高级检索

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于J-Earth领域编程框架的高效并行大气环流模式(H-GAMIL)构建

普业 张爱清 莫则尧 郭红 王斌

普业, 张爱清, 莫则尧, 郭红, 王斌. 基于J-Earth领域编程框架的高效并行大气环流模式(H-GAMIL)构建[J]. 气候与环境研究, 2017, 22(1): 71-79. doi: 10.3878/j.issn.1006-9585.2016.16020
引用本文: 普业, 张爱清, 莫则尧, 郭红, 王斌. 基于J-Earth领域编程框架的高效并行大气环流模式(H-GAMIL)构建[J]. 气候与环境研究, 2017, 22(1): 71-79. doi: 10.3878/j.issn.1006-9585.2016.16020
Ye PU, Aiqing ZHANG, Zeyao MO, Hong GUO, Bin WANG. The Development of High Performance Atmospheric General Circulation Model (H-GAMIL) Base on Programming Framework J-Earth[J]. Climatic and Environmental Research, 2017, 22(1): 71-79. doi: 10.3878/j.issn.1006-9585.2016.16020
Citation: Ye PU, Aiqing ZHANG, Zeyao MO, Hong GUO, Bin WANG. The Development of High Performance Atmospheric General Circulation Model (H-GAMIL) Base on Programming Framework J-Earth[J]. Climatic and Environmental Research, 2017, 22(1): 71-79. doi: 10.3878/j.issn.1006-9585.2016.16020

基于J-Earth领域编程框架的高效并行大气环流模式(H-GAMIL)构建

doi: 10.3878/j.issn.1006-9585.2016.16020
基金项目: 

国家自然科学基金项目 d41205079

国家重点基础研究发展计划项目(973计划) 2015CB954102

详细信息
    作者简介:

    普业,男,1978年出生,博士研究生,高级工程师,主要从事高性能计算、数值模拟研究,E-mail:puye@lasg.iap.ac.cn

  • 中图分类号: P435

The Development of High Performance Atmospheric General Circulation Model (H-GAMIL) Base on Programming Framework J-Earth

Funds: 

the National Natural Science Foundation of China d41205079

National Program on Key Basic Research Program of China (973 Program) 2015CB954102

  • 摘要: 高性能并行计算,一直是气候系统模式研究中的重要科学前沿。然而,随着计算机技术的发展,计算机体系结构越来越复杂,以及气候系统模式本身的复杂性使得开发高性能气候系统模式成为一个难题。采用了一种全新的并行程序设计方法:基于针对气象科学中的计算特点发展出的领域编程框架J-Earth,构建了高性能并行大气环流模式H-GAMIL。在对原模式GAMIL分析和解构以后,结合J-Earth的特点,采用面相对象的编程方式,对H-GAMIL进行重构。重构后的H-GAMIL具有现代软件标准化模块化的特点,并具有自动二维剖分、MPI(Message PassingInterface)和OpenMP(Open Multi-Processing)混合并行实现、并行输出等新功能。解决了原模式一维剖分受到处理器限制,串行IO等问题。在此基础上,本文对H-GAMIL进行了并行效率测试,结果显示,模式水平分辨率为1°(纬度)×1°(经度)时,可使用上千处理器核,并行效率达40%以上且负载平衡达70%。测试结果表明了H-GAMIL具有较好的并行效率和可扩展性;同时对H-GAMIL模拟性能进行对比分析,其结果表明H-GAMIL保持了原模式的守恒性,并且具有与原模式相当的模拟性能,能够满足气候研究的计算需求,达到实用程度。
  • 图  1  并行应用软件框架示意图

    Figure  1.  A schematic diagram of the parallel application software framework

    图  2  J-Earth 并行框架中的(a)网格层和(b)网格片

    Figure  2.  (a)Patches and(b)levels in the J-Earth parallel software framework

    图  3  GAMIL 模式主要结构

    Figure  3.  Main structure of GAMIL model

    图  4  H-GAMIL 的层次化模块结构

    Figure  4.  Hierarchical module structure of H-GAMIL model

    图  5  H-GAMIL 网格片patch 示意图

    Figure  5.  A schematic diagram of patch in H-GAMIL model

    图  6  GAMIL 与H-GAMIL 的并行效率测试

    Figure  6.  The parallel test between GAMIL and H-GAMIL model

    图  7  1°(纬度)×1°(经度)H-GAMIL 并行效率及负载平衡效率测试

    Figure  7.  The parallel efficiency and load balance test of 1°×1° H-GAMIL model

    图  8  (a)H-GAMIL 全球年平均降水的模拟结果,(b)原GAMIL 模式全球年平均降水的模拟结果,(c)H-GAMIL 与GAMIL 模拟的差值

    Figure  8.  (a)Annual mean rainfall from H-GAMIL simulation.(b)The same as(a)but from GAMIL.(b)Differences between H-GAMIL and GAMIL simulations

