针对变电压处理器的电池感知任务调度算法  

Battery-aware task scheduling algorithm on DVS enabled processors

作　　者：解玉凤[1] 刘雷波[1] 戴锐[1] 魏少军[1] 

机构地区：[1]清华大学微电子学研究所,北京100084

出　　处：《清华大学学报（自然科学版）》2008年第1期132-136,共5页Journal of Tsinghua University(Science and Technology)

基　　金：国家自然科学基金资助项目(60676012;60506007)

摘　　要：为了解决变电压处理器上以延长电池使用时间为目标的任务调度问题,提出一种基于控制步的电池感知任务调度算法。实验结果表明:在电池感知优化最有效率的范围内,该算法跟传统的简化算法相比,在计算复杂度相当的情况下,电池代价降低6.1%,实际消耗电荷降低9.4%,而跟迭代算法相比,电池代价下降2.5%,实际消耗电荷略微增加,但是算法的运行时间平均下降99.7%。该算法能有效地降低系统能耗和延长电池使用时间。Task scheduling on variable voltage processor is used to extend battery lifetime. This paper describes a control-step-based battery-aware task scheduling algorithm to increase battery lifetime. Tests show that, within the most efficient task granularity range for battery-aware optimization, the algorithm has comparable CPU runtime to traditional simplified algorithm with a 6.1% reduction on battery cost and a 9. 4% reduction on the actual charge consumption. Compared with the iterative algorithm, the algorithm reduces battery cost by 2. 5% with a small increment on the actual charge consumption, while reduces CPU runtime by an average of 99.7%. Therefore, the algorithm efficiently reduces energy consumption and extends battery lifetime.

关 键 词：数字系统 实时系统 电池感知 任务调度 动态电压调整 

分 类 号：TN47[电子电信—微电子学与固体电子学] TP316[自动化与计算机技术—计算机软件与理论]