The Application of High Performance Computing Technology in Meteorological Field

Zhao Licheng; Shen Wenhai; Xiao Huadong; Wang Bin; Sun Jing; Wei Min; Li Juan; Shen Yu

doi:10.11898/1001-7313.20160504

Vol.27, NO.5, 2016

Turn off MathJax

Article Contents

Article Navigation > Journal of Applied Meteorological Science > 2016 > 27(5): 550-558

Zhao Licheng, Shen Wenhai, Xiao Huadong, et al. The application of high performance computing technology in meteorological field. J Appl Meteor Sci, 2016, 27(5): 550-558. DOI: 10.11898/1001-7313.20160504.

Citation:

Zhao Licheng, Shen Wenhai, Xiao Huadong, et al. The application of high performance computing technology in meteorological field. J Appl Meteor Sci, 2016, 27(5): 550-558. DOI: 10.11898/1001-7313.20160504.

Citation:

PDF( 448 KB)

The Application of High Performance Computing Technology in Meteorological Field

DOI: 10.11898/1001-7313.20160504

National Meteorological Information Center, Beijing 100081

Received Date: 2016-06-13
Rev Recd Date: 2016-07-25
Publish Date: 2016-09-30

Abstract

Abstract

High performance computing (HPC), as one of fastest growing branches of information technology, solves complex computational problems via the usage of super-computer and parallel processing. The newest edition of the TOP500 list of the world's top supercomputers shows full entry into PFLOPS era. Most of them adopt cluster structure, with internal nodes interconnected by Infiniband technology. The Intel processor chip and Linux operating system occupy the dominant position. Leading HPC vendors have begun to explore ExaFlops supercomputers.Meteorology is one of the key areas on the application of HPC. Whether for short-term numerical weather forecast systems or for long-term numerical climate prediction systems, it is inseparable from the powerful computing and storage capacity support. Numerical weather prediction services have been timeliness requirement, and HPC technology overcomes efficiently restrictions of high-resolution, high-precision numerical models, and thus plays an essential supporting role. For decades, due to strong demands for HPC resources from numerical weather prediction research and operations, high performance computers and application see a rising requirement and rapid development in CMA and foreign meteorological department. Requirement for computing capabilities and system reliability is rising sharply. 22 high performance computer systems on the TOP500 list are used for the weather and climate fields. The development of meteorological numerical forecast models is characterized by higher resolution, more complicated physical processes, ensemble forecasts and model couplers. HPC capability building is especially important, as well as optimizing processor architecture, operating system, compiler and parallel environment, scientific computing library, etc. HPC resources for meteorological fields are always scarce, therefore, future management will be gradually transitioned from the current local pre-allocation way to the unified scheduling and sharing of local and remote resources. If HPC resources in the operational center of CMA headquarters and regional centers is put into unified management, monitoring, and sharing, it will solve resources integration, sharing and collaborative management, and other problems.HPC technology and weather prediction applications will be even more closely integrated, and expect innovative development in the future. In order to keep pace with numerical weather prediction research and operations, CMA will upgrade HPC capabilities and be committed to optimizing system deployment and managing computing resources to maximize efficiency.
- high performance computing;
- meteorological numerical models;
- supercomputers

FullText(HTML)

References(75)

Figures and Tables

Table 1 Development plan of ECMWF IFS model

IFS分辨率	预期业务化时间	格点间距/km	积分步长/s	估算CPU核数
T2047H	2014—2015年	10	450	6000
T3999NH	2023—2024年	5	249	80000
T7999NH	2031—2032年	2.5	30~120	1~4000000
注：H表示静力平衡，NH表示非静力平衡；估算CPU核数以IBM POWER7芯片为基准，1 h内完成10 d预报需要的CPU核数。

