计算机科学 ›› 2013, Vol. 40 ›› Issue (8): 34-37.

• 综述 • 上一篇    下一篇

WPP-L2:多核处理器中共享Cache低功耗路预测算法

方娟,郭媚,杜文娟   

  1. 北京工业大学计算机学院 北京100124;北京工业大学计算机学院 北京100124;北京工业大学计算机学院 北京100124
  • 出版日期:2018-11-16 发布日期:2018-11-16
  • 基金资助:
    本文受国家自然科学基金项目(61202076),北京市教委科技计划面上项目(KM201210005022),北京市中青年骨干人才培养项目(007000543112512)资助

WPP-L2:A Way-predicting Algorithm for Shared L2Cache of Multi-core Processors

FANG Juan,GUO Mei and DU Wen-juan   

  • Online:2018-11-16 Published:2018-11-16

摘要: 针对片上多核处理器下的二级共享Cache的能耗问题提出了基于Cache划分的路预测Cache结构WPP-L2,该结构首先对共享Cache进行公平性划分,然后采用路预测的方法降低了预测命中和失效时各自的能耗开销。实验表明,在基本保持多核处理器性能的同时,8核处理器系统下WPP-L2Cache比基于路预测的L2Cache的能耗延迟乘积EDP(Energy Delay Product)平均下降24.7%,比传统的L2Cache的EDP平均下降66.1%,极大地降低了L2Cache功耗。

关键词: 多核处理器,低功耗,路预测

Abstract: This paper proposed a way-predicting L2cache mechanism based on cache partitioning for the problem of multi-core processors power,which is WPP-L2.WPP-L2partitions the shared cache for fairness,and then uses cache way-predicting techniques to reduce the energy consumption when the prediction hits and misses.The results turn out,WPP-L2cache could reduce EDP(Energy Delay Product)by 24.7% compared with the way-predicting L2cache,and reduce EDP by 66.1% compared with the traditional L2cache under the eight-core processors,which lowers the energy consumption of L2Cache and maintains the performance at the same time.

Key words: Multi-core processors,Low-power,Way-predicting

[1] Ijaykr V,Ishman N,Kandemir M,et al.Energy-driven integra-ted hardware-software optimizations using simple-power[A]∥Proceedings of the 27th Annual International Symposium on Computer Architecture,2000[C].California:IEEE,2000:95-106
[2] Lu Jun-yang,Guo Yao.Energy-Aware Fixed-Priority Multi-core Scheduling for Real-Time Systems [A]∥The 17th IEEE International Conference on Embedded and Real-Time Computing Systems and Applications,2011[C].California:IEEE,2011:277-281
[3] Wang Wei-xun,Mishra P,Ranka S.Dynamic Cache Reconfiguration and Partitioning for Energy Optimization in Real-Time Multi-core Systems[A]∥Design Automation Conference.New Jersey,2011[C].IEEE,2011:948-953
[4] Reddy R,Petrov P.Cache partitioning for energy-efficient andinterference-free embedded multitasking[J].ACM TECS,2010,9(3):1-35
[5] Inoue K,Ishihara T,Murakami K.Way-Predicting Set-associative Cache for High Performance and Low Energy Consumption[A]∥Proceedings on Low Power Electronics and Design.New Jersey,1999[C].IEEE,1999:273-275
[6] Chen H-C,Chiang J-S.Low-power Way-predicting Cache Using Valid-bit Predecision for Parallel Architectures[A]∥Procee-dings of the 19th International Conference on Advanced Information Networking and Applications,2005[C].California:IEEE,2005:28-30
[7] Chung C-M,Kim J.Low-power L2cache design for multi-core processors [J].Electronics Letters 29th,2010,46(9):618-120
[8] Suh G E,Rudolph L,Devadas S.Dynamic Partitioning of Shared Cache Memory [J].Journal of Supercomputing,2004,28(1):7-26
[9] 所光,杨学军.面向多线程多道程序的加权共享Cache划分[J].计算机学报,2008,31(11):1938-1947
[10] Muralidhara S P,Kandemir M,Raghavan P.Intra-ApplicationCache Partitioning[A]∥IEEE International Parallel & Distri-buted Processing Symposium,2010[C].New Jersey:IEEE,2010:1-12
[11] Kim S,Chandra D,Solihin Y.Fair cache sharing and partitioning in a chip multiprocessor architecture[A]∥Proceedings of 13th International Conference on PACT,2004[C].California:IEEE,2004:111-122
[12] Hsu L R,Reinhard S K,Iyer R.Communist utilitarian,and capitalist cache policies on CMPs:Caches as a shared resource[A]∥Proc of PACT-15,2006[C].California:IEEE,2006:13-22
[13] Gabor R,Weiss S,Mendelson A.Fairness and throughput inswitch on event multithreading[A]∥39th IEEE/ACM Inter-national Symposium on Microarchitecture,2006[C].California:IEEE,2006:149-160
[14] Iyer,Ravi.A framework for enabling QoS in shared caches of CMP platforms[A]∥Proceedings of ICS-18,2004[C].USA:ACM,2004:257-266
[15] 隋秀峰,吴俊敏,等.ARP:同时多线程处理器中共享Cache自适应运行划分机制[J].计算机研究与发展,2008,45(7):1269-1277

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!