用研究[J].计算机技术与发展,2020,30(12):187-191.[doi:10. 3969 / j. issn. 1673-629X. 2020. 12. 033]
 ZHANG En-hong,YIN Hai-yan.Research on Application of High Performance Computer in Meteorological Industry in South China[J].,2020,30(12):187-191.[doi:10. 3969 / j. issn. 1673-629X. 2020. 12. 033]
点击复制

高性能计算机在华南气象行业的应用研究()
分享到:

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
30
期数:
2020年12期
页码:
187-191
栏目:
应用开发研究
出版日期:
2020-12-10

文章信息/Info

Title:
Research on Application of High Performance Computer in Meteorological Industry in South China
文章编号:
1673-629X(2020)12-0187-05
作者:
张恩红尹海燕
广东省气象探测数据中心,广东 广州 510641
Author(s):
ZHANG En-hongYIN Hai-yan
Guangdong Meteorological Observation Data Center,Guangzhou 510641,China
关键词:
高性能计算机气象应用loadlevel共享存储海量数据
Keywords:
high-performance computersmeteorological applicationsloadlevelshared storagemassive data
分类号:
TP39
DOI:
10. 3969 / j. issn. 1673-629X. 2020. 12. 033
摘要:
为了提高产能,近些年来各行业都在建设高性能计算机系统。 该文提到的高性能计算机系统是指中国气象局在华南区域气象中心建设的 IBM 高性能计算机子系统。 重点阐述了如何优化计算资源和存储资源的配置与作业调度管理系统,使得系统在华南气象数值预报模式计算中提供较高的运行效率、较高的节点使用率。 为了解决海量数据的传输与数据处理,及时准确输出数值预报产品,采用对用户类型的划分、计算节点分组的设计、存储资源独享与共享结合的方法。从系统运行结果来看,取得了显著的成效,提高了存储空间的使用效率,提高了节点的使用率,改善了网络的使用环境。不同类型用户的计算节点可用率提升 157% 至 274% ;存储的总需求量比旧系统少了 100 T,提高了 40% 的使用率,数据传输总量减少 55 T,只占旧方案的 45% 。
Abstract:
In order to increase production capacity,various industries have been building high-performance computer systems in recent years. The high performance computer system mentioned refers to the  IBM high performance computer subsystem built by the China Meteorological Administration in the South China Regional Meteorological Center. We focus on how to optimize the allocation of computing resources and storage resources and the job scheduling and management system,so that the system can provide higher operation efficiency and higher node utilization rate in the calculation of meteorological numerical forecast model in South China. In order to solve the problem of mass data transmission and data processing and output numerical prediction products timely and accurately, the methods of dividing user types,grouping computing nodes,and combining exclusive and shared storage resources are adopted. From the running results of the system,remarkable results have been achieved,the use efficiency of storage space has been improved,the utilization rate of nodes has been improved,and the use environment of the network has been improved. The availability of computing nodes for different types of users has increased by 157% to 274% . The total demand for storage is 100T less than that of the old system,an increase of 40% of the utilization rate,and the total amount of data transmission is reduced by 55T,accounting for only 45% of the old scheme.

相似文献/References:

[1]王俊超,彭涛,冯光柳. 曙光高性能计算机在数值预报模式中的应用[J].计算机技术与发展,2014,24(10):178.
 WANG Jun-chao,PENG Tao,FENG Guang-liu. Application of HPC in Numerical Prediction Model[J].,2014,24(12):178.

更新日期/Last Update: 2020-12-10