    图  9  (a)H-GAMIL 全球年平均850 hPa 温度场的模拟结果,(b)原 GAMIL 模式全球年平均850 hPa 的模拟结果,(c)H-GAMIL 与GAMIL模拟的差值

    Figure  9.  (a)Annual mean temperature from H-GAMIL simulation.(b)The same as(a)but from GAMIL.(c)Differences between H-GAMIL and GAMIL simulations

    表  1  H-GAMIL 并行加速比和并行效率

    Table  1.   The speed up and parallel performance of H-GAMIL

    CPU 数 耗时/s 加速比 并行效率
    12 807.3 1 100%
    36 269.8 2.99 99.7%
    72 145.2 5.60 93.3%
    144 82.0 9.85 82.1%
    192 66.4 12.16 76%
    384 46.5 17.36 54.3%
    576 34.8 23.20 48.3%
    768 31.3 25.79 40.3%
    1035 23.4 34.5 40%
    下载: 导出CSV
  • [1] Bacmeister J T, Wehner M F, Neale R B, et al. 2014. Exploratory high-resolution climate simulations using the community atmosphere model (CAM)[J]. J. Climate, 27(9):3073-3099, doi: 10.1175/JCLI-D-13-00387.1.
    [2] 毕训强, 曾庆存. 1997. 气候模式并行计算[J]. 气候与环境研究, 2(2):100-105. doi: 10.3878/j.issn.1006-9585.1997.02.01

    Bi Xunqiang, Zeng Qingcun. 1997. Parallel computing by using climate models[J]. Climatic and Environmental Research (in Chinese), 2(2):100-105, doi: 10.3878/j.issn.1006-9585.1997.02.01.
    [3] 程汤培, 莫则尧, 邵景力. 2013. 基于JASMIN的地下水流大规模并行数值模拟[J]. 计算物理, 30(3):317-325. doi: 10.3969/j.issn.1001-246X.2013.03.001