DownLoad: Download CSV

References

[1]	王彬.高性能计算技术在气象部门的应用.计算机工程与设计, 2014, 35(4):1476-1479. http://www.cnki.com.cn/Article/CJFDTOTAL-SJSJ201404067.htm
[2]	洪文董.高性能计算机的发展与气象应用.计算机工程与应用, 2004, 40(5):32-35. http://www.cnki.com.cn/Article/CJFDTOTAL-JSGG200405010.htm
[3]	王文义, 王若雨, 董绍静.高性能科学计算的特征分析及其实用方法研究.计算机科学, 2008, 35(9):217-219. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJA200809060.htm
[4]	王彬, 宗翔, 田浩.国家气象计算网格的设计与建立.应用气象学报, 2010, 21(5):632-640. doi: 10.11898/1001-7313.20100514
[5]	沈文海.从云计算看气象部门未来的信息化趋势.气象科技进展, 2012, 2(2):49-56. http://www.cnki.com.cn/Article/CJFDTOTAL-QXKZ201202017.htm
[6]	Bennett P M.Sustained Systems Performance Monitoring at the US Department of Defense High Performance Computing Modernization Program//State of the Practice Reports.ACM, 2011:3.
[7]	Jordan H, Thoman P, Durillo J J, et al.A Multi-objective Auto-tuning Framework for Parallel Codes//High Performance Computing, Networking, Storage and Analysis (SC).2012 International Conference for IEEE, 2012:1-12.
[8]	Sarood O, Kale L V.A 'cool' load balancer for parallel applications//Proceedings of 2011 International Conference for High Performance Computing, Networking, Storage and Analysis.ACM, 2011:21.
[9]	杨广文, 武永卫, 朱晶.一种全局统一的层次化网格资源模型.计算机研究与发展, 2003, 40(12):1763-1769. http://www.cnki.com.cn/Article/CJFDTOTAL-JFYZ200312018.htm
[10]	廖湘科, 谭郁松, 卢宇彤, 等.面向大数据应用挑战的超级计算机设计.上海大学学报:自然科学版, 2016, 22(1):3-16. http://www.cnki.com.cn/Article/CJFDTOTAL-SDXZ201601002.htm
[11]	Foster I, Zhao Y, Raicu I, et al.Cloud Computing and Grid Computing 360-degree Compared//Grid Computing Environments Workshop, 2008.IEEE, 2008:1-10.
[12]	陈康, 郑纬民.云计算:系统实例与研究现状.软件学报, 2009, 20(5):1337-1348. http://www.cnki.com.cn/Article/CJFDTOTAL-RJXB200905028.htm
[13]	季刚.集群技术及负载均衡调度算法研究.计算机时代, 2012 (8):37-43. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJS201208013.htm
[14]	气候变化应对决策支撑系统工程第二批高性能计算机系统技术及应用调研小组. 高性能计算机技术及应用调研报告. 2016.
[15]	InfiniBandSM Trade Association.InfiniBandTM Architecture Specification Volume 1, Ser Release 1.0.2000.
[16]	谢向辉, 彭龙根, 吴志兵, 等.基于InfiniBand的高性能计算机技术研究.计算机研究与发展, 2005, 42(6):905-912. http://www.cnki.com.cn/Article/CJFDTOTAL-JFYZ200506000.htm
[17]	敖志刚, 解文彬, 胡琨, 等.新一代以太网的结构模型和并行传输设计.电子科技大学学报, 2013, 42(5):773-777. http://www.cnki.com.cn/Article/CJFDTOTAL-DKDX201305026.htm
[18]	王彬.第15届ECMWF高性能计算机在气象中的应用研讨会简介.气象科技合作动态, 2013(2):11-17. http://www.cqvip.com/QK/85781X/201302/45544126.html
