[1]王 兴,王 新,苗春生,等.基于GPU加速的雷暴追踪外推方法研究[J].南京师范大学学报(工程技术版),2015,15(01):035-42.
 Wang Xing,Wang Xin,Miao Chunsheng,et al.Research on Thunderstorm Track and ExtrapolationBased on GPU-Acceleration[J].Journal of Nanjing Normal University(Engineering and Technology),2015,15(01):035-42.
点击复制

基于GPU加速的雷暴追踪外推方法研究
分享到:

南京师范大学学报(工程技术版)[ISSN:1006-6977/CN:61-1281/TN]

卷:
15卷
期数:
2015年01期
页码:
035-42
栏目:
计算机工程
出版日期:
2015-03-20

文章信息/Info

Title:
Research on Thunderstorm Track and ExtrapolationBased on GPU-Acceleration
作者:
王 兴1王 新2苗春生1王介君1
(1.南京信息工程大学大气科学学院,江苏 南京 210044)(2.北京华风气象影视信息集团公司,北京 100081)
Author(s):
Wang Xing1Wang Xin2Miao Chunsheng1Wang Jiejun1
(1.School of Atmospheric Science,Nanjing University of Information Science and Technology,Nanjing 210044,China)(2.Huafeng Group of Meteorological Audio and Video Information,CMA,Beijing 100081,China)
关键词:
光流计算开放运算语言并行化临近预报雷达资料
Keywords:
optical flow computationOpenCLparallel computingnowcastingradar data
分类号:
TP391.4
文献标志码:
A
摘要:
基于气象雷达的雷暴识别与追踪是临近预报中重要的方法之一. 为解决传统算法实时性差的问题,运用OpenCL构建异构计算模型对算法进行并行化改进. 通过对算法分支结构优化、OpenCL设备内存优化,以及针对VLIW的优化,分步阐述算法优化的过程和原理. 这些方法不仅使得基于光流的计算速度大幅提升,还可为其他基于OpenCL异构计算的优化提供参考. 以AMD两代不同架构的GPU和Intel XEON CPU作为测试平台测试,结果表明,改进后的算法程序在硬件同等功耗的情况下,计算速度提高了10至18倍.
Abstract:
The thunderstorms recognition and tracking based on weather radar is one of the important methods in weather nowcasting. In order to solve the problem that the traditional algorithm is poor real-time,this paper discusses the heterogeneous computing model based on GPU,and presents that parallel the algorithm with OpenCL to achieve high performance. By the methods of branching structure optimization,OpenCL device memory optimization,and optimized for VLIW,this paper expounds the optimization of the algorithm step by step. These methods not only boost the speed of optical flow computation,but also provide a reference for other optimization based on OpenCL heterogeneous computing. We use AMD’s two generations of different architecture of GPU and Intel XEON CPU as a test platform. The tests results show that the computing performance improve 10-18 times under the circumstance of same power consumption.

参考文献/References:

