期刊文献+

面向众核多级访存资源的静态数据布局优化模型 被引量:3

A STATIC DATA PLACEMENT OPTIMISATION MODEL ORIENTED TOWARDS MULTI-CORE HIERARCHICAL ACCESSIBLE RESOURCES
下载PDF
导出
摘要 受功耗、面积的限制,高性能众核处理器倾向于将片上SRAM组织成SPM这种非cache形式,与片外主存构成多级存储架构。这种存储架构需要软件显示管理应用程序中的数据存储和传输。为此,提出了一种众核多级访存资源的静态数据布局优化模型。该模型首先选择应用程序中可进行访存优化的数组变量,对这些变量进行优化收益的评估,然后建立一个类0-1背包优化问题的目标模型,最后针对该优化模型提出了一个实用的近似算法。实验结果验证了该模型的有效性。 Constrained by power consumption and surface area,high-performance multi-core processor manufacturers tend to arrange on-chip SRAM into a non-cache form such as SPM to constitute a hierarchical storage architecture together with standalone memories.Such storage architecture requires data storage and transfer functions of the software display management application.The article puts forward a static data placement and optimisation model oriented towards multi-core hierarchical accessible resources.The model first determines array variables available for access optimisation in an application,draws assessments on benefits of optimisation with those variables,then establishes an object model similar to 0-1 bag optimisation,and finally concludes a practical approximate algorithm aiming at the optimised model.Experiments have validated the method.
出处 《计算机应用与软件》 CSCD 2011年第7期53-56,共4页 Computer Applications and Software
基金 国家重点基础研究发展计划资助项目(2007CB310900)
关键词 SPM 多级访存资源 数据布局 数据传输 SPM Hierarchical accessible resources Data placement Data transfer
  • 相关文献

参考文献7

  • 1Balart J, Gonzalez M, Martorell X, et al. A novel asynchronous software cache implementation for the Cell-BE Processor[ C ]// Proceedings of 'the 2007 Workshop on Languages and Compilers for Parallel Compu- ting. Urbana, Illinois ,2007 : 125 - 140.
  • 2Silberstein M, Schuster A, Geiger D, et al. Efficient computation of sum- product s on GPUs through software-managed cache[ C]// Proceedings of t he 22nd ACM International Conference on Supercomputing. Island of Kos, Greece,2008:309 - 318.
  • 3Tong Chen, Zehra Sura, Kathryn O'Brien, et 8.1. Optimizing the use of static buffers for DMA on a Cell chip[ C]//19th International Work-shop on Languages and Compilers for Parallel Computing, November 2 - 4,2006, New Orleans, Louisiana.
  • 4Knight T J, Park J Y, Ren M, et al. Compilation for explicitly managed memory hierarchies [ C ]// Proceedings of the 12th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming ( PpoPP' 07 ) . New York, N Y, USA ,2007:226 - 236.
  • 5Houston M, Park J Y, Ren M, et al. A portable runtime interface for multi-level memory hierarchies [ C ]// Proceedings of the 13th ACM S/GPLAN Symposium on Principles and Praetiee of Parallel Programming ( PPOPP 2008 ). Salt Lake City, Utah, USA ,2008 : 143 - 152.
  • 6冯国富,董小社,胡冰,王旭昊,王恩东.一种支持多种访存技术的CBEA片上多核MPI并行编程模型[J].计算机学报,2008,31(11):1965-1974. 被引量:6
  • 7冯国富,董小社,丁彦飞,王旭昊.面向Cell宽带引擎架构的异构多核访存技术[J].西安交通大学学报,2009,43(2):1-5. 被引量:10

二级参考文献8

  • 1袁伟,张云泉,孙家昶,李玉成.国产万亿次机群系统NPB性能测试分析[J].计算机研究与发展,2005,42(6):1079-1084. 被引量:13
  • 2胡伟武,赵继业,钟石强,杨旭,Elio Guidetti,吴永强.Implementing a 1GHz Four-Issue Out-of-Order Execution Microprocessor in a Standard Cell ASIC Methodology[J].Journal of Computer Science & Technology,2007,22(1):1-14. 被引量:14
  • 3GSCHWlND M. Chip multiprocessing and the cell broadband engine [C]//Proceedings of ACM Compu- ting Frontiers. New York, USA: ACM Press, 2006: 1-8.
  • 4OHARA M,INOUE H,SOHDA Y, et al. MPI microtask for programming the cell broadband engine processor [J]. IBM Systems Journal, 2006, 45 (1) : 85- 102.
  • 5BELLENS P.PEREZ J M.BADIA R M,et al.CellSa:a programming model for the cell BE architecture[C]//Proceedings of the ACM/IEEE SC 2006 Conference on High Performance Networking and Computing.New York.USA:ACM Press,2006:86.
  • 6EICHENBERGER A E.O'BREN J R.O'BRIEN K M,et al.Using advanced complier technology to exploit the porformance of the cell broadband cngine[trademark]architecture[J].IBM Systems Journal,2006,45(1):59-84.
  • 7VAN DER SPOEI. D, LINDABI. E, HESS B, et al. Gromacs: fast, flexible, and free [J]. Journal of Computational Chemistry, 2005,26(16) : 1701-1718.
  • 8胡伟武,唐志敏.龙芯1号处理器结构设计[J].计算机学报,2003,26(4):385-396. 被引量:53

共引文献14

同被引文献21

  • 1IBM Microelectronics: Cell Broadband Engine[ OL]. http://www-O1. ibm. com/. chips/techlib/techlib, nsf/products/Cell Broadband En- gine.
  • 2Maciej C. Cell Programming: PEACE Workshop [ C ]//New Languages & Future Technology Prototypes, March 1-2, LRZ, Germany ,2010.
  • 3Perez J M, Bellens P,Badia R M, et al. CellSs Making it easier to program the Cell Broadband Engine[ J]. IBM J. RES. &DEV. 2007, 51(5).
  • 4Tallada M G. OpenMP on the IBM Cell BE[ C]//15th Meeting of Sci- comP Barcelona Supercomputing Center(BSC) ,2009,18- 22.
  • 5IBM. Accelerated Library Framework for Cell Broadband Engine Pro- grammer' s Guide and API Reference. Software Development kit for Muhicore Acceleration Version 3. 0 [ OL]. http://moss, csc. ncsu. edu/- mueller/cluster/ps3/SDK3.0/docs/lib/ALF_Prog_Guide_API -v3.0.pdf.
  • 6Dakar T. SHOC:The Scalable HeterOgeneous Computing Benchmark Suite [ R ]. Future Technologies Group, Oak Ridge National Laboratory, Novermber 2011.
  • 7胡志刚,石金锋,蒋湘涛.针对能耗热点的SPM静态分配管理策略[J].计算机工程与应用,2010,46(3):58-61. 被引量:4
  • 8李建江,刘珍珍,王珏.基于IBM Cell多核平台的OpenMP数组私有化技术研究[J].计算机研究与发展,2010,47(8):1434-1441. 被引量:3
  • 9刘勇,陆林生,何王全.一种简便的栈式片上内存动态管理方法[J].计算机工程与科学,2010,32(9):111-114. 被引量:1
  • 10孙守航,杨灿群.Cell处理器上软件缓存的设计与实现[J].计算机工程,2011,37(2):45-47. 被引量:3

引证文献3

二级引证文献1

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