[19]	李国杰.21世纪上半叶信息科学技术展望.中国科学院院刊, 2010, 25(1):78-86. http://www.cnki.com.cn/Article/CJFDTOTAL-KYYX201001012.htm
[20]	孙婧, 沈瑜.气象应用的高性能计算机性能需求推算方法.计算机技术与发展, 2015, 25(6):206-210. http://www.cnki.com.cn/Article/CJFDTOTAL-WJFZ201506046.htm
[21]	谢彬, 钟敏, 刘军.高性能计算机群的性能评测与优化.高能量密度物理, 2006(4):170-173. http://cdmd.cnki.com.cn/Article/CDMD-10183-2010032013.htm
[22]	Karlin I, Bhatele A, Keasler J, et al.Exploring Traditional and Emerging Parallel Programming Models Using a Proxy Application//Parallel & Distributed Processing (IPDPS).2013 IEEE 27th International Symposium on IEEE, 2013:919-932.
[23]	Massie M L, Chun B N, Culler D E.The ganglia distributed monitoring system:Design, implementation, and experience.Parallel Computing, 2004, 30(7):817-840. doi: 10.1016/j.parco.2004.04.001
[24]	Katz D S, Hart D, Jordan C, et al.Cyberinfrastructure Usage Modalities on the TeraGrid//2011 IEEE International Parallel & Distributed Processing Symposium Workshops and PhD Forum IEEE.2011:932-939.
[25]	Asanovic K, Bodik R, Catanzaro B C, et al.The Landscape of Parallel Computing Research:A View From Berkeley.Technical Report UCB/EECS-2006-183, EECS Department, University of California, Berkeley, 2006. https://www.eecis.udel.edu/~cavazos/cisc879-spring2008/YingY.pdf
[26]	Ge R, Feng X, Song S, et al.Powerpack:Energy profiling and analysis of high-performance systems and applications.IEEE Transactions on Parallel and Distributed Systems, 2010, 21(5):658-671. doi: 10.1109/TPDS.2009.76
[27]	肖华东, 孙婧, 魏敏, 等.高性能计算机系统相对持续性能度量模型.计算机工程与应用, 2015, 51(5):33-37. http://www.cnki.com.cn/Article/CJFDTOTAL-JSGG201505008.htm
[28]	魏敏, 孙婧, 沈瑜, 等.高性能计算系统性能评测方法及其应用.应用气象学报, 2013, 24(6):753-760. doi: 10.11898/1001-7313.20130612
[29]	林一松, 杨学军, 唐滔, 等.一种基于关键路径分析的CPU-GPU异构系统综合能耗优化方法.计算机学报, 2012, 35(1):123-133. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJX201201012.htm
[30]	郭妙, 金之雁, 周斌.基于通用图形处理器的GRAPES长波辐射并行方案.应用气象学报, 2012, 23(3):348-354. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20120311&flag=1
[31]	吴恩华, 柳有权.基于图形处理器 (GPU) 的通用计算.计算机辅助设计与图形学学报, 2004, 16(5):601-612. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJF200405001.htm
[32]	王海峰, 曹云鹏.GPU集群能耗优化控制模型研究.电子学报, 2015, 43(10):1904-1910. doi: 10.3969/j.issn.0372-2112.2015.10.004
[33]	贺永翔, 刘昕, 赵海波, 等.气体动力学直接模拟Monte Carlo的高效GPU并行计算.计算物理, 2015, 32(2):169-176. http://www.cnki.com.cn/Article/CJFDTOTAL-JSWL201502004.htm
[34]	李佳佳.异构GPU集群的并行编程模型及实现.上海:复旦大学, 2013.
