期刊文献+

一种基于ESVW技术的新型载入存储队列设计

A Novel Design of Load Store Queue with Enhanced Store Vulnerability Window
下载PDF
导出
摘要 在超标量处理器设计中,SVW技术通过降低重新执行的载入指令数目提高系统性能.本文提出一种基于ESVW技术的新型载入存储队列设计方案,使得SVW结构中的SSBF不仅记录SSN值,同时记录存储地址和数值,重新执行的载入指令就可以从中获取数据而不必访问缓存,进一步降低指令的重新执行率.实验结果显示,最优情况下,超过99%的载入指令免于二次访问缓存,系统性能提高约2%. In superscalar process design,Store Vulnerability Window technique improves system performance by reduce load re-execute rate.In this paper,we present a novel design of load store queue with Enhanced SVW.We observe that if store sequence Bloom filter(SSBF)unit in SVW is modified to keep not only store sequence number(SSN)but also complete memory addresses and data values,trailing loads can get data from SSBF rather than cache,therefore data cache access rate will be further reduced.Experiments show that,in the best case,over 99% re-execute loads are kept from memory re-access,and system performance is improved by 2%.
出处 《微电子学与计算机》 CSCD 北大核心 2013年第7期20-23,共4页 Microelectronics & Computer
基金 国家"八六三"计划项目(2009AA011705)
关键词 超标量处理器 LSQ SVW SSBF superscalar processor LSQ SVW SSBF
  • 相关文献

参考文献6

  • 1Chrysos G, Emer J. Memory dependence prediction u- sing store sets [C]/// InISCA- 25. Spain: Barcelona, 1998:142-153.
  • 2Lei Jin, Sangyeun Cho. Macro data load., an efficient mechanism for enhancing loaded data reuse[J]. Com- puters, IEEE Transactions on, 2011, 60 (4) : 526 - 537.
  • 3Gandhi A, Akkary H, Rajwar R, et al. Scalable load and store processing in latency tolerant processors[C] // Proc. 32nd International Symposium on ComputerArchitecture USA: Madison, Wisconsin, 2005:446 - 457.
  • 4Gharachorloo K, Gupta A, Hennessy J. Two tech- niques to enhance the performance of memory consis- tency models[C]// Proc. of the International Confer- ence on Parallel Processing. USA: Austion, 1991:355 - 364.
  • 5Cain H, Lipasti M. Memory ordering: A value based definition[C]/// Proc. 31st International Symposium on Computer Architecture. Gernany: Munich, 2004:90 - 101.
  • 6Roth A. Store vulnerability window (SVW) : re-exe- cution filtering for enhanced load optimization[C]/// Proc. 32nd International Symposium on Computer Ar- chitecture. USA: Madison, Wadison, Wisconsin, 2005 : 458 - 468.

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