常带宽服务器混合任务低功耗调度算法

发布时间：2018-04-02 02:08

本文选题：动态电压调节　切入点：动态功耗管理　出处：《计算机研究与发展》2015年09期

【摘要】：针对包含有截止期限限制的周期任务和有响应时间要求的非周期任务的实时系统混合任务集,提出常带宽服务器混合任务低功耗调度算法(constant bandwidth server mix task low power scheduling algorithm,CBSMTLPSA).该算法是2阶段调度算法,并且结合了动态电压调节(dynamic voltage scaling,DVS)技术和动态功耗管理(dynamic power management,DPM)技术.离线阶段确定任务的离线速度,充分利用处理器的资源;在线阶段通过回收周期任务提早完成的空闲时间以及服务器产生的空闲时间,利用DVS技术调节处理器的运行速度,并且当处理器处于空闲状态时,判断是否使用DPM技术以达到进一步降低能耗的目的.仿真实验表明所提出的CBSMTLPSA算法比CBS/DRA-W(constant bandwidth server for dynamic reclaim algorithm base workload)算法节约6.02%~34.14%的能耗;CBSMTLPSA算法的能耗与非周期任务的响应时间的乘积比CBS/DRA-W算法低5.86%~34.06%.
[Abstract]:For a real-time system hybrid task set containing periodic tasks with deadlines and aperiodic tasks with response time requirements, This paper presents a low power bandwidth server mix task low power scheduling algorithm for constant bandwidth servers, which is a two-stage scheduling algorithm. In addition, dynamic voltage scaling voltage (DVS) technology and dynamic power management (DPM) technology are combined to determine the off-line speed of the task in the off-line phase and make full use of the processor resources. The online phase uses DVS technology to adjust the speed of the processor, and when the processor is idle, the idle time of the recovery cycle task is completed earlier and the idle time generated by the server is used to adjust the speed of the processor. The simulation results show that the proposed CBSMTLPSA algorithm saves 6.02% of the energy consumption compared with the CBS/DRA-W(constant bandwidth server for dynamic reclaim algorithm base workload algorithm, and the response time of the aperiodic task and the energy consumption of the CBS/DRA-W(constant bandwidth server dynamic reclaim algorithm base workload algorithm is less than that of the CBS/DRA-W(constant bandwidth server for dynamic reclaim algorithm base workload algorithm, which is 6.02% or 34.14% less than that of the CBS/DRA-W(constant bandwidth server dynamic reclaim algorithm base workload. The product of CBS/DRA-W algorithm is 5.86% lower than that of CBS/DRA-W algorithm.
【作者单位】：中国科学院沈阳计算技术研究所;中国科学院大学;
【基金】：“核高基”国家科技重大专项基金项目(2012ZX01029001-002)
【分类号】：TP368.5;TP301.6

【参考文献】