异构系统编程方法综述

摘要/Abstract

摘要：

由通用微处理器和专用加速器构成的异构并行系统比传统的同构系统显示出更高的性能和效能，目前已成为高性能计算领域的重要发展趋势。然而，编程问题成为限制异构系统发展的一大挑战，因此引起学术界和业界广泛关注。本文首先阐述了异构系统编程问题的来源；其次分析了近年来面向异构系统编程方法的研究工作，并对其进行分类；最后总结了异构系统编程方法未来的研究问题。

关键词: 异构系统, 编程模型

Abstract:

Heterogeneous parallel systems integrated with general purpose CPUs and domainspecific accelerators feature higher performance and energyefficiency than their homogeneous counterparts. Nowadays, heterogeneous parallel systems have become an important trend in the high performance computing field. However, the programming model problem becomes one of the most important challenges of heterogeneous systems, and thus attracts much attention from academia and industry. In this paper, we first discuss the source of the programming issue for heterogeneous systems. Then we introduce and classify the researches of the programming models for heterogeneous systems in recent years. Finally, the programming issues to be further studied for heterogeneous systems are discussed.

Key words: heterogeneous system;programming model

唐滔，杨学军. 异构系统编程方法综述[J]. J4, 2012, 34(3): 29-34.

TANG Tao,YANG Xuejun. A Survey of the Programming Methods for Heterogeneous Systems[J]. J4, 2012, 34(3): 29-34.

参考文献［31］

［1］	Hofstee H P. Power Efficient Processor Architecture and the Cell Processor［C］∥Proc of HPCA, 2005:258262.
［2］	Thakkar S, Huff T. The Internet Streaming SIMD Extensions［M］. Intel Technology Journal Q2, 1999,32(12):18.
［3］	Advanced Micro Devices, AMD CPU Roadmap［EB/OL］.［20101016］.http://developer.amd.com/assets/Develop_Brighton_Justin_Boggs1.pdf.
［4］	Fan Z, Qiu F, Kaufman A, et al. Gpu Cluster for High Performance Computing［C］∥Proc of the 2004 ACM/IEEE Conference on Supercomputing,2004:47.
［5］	Top500［EB/OL］.［20100317］.http://www.top500.org/lists/2010/11.
［6］	Kunzman D, Zheng G, Bohm E, et al. Charm++, Offload API, and the Cell Processor［C］∥Proc of the Workshop on Programing Models for Vbiquitous Paralelism,2006.
［7］	Lastovetsky A. Adaptive Parallel Computing on Heterogeneous Networks with MPC［J］. Parallel Computing, 2002,28(10):13691407.
［8］	Gaissaryan S, Lastovetsky A. An ANSI C Superset for Vector and Superscalar Computers and Its Retargetable Compiler［J］.The Journal of C Language Translation, 1994,5(3):183198.
［9］	Hou Qiming, Zhou Kun, Guo Baining. BSGP: BulkSynchronous GPU Programming［C］∥Proc of SIGGRAPH’08,2008.
［10］	AMD Brook［EB/OL］.［20080901］.http://ati.amd.com/technology/streamcomputing/AMDBrookplus.
［11］	Kirk D. NVIDIA Cuda Software and GPU Parallel Computing Architecture［C］∥Proc of the 6th International Symposium on Memory Management,2007:103104.
［12］	Open Computing Language［EB/OL］.［20090716］.http://www.khronos.org/.
［13］	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.
［14］	Buck I. Brook Spec v0.2［R］. TechReport, Stanford University, 2003.
［15］	Hou Qiming, Zhou Kun, Guo Baining. SPAP:A Programming Language for Heterogeneous ManyCore Systems［R］. Technical Report, 2010.
［16］	Ayguade E. A Proposal to Extend the OpenMP Tasking Model for Heterogeneous Architectures［C］∥Proc of the 5th International Workshop on OpenMP: Evolving OpenMP in an Age of Extreme Parallelism,2009:154167.
［17］	O’Brien K, O’Brien K, Sura Z, et al. 2008. Supporting OpenMP on Cell［J］. International Journal of Parallel Program,2008,36(3):289311.
［18］	Bellens P, Perez J M, Badia R M,et al. CellSs: A Programming Model for the Cell BE Architecture［C］∥Proc of the 2006 ACM/IEEE Conference on Supercomputing,2006.
［19］	Lee Seyong, Eigenmann R. OpenMPC: Extended OpenMP Programming and Tuning for GPUs［C］∥Proc of the 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis. IEEE Computer Society, 2010:111.
［20］	Lee Seyong, Min SeungJai, Eigenmann R. OpenMP to GPGPU: A Compiler Framework for Automatic Translation and Optimization［C］∥Proc of the 14th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming,2009:101110.
［21］	Wang P H, Collins J D, Chinya G N, et al. EXOCHI: Architecture and Programming Environment for a Heterogeneous MultiCore Multithreaded System［C］∥Proc of the 2007 ACM SIGPLAN Conference on Programming Language Design and Implementation, 2007:156166.
［22］	Dolbeau R, Bihan S, Bodin F. HMPP: A Hybrid MultiCore Parallel Programming Environment［C］∥Proc of the Workshop on General Purpose Processing on Graphics Processing Units, 2007.
［23］	The Portland Group. PGI Fortran and C Accelerator Compilers and Programming Model Technology Preview［R］.2008.
［24］	Limberg T, Ristau B, Fettweis G. A RealTime Programming Model for Heterogeneous MPSoCs［C］∥Proc of the 8th International Workshop on Embedded Computer Systems: Architectures, Modeling, and Simulation, 2008:7584.
［25］	Saha B, Zhou Xiaocheng, Chen Hu, et al. Programming Model for a Heterogeneous x86 Platform［C］∥Proc of the 2009 ACM SIGPLAN Conference on Programming Language Design and Implementation, 2009:431440.
［26］	Seiler L, Carmean D, Sprangle E, et al. Larrabee: A ManyCore x86 Architecture for Visual Computing［J］. ACM Transactions on Graphics, 2008,27(3):115.
［27］	Linderman M D, Collins J D, Wang H, et al. Merge: A Programming Model for Heterogeneous MultiCore Systems［J］. SIGARCH Comput Archit News, 2008,42(2):287296.
［28］	Dean J, Ghemawat S. MapReduce: Simplified Data Processing on Large Clusters［J］. Commun ACM, 2008,51(1):107113.
［29］	Lastovetsky A, Reddy R. HeteroMPI: Towards a MessagePassing Library for Heterogeneous Networks of Computers［J］. J Parallel Distrib Comput, 2006,66(2):197220.
［30］	Peisert S, Baden S. A Programming Model for Automated Decomposition on Heterogeneous Clusters of Multiprocessors［R］. Technical Report, UCSD CSE, 2001.
［31］	Intel. Next Generation Processors, in Manufacturing, Chip design Expertise Driving Innovation and Integration, Historic Change to Computers［R］. 2009.