    Cheng Tangpei, Mo Zeyao, Shao Jingli. 2013. JASMIN-based massive parallel computing of large scale groundwater flow[J]. Chinese Journal of Computational Physics (in Chinese), 2013, 30(3):317-325, doi:10.3969/j.issn.1001-246X.2013. 03.001.
    [4] Dong L, Li L J, Huang W Y, et al. 2012. Preliminary evaluation of cloud fraction simulations by GAMIL2 using COSP[J]. Atmospheric and Oceanic Science Letters, 5(3):258-263, doi:10.1080/16742834.2012. 11447002.
    [5] Kodam C, Yamada Y, Noda A T, et al. 2015. A 20-year climatology of a NICAM AMIP-type simulation[J]. J. Meteor. Soc. Japan, 93(4):393-424, doi: 10.2151/jmsj.2015-024.
    [6] Kucharski F, Scaife A A, Yoo J H, et al. 2009. The CLIVAR C20C project:Skill of simulating Indian monsoon rainfall on interannual to decadal timescales. Does GHG forcing play a role?[J]. Climate Dyn., 33(5):615-627, doi: 10.1007/s00382-008-0462-y.
    [7] Li L J, Wang B, Zhou T J. 2007a. Impacts of external forcing on the 20th century global warming[J]. Chinese Science Bulletin, 52(22):3148-3154, doi: 10.1007/s11434-007-0463-y.
    [8] Li L J, Wang B, Wang Y Q, et al. 2007b. Improvements in climate simulation with modifications to the Tiedtke convective parameterization in the grid-point atmospheric model of IAP LASG (GAMIL)[J]. Advances in Atmospheric Sciences, 24(2):323-335, doi: 10.1007/s00376-007-0323-3.
    [9] Li L J, Wang B. 2010. Influences of two convective schemes on the radiative energy budget in GAMIL1.0[J]. Acta Meteorologica Sinica, 24(3):318-327. http://cn.bing.com/academic/profile?id=696c7ce23dc384c7b1b28eb4b190469a&encoded=0&v=paper_preview&mkt=zh-cn
    [10] Li L J, Xie X, Wang B, et al. 2012. Evaluating the performances of GAMIL1.0 and GAMIL2.0 during TWP-ICE with CAPT[J]. Atmospheric and Oceanic Science Letters, 5(1):38-42. doi: 10.1080/16742834.2012.11446963
    [11] Li L J, Wang B, Dong L, et al. 2013. Evaluation of grid-point atmospheric model of IAP LASG version 2(GAMIL2)[J]. Advances in Atmospheric Sciences, 30(3):855-867, doi: 10.1007/s00376-013-2157-5.
    [12] Mo Z Y, Zhang A Q, Cao X L, et al. 2010. Jasmin:A parallel software infrastructure for scientific computing[J]. Frontiers of Computer Science in China, 4(4):480-488, doi: 10.1007/s11704-010-0120-5.
    [13] Pei W B. 2007. The construction of simulation algorithms for Laser Fusion[J]. Communication in Computational Physics, 2(2):255-270. http://cn.bing.com/academic/profile?id=bfd55f6d99e99f9f29909252bf003323&encoded=0&v=paper_preview&mkt=zh-cn
    [14] Scaife A A, Kucharski F, Folland C K, et al. 2009. The CLIVAR C20C project:Selected twentieth century climate events[J]. Climate Dyn., 33(5):603-614, doi: 10.1007/s00382-008-0451-1.
    [15] 王斌. 2009. 一种典型的高性能计算:地球系统模拟[J]. 物理, 38(8):569-574. doi: 10.3321/j.issn:0379-4148.2009.08.006

    Wang Bin. 2009. A typical type of high-performance computation:Earth system modeling[J]. Physics (in Chinese), 38(8):569-574, doi: 10.3321/j.issn:0379-4148.2009.08.006.
    [16] Wang B, Wan H, Ji Z Z, et al. 2004. Design of a new dynamical core for global atmospheric models based on some efficient numerical methods[J]. Science in China Series A:Mathematics, 47(S1):4-21, doi: 10.1360/04za0001.
    [17] 王斌, 周天军, 俞永强, 等. 2008. 地球系统模式发展展望[J]. 气象学报, 66(6):857-869. doi: 10.3321/j.issn:0577-6619.2008.06.002

    Wang Bin, Zhou Tianjun, Yu Yongqiang, et al. 2008. A perspective on earth system model development[J]. Acta Meteorologica Sinica (in Chinese), 66(6):857-869, doi: 10.3321/j.issn:0577-6619.2008.06.002.
    [18] 王鹏飞, 马晓光, 李薇, 等. 2006. 气候数值模式在计算网格环境下的应用[J]. 气候与环境研究, 11(6):745-753. doi: 10.3969/j.issn.1006-9585.2006.06.009

    Wang Pengfei, Ma Xiaoguang, Li Wei, et al. 2006. Climatic numerical model applied in computing grid environments[J]. Climatic and Environmental Research (in Chinese), 11(6):745-753, doi: 10.3969/j.issn.1006-9585.2006.06.009.
    [19] 张昕, 季仲贞, 王斌. 2001. OpenMP在MM5中尺度模式中的应用试验[J]. 气候与环境研究, 6(1):84-90. doi: 10.3969/j.issn.1006-9585.2001.01.010

    Zhang Xin, Ji Zhongzhen, Wang Bin. 2001. Some study on application of OpenMP in mesoscale meteorological Model-MM5[J]. Climatic and Environmental Research (in Chinese), 6(1):84-90, doi: 10.3969/j.issn.1006-9585.2001.01.010.
  • 加载中
图(9) / 表(1)
计量
  • 文章访问数:  2282
  • HTML全文浏览量:  15
  • PDF下载量:  1646
  • 被引次数: 0
出版历程
  • 收稿日期:  2016-01-24
  • 网络出版日期:  2016-04-19
  • 刊出日期:  2017-01-01

目录

    /

    返回文章
    返回