期刊文献+
任意字段
题名或关键词
题名
关键词
文摘
作者
第一作者
机构
刊名
分类号
参考文献
作者简介
基金资助
栏目信息

帮助线程预取技术研究综述 被引量：3

Survey of Helper Thread Prefetching
下载PDF
导出
摘要 帮助线程预取是当前多核平台提高非规则数据密集应用预取效果性能的关键技术之一,近年来已成为国内外的研究热点。针对非规则数据密集应用访存规律的非连续局部性特征,帮助线程预取技术利用CMP平台的最后一级共享缓存(LLC)将应用的非连续局部性转换为瞬时的连续时空局部性(即时局部性),从而达到通过线程级数据预取提高程序性能的目的。归纳了帮助线程预取技术的分类,概括和比较了不同帮助线程实现技术的优势和局限性,深入分析和探讨了现有的几种典型帮助线程技术的预取控制策略。最后从帮助线程实时控制、参数动态选取和优化方面指出了帮助线程预取技术的研究方向。 Helper thread prefetching is one of the key techniques to improve the prefetch effect of non-irregular data in- tensive applications. It has become a hot research topic at all over the world in recent years. Aiming at the memory ac- cess characteristic of discontinuous locality of non-irregular data intensive applications, helper threading could effectively convert discontinuous locality into continuous-instant spatial or temporal locality by using the shared LLC of CMP plat- form. And as a result, the application＇s performance can be improved. In this paper, the classification of helper thread prefetching techniques was summarized from the perspective of implementation method. The limitation and superiority of different types of prefetching were compared and surveyed. The current helper thread prefetching control policy was systematically analyzed and compared. Finally, several major issues and research directions of helper thread prefetching for further exploration were also pointed out.
作者 张建勋 古志民
机构地区 北京理工大学计算机学院 天津中医药大学网络中心
出处 《计算机科学》 CSCD 北大核心 2013年第7期19-23,39,共6页 Computer Science
基金 国家自然科学基金项目(61070029)资助
关键词 帮助线程 数据预取 CMP(Chip Multi-Processor)平台 非规则数据密集应用 Helper thread, Data prefetching, CMP platfotto, Non-irregular data intensive application
分类号 TP303 [自动化与计算机技术—计算机系统结构]
  • 相关文献

参考文献33

  • 1Bryant R E. Data-Intensive Supercomputing; The case for DISC [EB/OL]. http://www, cs. cmu. edu/bryant,2012-12-13.
  • 2谭光明.非规则计算中的局部性和并行性[D].北京:中国科学院计算技术研究所,2008.
  • 3Annavaram M, Patel J M, Davidson E S. Data pretetchmg by de- pendence graph pre-computation [ A] // Proceedings of the 28th Annual International Symposium on Computer Architecture (Gotehorg, Sweden),2001[C-]. New YorkACM,2001:52-61.
  • 4Collins J D, Tullsen D M,Wang H, et al. Dynamic speculative preeomputation[A].//roceedings of the 34th International Symposium on Microarehitecture (Austin, Tex. ), 2001 [C]. New York: ACM, 2001 : 306-317.
  • 5Collins J D,Wang H,Tullsen D M,et al. Speculative preeompu- tation: Long-range prefetching of delinquent loads[A] ff Pro- ceedings of the 28th Annual International Symposium on Com- puter Architecture (Goteborg, Sweden), 2001[C]. New York: ACM, 2001 : 14-25.
  • 6Liao S S W,Wang P H,Wang H,et a]. Post-pass binary adapta- tion for software-based speculative precomputation[]//Pro- ceedings of the ACM SIGPLANConference on Programming Language Design and Implementation (Berlin, Germany), 2002 [C]. New York: ACM, 2002 : 117-128.
  • 7Kim D , Yeung D. Design and evaluation of compiler algorithms for pre-execution[A] // Proceedings of the 10th International Conference on Architectural Support for Programming Langua- ges and Operating Systems (San Jose, Calif. ), 2002 [C]. New York: ACM, 2002 : 159-170.
  • 8Luk C-K. Tolerating Memory Latency through software-con- trolled pre-exeeution in simultaneous multithreading processors [A]//Proceedings of the 28th Annual International Symposium on Computer Architecture (Goteborg, Sweden), 2001 EC]. New York: ACM, 2001 : 40-51.
  • 9Moshovos A, Pnevmatikatos D N, Baniasadi A. Slice-processors: An implementation of operation-based prediction-A]//Procee- dings of the International Conference on Supercomputing (Sor- rento, Italy), 2001[C]. New York ACM, 2001 : 321-334.
  • 10Roth A, Sohi G S. Speculative data-driven multithreading [A]// Proceedings of the 7th International Conference on High Per- formance Computer Architecture (Monterrey, Mexico), 9.001 [C]. Los Alamitos, Calif: IEEE Computer Society Press, 2001: 191-202.

二级参考文献1

同被引文献13

引证文献3

二级引证文献2

计算机科学
2013年 第7期

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