[1] 陆林. 我国公共气象服务能力建设研究[D]. 云南:云南大学公共管理学院,2013.
Lu Ling. Research of the construction ability for public meteorological service in China [D]. Yun Nan:School of Public Administration,Yunnan University,2013.(in Chinese)
[2]陈明轩,王迎春,俞小鼎,等. 交叉相关外推算法的改进及其在对流临近预报中的应用[J]. 应用气象学报,2007,18(5):690-700.
Chen Mingxuan,Wang Yingchun,Yu Xiaoding. Improvement and application test of TREC algorithm for convective storm nowcast [J]. Journal of Applied Meteorological Science,2007,18(5):690-700.
[3]Johnson J T,MacKeen P L,Witt A,et al. The storm cell identification and tracking algorithm:an enhanced WSR-88D algorithm[J]. Weather and Forecasting,1998,13:263-276.
[4]韩雷,王洪庆,林隐静. 光流法在强对流天气临近预报中的应用[J]. 北京大学学报:自然科学版,2008,44(5):751-755.
Han Lei,Wang Hongqing,Lin Yinjing. Application of optical flow method to nowcasting convective weather[J]. Acta Scientiarum Naturalium Universitatis Pekinensis,2008,44(5):751-755.(in Chinese)
[5]Owens J D,Luebke D,Govindaraju N,et al. A survey of general-purpose computation on graphics hardware[J]. Computer Graphics Forum,2007,26(1):80-113.
[6]Strecha C,Fransens R,Gool L V,A probabilistic approach to large displacement optical flow and occlusion detection,statistical methods in video processing[J]. Lecture Notes in Computer Science,2004,3 247:1-82.
[7]Xiao J,Cheng H,Sawhney H,et al. Bilateral filtering-based optical flow estimation with occlusion detection[J]. ECCV,Lecture Notes in Computer Science,2006,3951:1-224.
[8]朱倩. 基于改进的光流场算法对运动目标的检测与跟踪技术研究[D]. 哈尔滨:哈尔滨工程大学信息与通信工程学院,2006.
Zhu Qian. Moving objects detection and tracking based on revised optical flow[D]. Harbin:College of Information and Communication Engineering,Harbin Engineering University,2006.(in Chinese)
[9]刘闵. 中尺度对流系统中多个云团的跟踪和预测[D]. 南京:南京理工大学模式识别与智能系统,2013.
Liu Min. Tracking and prediction of multiple-clouds in the mesoscale convective system[D]. Nanjing:Pattern Recognition and Intelligent Systems,Nanjing University of Science and Technology,2013.(in Chinese)
[10]江志军,易华蓉. 一种基于图像金字塔光流的特征跟踪方法[J]. 武汉大学学报:信息科学版,2007,32(8):680-683.
Jiang Zhijun,Yi Huarong. An image pyramid-based feature detection and tracking algorithm[J]. Geomatics and Information Science of Wuhan University,2007.(in Chines)
[11]张云泉,张先轶,龙国平,等. OpenCL异构计算[M]. 北京:清华大学出版社,2012:1-148.
Zhang Yunquan,Zhang Xianyi,Long Guoping,et al. Heterogeneous Computing with OpenCL[M]. Beijing:Tsinghua University Press,2012:1-148.(in Chinese)
[12]Advanced Micro Devices,Inc. AMD Accelerated Parallel Processing OpenCL Programming Guide[M]. [S.l.]. [s.n.]. November 2013.
[13]Aaftab Munshi,Benedict Gaster,Timothy G Mattson,et al. OpenCL Programming Guide[M]. Boston:Addison-Wesley Professional,2011.
[14]李约炯. 跨平台的多核与众核编程讲义[M]. 上海:AMD上海研发中心,2010.
Li Yuejiong. Cross-platform and Multi-core Programming Notes[M]. Shanghai:AMD Shanghai r&d Center,2010.(in Chinese)
[15]佚名. AMD的显卡架构与OpenCL性能之间的一点思考[EB/OL]. [2014-03-27]. http://pengx17.me/opencl/2013/09/25/amd-architect/.
Anon. Thinking about AMD Graphics between architecture and OpenCL Performance[EB/OL]. [2014-03-27]. http://pengx17. me/opencl/2013/09/25/amd-architect/.(in Chinese)
[16]仇德元. 从GLSL、CUDA到OpenCL[M]. 北京:机械工业出版社,2011.
Qiu Deyuan. From the GLSL,CUDA to OpenCL[M]. Beijing:China Machine Press,2011.(in Chinese)
[17]Guo Yidong,Liu Bozhong,Qiu Weidong. General-purpose oriented GPU optimization techniques on AMD platforms[C]//Proceedings of 2013 IEEE International Conference on Computer Science and Automation Engineering. Guangzhou:IEEE Beijing Section,2013:1 284-1 288.
[18]邵晨. 基于雷达与闪电观测的短历时强降水分类研究[D]. 南京:南京信息工程大学大气科学学院,2012.
Shao Chen. Classification of short-time heavy rain using radar and lightning observations[D]. Nanjing:School of Meteorology Science,Nanjing University of Information Science and Technology,2012.(in Chinese)

备注/Memo

备注/Memo:
收稿日期:2014-08-20.基金项目:国家科技支撑计划课题(2012BAH05B01)、公益性行业科研计划课题(GYHY201206030)、大学生实践创新训练计划项目(201410300119). 通讯联系人:王兴,博士,工程师,研究方向:气象信息安全技术. E-mail:sweetdreamworks@qq.com
更新日期/Last Update: 2015-03-30