[35]	高飞, 刘轶.基于Intel MIC众核架构的视频字幕提取算法并行加速.计算机工程与科学, 2015, 37(4):634-640. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJK201504003.htm
[36]	沈铂, 张广勇, 吴韶华, 等.基于MIC平台的offload并行方法研究.计算机科学, 2014, 41(增刊Ⅰ):477-480. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJA2014S1118.htm
[37]	吕慧伟, 程元, 白露, 等.众核处理器和众核集群的并行模拟.计算机研究与发展, 2013, 50(5):1110-1117. doi: 10.7544/issn1000-1239.2013.20111270
[38]	郑方, 沈莉, 李宏亮, 等.面向高性能计算的众核处理器轻量级错误恢复技术研究.计算机研究与发展, 2015, 52(6):1316-1328. doi: 10.7544/issn1000-1239.2015.20150119
[39]	Lorenz E N.Reflections on the conception, birth, and childhood of numerical weather prediction.Annu Rev Earth Planet Sci, 2006, 34:37-45. doi: 10.1146/annurev.earth.34.083105.102317
[40]	郑明典.数值天气预报近期的发展趋势.物理双月刊, 2001, 23(3):422-426. http://www.cnki.com.cn/Article/CJFDTOTAL-QXKJ704.000.htm
[41]	Bauer P, Thorpe A, Brunet G.The quiet revolution of numerical weather prediction.Nature, 2015, 525(7567):47-55. doi: 10.1038/nature14956
[42]	魏敏.气象高性能计算应用服务环境适应性研究.气象, 2015, 41(1):92-97. doi: 10.7519/j.issn.1000-0526.2015.01.011
[43]	魏敏, 王彬, 孙婧, 等."天河一号"系列超级计算机系统气象领域适用性分析.气象科技进展, 2012, 2(1):31-35. http://www.cnki.com.cn/Article/CJFDTOTAL-QXKZ201201014.htm
[44]	李黄, 王春虎.气象通信网络和计算机系统工程.中国工程科学, 2000, 2(9):101-106. http://www.cnki.com.cn/Article/CJFDTOTAL-GCKX200009016.htm
[45]	赵立成.气象信息系统.北京:气象出版社, 2011:62;72-84.
[46]	王彬. 面向气象计算的NMIC Grid. 2010国际信息技术与应用论坛论文集. 计算机科学, 2010, 37(7A): 23-25.
[47]	肖文名, 李永生, 陈晓宇, 等.高性能计算系统性能评测关键问题探讨.计算机系统应用, 2008, 17(3):115-118. http://www.cnki.com.cn/Article/CJFDTOTAL-XTYY200803031.htm
[48]	吕爽, 衡志炜, 马艳军.西南区域气象中心IBM高性能计算机管理及应用.高原山地气象研究, 2015, 35(2):71-76. http://www.cnki.com.cn/Article/CJFDTOTAL-SCCX201502012.htm
[49]	王彬, 肖文名, 李永生, 等.华南区域中心计算资源管理系统的建立与应用.气象, 2011, 37(6):764-770. doi: 10.7519/j.issn.1000-0526.2011.06.016
[50]	袁卫华, 赵玉娟, 孟冬梅, 等.天津市中尺度数值预报业务系统的设计与实现.气象科技, 2011, 39(6):828-833. http://www.cnki.com.cn/Article/CJFDTOTAL-QXKJ201106030.htm
[51]	袁金南, 王在志, 薛纪善.广州区域数值预报模式并行化计算.应用气象学报, 2004, 15(5):557-563. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20040568&flag=1
[52]	陈德辉, 沈学顺.新一代数值预报系统GRAPES研究进展.应用气象学报, 2006, 17(6):773-777. doi: 10.11898/1001-7313.20060614
[53]	伍湘君, 金之雁, 黄丽萍, 等.GRAPES模式软件框架与实现.应用气象学报, 2005, 16(4):539-546. doi: 10.11898/1001-7313.20050415
[54]	伍湘君, 金之雁, 陈德辉, 等.新一代数值预报模式GRAPES的并行计算方案设计与实现.计算机研究与发展, 2007, 44(3):510-515. http://www.cnki.com.cn/Article/CJFDTOTAL-JFYZ200703020.htm
[55]	王彬, 宗翔, 魏敏.一个精细粒度实时计算资源管理系统.应用气象学报, 2008, 19(4):507-511. doi: 10.11898/1001-7313.20080416
