基于CUDA的泥石流模拟计算研究
|
摘要
针对泥石流仿真过程中海量数据计算问题,采用CUDA技术即结合CPU与GPU的优点研究了一种协同计算方法以提高数据计算效率和仿真性能。同时,搭建了基于GPU的泥石流仿真计算平台,对优化的CUDA并行计算方法进行验证。实验结果表明,该方法对海量数据的计算具有快速准确、低成本、低功耗的特点,能为灾害预测提供及时准确的决策支持,满足了高密集型计算的需求。
In order to solve the mass data computing problems in the simulation process of debris flow,a collaborative computing method,employing the CUDA technology which combined the advantages of CPU and GPU is proposed to enhance the computing efficiency and the simulation performance.Besides,the debris flow simulation computing platform is built based on GPU to valid the optimized CUDA parallel method.And the simulation results show that the optimized CUDA parallel method has the characteristics of fast,accurate,low cost and low power consumption and it provide timely and accurate decision support for disaster prediction and meet the requirements of high intensive computing.
In order to solve the mass data computing problems in the simulation process of debris flow,a collaborative computing method,employing the CUDA technology which combined the advantages of CPU and GPU is proposed to enhance the computing efficiency and the simulation performance.Besides,the debris flow simulation computing platform is built based on GPU to valid the optimized CUDA parallel method.And the simulation results show that the optimized CUDA parallel method has the characteristics of fast,accurate,low cost and low power consumption and it provide timely and accurate decision support for disaster prediction and meet the requirements of high intensive computing.
引文
[1]Hwu W W,Keutzer K,Mattson T.The Concourrency challenge[C].IEEE Design and Test of Computers,2008:312-320.
[2]张舒,褚艳丽,赵开勇,等.GPU高性能运算之CUDA[M].北京:中国水利水电出版社,2009.
[3]Wing J.Computational thinking[J].Communications of theACM,2006,49(3):33-35.
[4]YANG Yu,GUAN Qun,HU Kai-heng,et al.GIS-based distributedhydrological computational system of debris-flow atershed[J].Computer Engineering,2010,36(5):261-265.
[5]GUAN Qun,WEI Fangqiang,HU Kaieng.Study of risk zoningsystem of debris flow based on real time numerical simulation[J].Application Research of Computer,2006,23(12):54-55.
[6]费祥俊,舒安平.泥石流运动机理与灾害防治[M].北京:清华大学出版社,2004.
[7]NVADIA.CUDA programinge guide 2.3[M].Samta Clara,CA:NVIDIA,2007.
[8]张正栋.SuperMap GIS应用与开发教程[M].武汉:武汉大学出4236 2011,Vol.32,No.12计算机工程与设计Computer Engineering and Design版社,2006.
[9]Rong Guodong,Tan Tiow-Seng,Cao Thanh-Tung,et al.Compu-ting two-dimensional Delaunay triangulation using graphicshardware[C].Proceedings of the Symposium on Interactive 3DGraphics and Games,2008.
[10]崔雪冰,张延红,李国徽.基于通用计算的GPU-CPU协作计算模式研究[J].微电子学与计算机,2009,26(8):30-33.
[11]苏凤环,刘洪江,韩用顺.汶川地震山地灾害遥感快速提取及其分布特点分析[J].遥感学报,2008,12(6):956-963.
[12]王纯祥,白世伟,江崎哲朗,等.泥石流的二维数学模型[J].岩土力学,2007,28(6):1236-1240.
[13]Davidb Kirk,Wen-mei W Hwu.Programming Massively ParallelProcessors:A Hands-on Approach[M].Morgan Kaufmann,2010.
[14]Draft standard for floatingpoint arithmetic P754.Piscataway,NJ:institute of electrical and electronics engineers[C].IEEE Micro-processor Standards Committee,2008.
[15]Message passing interface forme.MPI:A message-passing inter-face standard,version2.2[EB/OL].http://www.mpi-forum.org/docs/mpi-2.2/mpi22-report.pdf,2009.
[16]AMD.Introductory tutorial to openCL.Sunnyvale,CA:advancedMicro devices[EB/OL].http://developer.amd.com/gpu/ATIS-treamSDK/pages/TutorialOpenCL.aspx,2009.
[2]张舒,褚艳丽,赵开勇,等.GPU高性能运算之CUDA[M].北京:中国水利水电出版社,2009.
[3]Wing J.Computational thinking[J].Communications of theACM,2006,49(3):33-35.
[4]YANG Yu,GUAN Qun,HU Kai-heng,et al.GIS-based distributedhydrological computational system of debris-flow atershed[J].Computer Engineering,2010,36(5):261-265.
[5]GUAN Qun,WEI Fangqiang,HU Kaieng.Study of risk zoningsystem of debris flow based on real time numerical simulation[J].Application Research of Computer,2006,23(12):54-55.
[6]费祥俊,舒安平.泥石流运动机理与灾害防治[M].北京:清华大学出版社,2004.
[7]NVADIA.CUDA programinge guide 2.3[M].Samta Clara,CA:NVIDIA,2007.
[8]张正栋.SuperMap GIS应用与开发教程[M].武汉:武汉大学出4236 2011,Vol.32,No.12计算机工程与设计Computer Engineering and Design版社,2006.
[9]Rong Guodong,Tan Tiow-Seng,Cao Thanh-Tung,et al.Compu-ting two-dimensional Delaunay triangulation using graphicshardware[C].Proceedings of the Symposium on Interactive 3DGraphics and Games,2008.
[10]崔雪冰,张延红,李国徽.基于通用计算的GPU-CPU协作计算模式研究[J].微电子学与计算机,2009,26(8):30-33.
[11]苏凤环,刘洪江,韩用顺.汶川地震山地灾害遥感快速提取及其分布特点分析[J].遥感学报,2008,12(6):956-963.
[12]王纯祥,白世伟,江崎哲朗,等.泥石流的二维数学模型[J].岩土力学,2007,28(6):1236-1240.
[13]Davidb Kirk,Wen-mei W Hwu.Programming Massively ParallelProcessors:A Hands-on Approach[M].Morgan Kaufmann,2010.
[14]Draft standard for floatingpoint arithmetic P754.Piscataway,NJ:institute of electrical and electronics engineers[C].IEEE Micro-processor Standards Committee,2008.
[15]Message passing interface forme.MPI:A message-passing inter-face standard,version2.2[EB/OL].http://www.mpi-forum.org/docs/mpi-2.2/mpi22-report.pdf,2009.
[16]AMD.Introductory tutorial to openCL.Sunnyvale,CA:advancedMicro devices[EB/OL].http://developer.amd.com/gpu/ATIS-treamSDK/pages/TutorialOpenCL.aspx,2009.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心