[1]丁科,谭营.GPU通用计算及其在计算智能领域的应用[J].智能系统学报,2015,10(01):1-11.[doi:10.3969/j.issn.1673-4785.201403072]
　DING Ke,TAN Ying.A review on general purpose computing on GPUs and its applications in computational intelligence[J].CAAI Transactions on Intelligent Systems,2015,10(01):1-11.[doi:10.3969/j.issn.1673-4785.201403072]
点击复制

GPU通用计算及其在计算智能领域的应用(/HTML)
分享到：

《智能系统学报》[ISSN:1673-4785/CN:23-1538/TP]

卷:
第10卷
期数:
2015年01期
页码:
1-11
栏目:
学术论文—智能系统
出版日期:
2015-03-25

文章信息/Info

Title:
A review on general purpose computing on GPUs and its applications in computational intelligence
作者:
丁科1; 2;  谭营1; 2
1. 北京大学 机器感知与智能教育部重点实验室, 北京 100871;
2. 北京大学 信息科学技术学院, 北京 100871
Author(s):
DING Ke1; 2;  TAN Ying1; 2
1. Key Laboratory of Machine Perception (MOE), Peking University, Beijing 100871, China;
2. School of Electronics Engineering and Computer Science, Peking University, Beijing 100871, China
关键词:
计算智能; 群体智能; 演化算法; 机器学习; 深度学习; 图形处理器; GPU通用计算; 异构计算; 高性能计算
Keywords:
computational intelligence; swarm intelligence; evolutionary algorithms; machine learning; deep learning; graphics processing unit (GPU); general purpose computing on GPUs; heterogonous computing; high performance computing (HPC)
分类号:
TP301
DOI:
10.3969/j.issn.1673-4785.201403072
文献标志码:
A
摘要:
在日趋复杂的图形处理任务的推动下,GPU已经演化成为具有众多计算核心、计算能力强大的通用计算设备,并被越来越多地应用于图形处理之外的计算领域。GPU具有高并行、低能耗和低成本的特点,在数据并行度高的计算任务中,相比与传统的CPU平台有着显著的优势。随着GPU体系结构的不断演进以及开发平台的逐步完善,GPU已经进入到高性能计算的主流行列。GPU通用计算的普及,使个人和小型机构能有机会获得以往昂贵的大型、超级计算机才能提供的计算能力,并一定程度上改变了科学计算领域的格局和编程开发模式。GPU提供的强大计算能力极大地推动了计算智能的发展,并且已经在深度学习和群体智能优化方法等子领域获得了巨大的成功,更是在图像、语音等领域取得了突破性的进展。随着人工智能技术和方法的不断进步,GPU将在更多的领域获得更加广泛的应用。
Abstract:
The GPU enjoys the characteristics of high parallelism, low energy consumption and cheap price. Compared with the traditional CPU platform, it is especially suitable for tasks with high data parallelism. GPU computing has come into the mainstream of high performance computation (HPC) due to the emerging of development platforms like CUDA and OpenCL. The GPU’s enormous computational power greatly promotes computational intelligence. A great success has been achieved in the fields such as deep learning and swarm intelligence optimization, and several breakthroughs have been seen in image, and speech recognition because of GPU. Though suffering some drawbacks, GPUs provide common people and small institutions with enormous computing power. This has changed the set-up of scientific computing and programming model because it could only be provided by expensive supercomputers. To help researchers in the field of computational intelligence better utilize GPUs, a detailed survey of GPGPU is given in this paper。First, the characteristics and advantages of GPUs against CPUs are presented. Then we briefly review the development of GPU hardware followed by a survey of the evolution of development tools for GPGPU; special attention is drawn to two major platforms, CUDA and OpenCL. We end this paper with our perspectives of the challenges and trends of GPGPU. We point out that embedding and cluster are two major trends for GPGPU and as both academia and industry continue to see increasing progress in artificial intelligence, the GPU will be more widely used in more domains.

参考文献/References:

[1] 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.
[2] OWENS J D, LUEBKE D, GOVINDARAJU N, et al. GPU computing[J]. Proceedings of the IEEE, 2008, 96(5): 879-899.
[3] SUTTER H. The free lunch is over: a fundamental turn toward concurrency in software[J]. Dr. Dobb’s Journal, 2005, 30(3): 202-210.
[4] ROSS P E. Why CPU frequency stalled[J]. Spectrum, 2008, 45(4): 72-78.
[5] BORKAR S. Getting gigascale chips: challenges and opportunities in continuing Moore’s Law[J]. Queue, 2003, 1(7): 26-33.
[6] NVIDIA. CUDA C programming guide v6.5[R]. Santa Clara, CA, USA: NVIDIA Corporation, 2014.
[7] JARARWEH Y, JARRAH M, BOUSSELHAM A, et al. GPU-based personal supercomputing[C]//2013 IEEE Jordan Conference on Applied Electrical Engineering and Computing Technologies. Amman, 2013: 1-5.
[8] KAPASI U J, RIXNER S, DALLY W J, et al. Programmable stream processors[J]. Computer, 2003, 36(8): 54-62.
[9] BUCK I, FOLEY T, HORN D, et al. Brook for GPUs: stream computing on graphics hardware[J]. ACM Transactions on Graphics, 2004, 23(3): 777-786.
[10] Microsoft. C++ accelerated massive parallelism[Z]. Redmond, WA, USA: Microsoft, 2013.
[11] NVIDIA. CUDA C best practices guide version 4.1[R]. Santa Clara, CA, USA: NVIDIA Corporation, 2012.
[12] NVIDIA. GPU-Accelerated Libraries.[OL/EB].[2015-01-05]. https://developer.nvidia.com/gpu-accelerated-libraries.
[13] JIA Y, SHELHAMER E, DONAHUE J, et al. Caffe: convolutional architecture for fast feature embedding[C]//Proceedings of the ACM International Conference on Multimedia,[s.l.], 2014: 675-678.
[14] GASTER B, HOWES L, KAELI D R,等. OpenCL异构计算[M]. 北京: 清华大学出版社, 2012: 10-35.
[15] KIRK D B, HWU W W. Programming massively parallel processors: a Hands-on approach[M]. Beijing: Tsinghua University Press, 2010: 205-220.
[16] MUNSHI A, GASTER B, MATTSON T G, et al. OpenCL Programming Guide[M]. Boston: Addison_Wesley Professional, 2011: 63-68.
[17] AMD上海研发中心. 跨平台的多核与从核编程讲义——OpenCL的方式[M]. 上海: AMD, 2010: 1-154.
[18] FARBER R. 高性能 CUDA应用设计与开发[M]. 北京:机械工业出版社, 2013: 1-49.
[19] ZEILER M, FERGUS R. Visualizing and understanding convolutional networks[C]//Proceedings of the 13th European Conference on Computer Vision. Zurich, Switzerland, 2014: 818-833.
[20] HINTON G, OSINDERO S, WELLING M, et al. Unsupervised discovery of nonlinear structure using contrastive backpropagation[J]. Nature, 2006, 30(4): 725-731.
[21] KRIZHEVSKY A, SUTSKEVER I, HINTON G. Imagenet classification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems 25. Reno, Nevada, USA, 2012: 1106-1114.
[22] COATES A, HUVAL B, WANG T, et al. Deep learning with COTS HPC systems[C]//Proceedings of the 30th International Conference on Machine Learning. Atlanta, USA, 2013: 1337-1345.
[23] ZHOU Y, TAN Y. GPU-based parallel particle swarm optimization[C]//IEEE Congress on Evolutionary Computation. Trondheim, Norway, 2009: 1493-1500.
[24] ZHOU Y, TAN Y. Particle swarm optimization with triggered mutation and its implementation based on GPU[C]//GECCO’10: Proceedings of the 12th Annual Conference on Genetic and Evolutionary Computation. Portland, Oregon, USA, 2010: 1-8.
[25] ZHOU Y, TAN Y. GPU-based parallel multi-objective particle swarm optimization[J]. International Journal of Artificial Intelligence, 2011, 7(A11): 125-141.
[26] DING K, TAN Y. A GPU-based parallel fireworks algorithm for optimization[C]//GECCO’13: Proceedings of the Fifteenth Annual Conference on Genetic and Evolutionary Computation Conference. Amsterdam, the Netherlands, 2013: 9-16.
[27] TAN Y, ZHU Y. Fireworks algorithm for optimization[C]//First International Conference of Swarm Intelligence. Beijing, China, 2010: 355-364.
[28] RYMUT B, KWOLEK B. GPU-supported object tracking using adaptive appearance models and particle swarm optimization[C]//International Conference on Computer Vision and Graphics, Warsaw, Poland, 2010: 227-234.
[29] MUSSI L, IVEKOVIC S, CAGNONI S. Markerless articulated human body tracking from multi-view video with GPU-PSO[C]//9th International Conference on Environmental Systems. York, UK, 2010: 97-108.
[30] NOBILE M S, BESOZZI D, CAZZANIGA P, et al. A GPU-based multi-swarm PSO method for parameter estimation in stochastic biological systems exploiting discrete-time target series[C]//10th European Conference on Evolutionary Computation, Machine Learning and Data Mining in Computational Biology. Málaga, Spain, 2012, 7246: 74-85.
[31] MAGHAZEH A, BORDOLOI UD, ELES P, et al. General purpose computing on low-power embedded GPUs: has it come of age[R]. Linkping University Electronic Press, 2013.
[32] HALLMANS D, SANDSTROM K, LINDGREN M, et al. GPGPU for industrial control systems[C]//2013 IEEE 18th Conference on Emerging Technologies Factory Automation. Cagliari, Italy, 2013: 1-4.

