期刊文献+

异构众核系统及其编程模型与性能优化技术研究综述 被引量:13

The Feature,Programming Model and Performance Optimization Strategy of Heterogeneous Many-Core System: A Review
下载PDF
导出
摘要 异构众核系统已成为当前高性能计算领域重要的发展趋势.针对异构众核系统,从架构、编程、所支持的应用三方面分析对比当前不同异构系统的特点,揭示了异构系统的发展趋势及异构系统相对于传统多核并行系统的优势;然后从编程模型和性能优化方面分析了异构系统存在的问题和面临的挑战,以及国内外研究现状,结合当前研究存在的问题和难点,探讨了该领域进一步深入的研究方向;同时对两种典型的异构众核系统CPU+GPU和CPU+MIC进行不同应用类型的Benchmark测试,验证了两种异构系统不同的应用特点,为用户选择具体异构系统提供参考,在此基础上提出将两种众核处理器(GPU和MIC)结合在一个计算节点内构成新型混合异构系统;该新型混合异构系统可以利用两种众核处理器不同的处理优势,协同处理具有不同应用特点的复杂应用,同时分析了在该混合异构系统下必须要研究和解决的关键问题;最后对异构众核系统面临的挑战和进一步的研究方向进行了总结和展望. The heterogeneous many-core system has emerged as a promising developing trend in the high performance computing area. In this paper,w e first revealed the developing trend and dominant position of the heterogeneous systems via analyzing their architectures,programming and application characteristics. Secondly,w e discussed the programming model and performance optimization of current heterogeneous systems,and summarized the relative research trends. Thirdly,w e verified the different application behaviors of the GPU and M IC heterogeneous system by running different types of Benchmark,w hich provides the reference for user to select the specific heterogeneous computing platform and,the basis of composing the hybrid heterogeneous system w hich combines the tw o types of many-core processor( GPU and M IC) into an individual computing node. This hybrid heterogeneous system can fully exploit the computing potential of the tw o types of manycore coprocessors to coordinate handling the complex application w ith different application behaviors. Finally,some open issues and future directions in the heterogeneous system were view ed.
出处 《电子学报》 EI CAS CSCD 北大核心 2015年第1期111-119,共9页 Acta Electronica Sinica
基金 国家自然科学基金(No.61173039 No.61202041) 国家863高技术研究发展计划(No.2012AA010904 No.2012AA01A306) 国家科技支撑计划(No.2011BAH04B03)
关键词 异构众核系统 高性能计算 异构计算 编程模型 性能优化 heterogeneous many-core system high-performance computing heterogeneous computing programming model performance optimization
  • 相关文献

参考文献4

二级参考文献105

  • 1吴恩华,柳有权.基于图形处理器(GPU)的通用计算[J].计算机辅助设计与图形学学报,2004,16(5):601-612. 被引量:225
  • 2吴恩华.图形处理器用于通用计算的技术、现状及其挑战[J].软件学报,2004,15(10):1493-1504. 被引量:141
  • 3NVIDIA Corporation. NVIDIA CUDA programming guide [EB/OL][2010-07-15]. http://www, nvidia. com/object/euda_ homenew. html.
  • 4YANG Yang, RAART K V, CASANOVA H. Multi round algorithms for scheduling divisible loads [J].IEEE Transactions on Parallel and Distributed Sys tems, 2005,16(11): 1092-1102.
  • 5TAO Yongcai,JIN Hai,WU Song,et al.Adaptive multi-round scheduling strategy for divisible workloads in grid environments[C] // Proceedings of the 23rd International Conference on Information Networking.New York,USA:ACM,2009:260-264.
  • 6SHET G A,SADAYAPPAN P,BERNHOLDT E D,et al.A framework for characterizing overlap of communication and computation in parallel applications[J].Cluster Computing,2008,11(1):75-90.
  • 7ANTHONY D,LORI P,MARTIN S.MPI-aware compiler optimizations for improving communicationcomputation overlap[C] //Proceedings of the 23th International Conference on Supercomputing New York,USA:ACM,2009:316-325.
  • 8YANG Yi, XIANG Ping, KONG Jingfei, et al. A GPGPU compiler for memory optimization and parallelism management[C]//Proceedings of the 2010 ACM SIGPLAN Conference on Programming Language Design and Implementation. New York, USA: ACM, 2010: 86-97.
  • 9MALONY A D, BIERSDORFF S, MAYANGLAMBAM S. An experimental approach to performance measurement of heterogeneous parallel applications using CUDA[C]//Proeeedings of the 24th ACM International Conference on Supercomputing. New York, USA: ACM, 2010; 127-136.
  • 10BAGHSORKHI S S, DELAHAYE M, PATEL S J, et al. An adaptive performance modeling tool for GPU architectures[C]//Proceedings of the 15th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming. New York, USA: ACM, 2010. 105 -114.

共引文献65

同被引文献59

引证文献13

二级引证文献31

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部