[1]	康宇晗, 时洋, 陈照云, 文梅. 面向迈创+MatrixZone异构系统的深度学习编程框架[J]. 计算机工程与科学, 2023, 45(07): 1149-1158.
[2]	陈香兰, 李曦, 汪超, 周学海. 实时机模型及时间语义指令集研究[J]. 计算机工程与科学, 2021, 43(04): 571-578.
[3]	王耀华, 郭阳. 面向HPC的高性能微处理器研究进展[J]. 计算机工程与科学, 2020, 42(10高性能专刊): 1742-1748.
[4]	张诗情，杨耀华，沈立，王志英. 通过部分页迁移实现CPU-GPU高效透明的数据通信[J]. 计算机工程与科学, 2019, 41(07): 1168-1175.
[5]	李大普1，陈光喜1,李肯立2. 异构系统中基于可用性的粒子群任务调度算法[J]. J4, 2012, 34(6): 44-49.
[6]	王桂彬. 基于硬件性能计数器的GPU功耗预测模型[J]. J4, 2012, 34(3): 46-50.
[7]	李成华，张新访，金海，向文. MapReduce：新型的分布式并行计算编程模型[J]. J4, 2011, 33(3): 129-135.
[8]	贾佳. 异构系统的异步应用级Checkpointing技术[J]. J4, 2011, 33(11): 54-59.
[9]	林一松,唐玉华,唐〓滔. GPGPU技术研究与发展[J]. J4, 2011, 33(10): 85-92.
[10]	朱海，李仁发，陈宇，刘彦. 支持过程级动态软硬件划分的RSoC设计与实现[J]. J4, 2010, 32(4): 131-135.
[11]	朱建军杨清冯建湘. 网格环境下动态编程模型研究[J]. J4, 2008, 30(6): 155-158.
[12]	王桂彬唐沛蓉唐玉华. 线性方程组迭代法在流处理器上的映射与分析[J]. J4, 2007, 29(7): 88-90.
[13]	张金[1] 易铭[1] 陈卓宁[2]. 基于组件技术的跨语言异构系统集成研究[J]. J4, 2004, 26(8): 106-109.