相似文献/References:

[1]丁永生.计算智能的新框架：生物网络结构[J].智能系统学报,2007,2(02):26.
　DING Yong-sheng.A new scheme for computational intelligence: bio-network architecture[J].CAAI Transactions on Intelligent Systems,2007,2(01):26.
[2]康 琦,汪 镭,刘小莉,等.基于群体智能框架理念的遗传算法总体模式描述[J].智能系统学报,2007,2(05):42.
　KANG Qi,WANG Lei,LIU Xiao-li,et al.General　mode　description genetic algorithms based on a framework of swarm intelligence[J].CAAI Transactions on Intelligent Systems,2007,2(01):42.
[3]杨东升,康 琦,刘 波,等.面向生产系统的残次品主次成因的群体智能分析[J].智能系统学报,2009,4(06):502.[doi:10.3969/j.issn.1673-4785.2009.06.006]
　YANG Dong-sheng,KANG Qi,LIU Bo,et al.Swarm intelligence analysis of primary and secondary causes of defective products for manufacturing system[J].CAAI Transactions on Intelligent Systems,2009,4(01):502.[doi:10.3969/j.issn.1673-4785.2009.06.006]
[4]夏琳琳,张健沛,初妍.计算智能在移动机器人路径规划中的应用综述[J].智能系统学报,2011,6(02):160.
　XIA Linlin,ZHANG Jianpei,CHU Yan.An application survey on computational intelligence for path planning of mobile robots[J].CAAI Transactions on Intelligent Systems,2011,6(01):160.
[5]陈杰,沈艳霞,陆欣.基于信息反馈和改进适应度评价的人工蜂群算法[J].智能系统学报,2016,11(2):172.[doi:10.11992/tis.201506024]
　CHEN Jie,SHEN Yanxia,LU Xin.Artificial bee colony algorithm based on information feedback and an improved fitness value evaluation[J].CAAI Transactions on Intelligent Systems,2016,11(01):172.[doi:10.11992/tis.201506024]
[6]秦全德,程适,李丽,等.人工蜂群算法研究综述[J].智能系统学报,2014,9(02):127.[doi:10.3969/j.issn.1673-4785.201309064]
　QIN Quande,CHENG Shi,LI Li,et al.Artificial bee colony algorithm: a survey[J].CAAI Transactions on Intelligent Systems,2014,9(01):127.[doi:10.3969/j.issn.1673-4785.201309064]
[7]谭营,郑少秋.烟花算法研究进展[J].智能系统学报,2014,9(05):515.[doi:10.3969/j.issn.1673-4785.201409010]
　TAN Ying,ZHENG Shaoqiu.Recent advances in fireworks algorithm[J].CAAI Transactions on Intelligent Systems,2014,9(01):515.[doi:10.3969/j.issn.1673-4785.201409010]
[8]顾大强,郑文钢.多移动机器人协同搬运技术综述[J].智能系统学报,2019,14(01):20.[doi:10.11992/tis.201801038]
　GU Daqiang,ZHENG Wengang.Technologies for cooperative transportation by multiple mobile robots[J].CAAI Transactions on Intelligent Systems,2019,14(01):20.[doi:10.11992/tis.201801038]
[9]李景灿,丁世飞.基于人工鱼群算法的孪生支持向量机[J].智能系统学报,2019,14(06):1121.[doi:10.11992/tis.201905025]
　LI Jingcan,DING Shifei.Twin support vector machine based on artificial fish swarm algorithm[J].CAAI Transactions on Intelligent Systems,2019,14(01):1121.[doi:10.11992/tis.201905025]
[10]邱华鑫,段海滨,范彦铭,等.鸽群交互模式切换模型及其同步性分析[J].智能系统学报,2020,15(2):334.[doi:10.11992/tis.201904052]
　QIU Huaxin,DUAN Haibin,FAN Yanming,et al.Pigeon flock interaction pattern switching model and its synchronization analysis[J].CAAI Transactions on Intelligent Systems,2020,15(01):334.[doi:10.11992/tis.201904052]

备注/Memo

备注/Memo:
收稿日期:2014-12-18;改回日期:。
基金项目:国家自然科学基金资助项目(61375119,61170057,60875080).
作者简介:丁科,男,1989年生,博士研究生,主要研究方向为群体智能、GPU通用计算、并行编程和机器学习;谭营,男,1964年生,教授,博士生导师,主要研究方向为计算智能、群体智能、机器学习、人工免疫系统、智能信息处理及信息安全应用。担任IJCIPT主编,IJSIR副主编,IEEE Trans on Cybernetics副主编等,IEEE Senior Member, IEEE CIS-ETTC委员,ICSI系列会议大会主席。主持国家“863”计划、国家自然科学基金、国际合作交流等科研项目30余项。获得2009年度国家自然科学二等奖,是中科院百人计划入选者。获国家发明专利授权3项,发表学术论文260余篇,出版专著5部。
通讯作者:谭营.E-mail:ytan@pku.edu.cn.
更新日期/Last Update: 2015-06-16