[56]	常飚, 王彬.气象计算资源管理系统设计.计算机工程, 2011, 37(8):281-284. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJC201108098.htm
[57]	刘立明, 王彬.气象网格环境下大数据的端到端传输机制研究.计算技术与自动化, 2014, 33(1):122-126. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJH201401025.htm
[58]	宗翔, 王彬.国家级气象高性能计算机管理与应用网络平台设计.应用气象学报, 2006, 17(5):629-634. doi: 10.11898/1001-7313.20060506
[59]	董波, 沈青, 肖德宝.云计算集群服务器系统监控方法的研究.计算机工程与科学, 2012, 34(10):68-72. doi: 10.3969/j.issn.1007-130X.2012.10.014
[60]	陈诗然, 胡凯, 张伟, 等.多集群并行作业的性能监测及分析.计算机工程, 2008, 34(13):75-77. doi: 10.3969/j.issn.1000-3428.2008.13.028
[61]	王彬, 常飚, 朱江, 等.气象计算网格平台资源监视模块的设计与实现.应用气象学报, 2009, 20(5):642-648. doi: 10.11898/1001-7313.20090517
[62]	王彬, 周斌, 魏敏.气象计算网格模式预报系统的建立与优化.计算机应用研究, 2010, 27(11):4182-4184. http://www.cnki.com.cn/Article/CJFDTOTAL-JSYJ201011050.htm
[63]	王彬.一个计算作业网格执行环境的分析、设计与应用.计算机应用研究, 2008, 25(8):2546-2549. http://www.cnki.com.cn/Article/CJFDTOTAL-JSYJ200808089.htm
[64]	陈萍, 王彬, 徐国市, 等.网格计算的应用支持平台和运行环境分析.计算机应用研究, 2004, 21(9):50-52;58. http://www.cnki.com.cn/Article/CJFDTOTAL-JSYJ200409017.htm
[65]	Dongarra J, London K, Moore S, et al.Using PAPI for Hardware Performance Monitoring on Linux Systems//Proc Conf on Linux Clusters:The HPC Revolution.2001:25-27.
[66]	魏敏, 徐金秀, 王在志.并行I/O技术在气候数值模式中的应用研究.计算机技术与发展, 2014, 24(12):11-15. http://www.cnki.com.cn/Article/CJFDTOTAL-WJFZ201412003.htm
[67]	吴竑, 王小鸽, 杨广文, 等.THREP:面向地球系统模式的插值算法研究平台.计算机工程与应用, 2013, 49(15):48-55. doi: 10.3778/j.issn.1002-8331.1301-0255
[68]	王卓薇, 许先斌, 赵武清, 等.基于GPU的GRAPES模型并行加速及性能优化.计算机研究与发展, 2013, 50(2):401-411. doi: 10.7544/issn1000-1239.2013.20110219
[69]	伍湘君, 黄丽萍.超级计算机上矩阵乘的并行计算与实现.应用气象学报, 2005, 16(1):122-128. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20050116&flag=1
[70]	金之雁, 王鼎兴.大规模数据并行问题的可扩展性分析.应用气象学报, 2003, 14(3):369-374. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20030345&flag=1
[71]	彭新东, 李兴良.多尺度大气数值预报的技术进展.应用气象学报, 2010, 21(2):129-138. doi: 10.11898/1001-7313.20100201
[72]	魏敏, 罗勇, 王兰宁, 等.海气耦合模式的优化方法研究.应用气象学报, 2005, 16(3):408-412. doi: 10.11898/1001-7313.20050316
[73]	金之雁, 王鼎兴.一种在异构系统中实现负载平衡的方法.应用气象学报, 2003, 14(4):410-418. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20030451&flag=1
[74]	刘灿灿, 张卫民, 骆志刚, 等.面向集合预报的高性能计算环境.计算机工程与科学, 2012, 34(2):87-92. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJK201202022.htm
[75]	孙健, 赵平.用WRF与MM5模拟1998年三次暴雨过程的对比分析.气象学报, 2003, 61(6):692-701. doi: 10.11676/qxxb2003.070