基于现场保存与恢复的双核冗余执行模型  

A Dual-Core Redundancy Execution Model Based on Context Saving and Recovery

作　　者：龚锐[1] 戴葵[1] 王志英[1] 

机构地区：[1]国防科技大学计算机学院,湖南长沙410073

出　　处：《计算机工程与科学》2009年第8期82-85,98,共5页Computer Engineering & Science

基　　金：国家973计划资助项目(2007CB310901);国家863计划资助项目(2007AA01Z101);国家自然科学基金资助项目(60773024)

摘　　要：本文提出了基于现场保存与恢复的双核冗余软错误恢复执行模型DCR。该执行模型在两个冗余的内核上执行相同的线程,并对store指令进行比较。本文对每个内核增加了硬件实现的现场保存与恢复机制,在检测到软错误以后可以恢复到上一现场保存点继续执行。实验结果表明,与传统的软错误恢复执行模型CRTR相比,DCR执行模型对核间通信带宽的需求降低了57.5%。在发生软错误的情况下,DCR能够恢复99.69%的软错误。A new approach to recovering soft errors is proposed in this paper. The dual-core redundancy based on context saving and recovery （DCR） executes two redundant threads of a given program on separate cores. The redundant cores are enhanced with hardware-implemented context saving and recovery, so that the soft errors can be recovered by re-execution from the last context saving point. The experimental results demonstrate that DCR can reduce 57. 5% inter-core communication bandwidth demand and recover 99. 69% soft errors compared to CRTR.

关 键 词：多核 执行模型 软错误恢复 现场保存 现场恢复 

分 类 号：TP302.8[自动化与计算机技术—计算机系统结构]