Journal of Systems Architecture ( IF 4.5 ) Pub Date : 2020-11-28 , DOI: 10.1016/j.sysarc.2020.101953 Sanjay Moulik , Zinea Das , Rajesh Devaraj , Shounak Chakraborty
Over the years, the nature of processing platforms is witnessing a significant shift in most of the battery supported real-time systems, which now support a combination of specialized multicores to meet the demands of modern applications. Devising energy-efficient schedulers has become a critical issue for such kinds of devices. Hence, this research presents a low-overhead heuristic strategy named SEAMERS, for DVFS based energy-aware scheduling for a set of real-time periodic tasks on a heterogeneous multicore platform. The presented strategy operates in four phases, namely Deadline Partitioning, Core Clustering, Task Allocation and Energy-Aware Scheduling. Our experimental analysis shows that the presented strategy improves upon the state-of-the-art in terms of energy savings (16% to 47% on average) and enables significant improvement in resource utilization.
中文翻译:
矫直机:一个小号EMI-分区ê nergy-甲洁具异构调度中号ulticor Ë ř EAL-时间š ystems
多年来,处理平台的性质正在见证大多数电池支持的实时系统的重大转变,这些实时系统现在支持组合专用多核以满足现代应用程序的需求。设计节能调度器已成为此类设备的关键问题。因此,本研究提出了一种名为SEAMERS的低开销启发式策略,用于基于DVFS的能源感知调度,用于异构多核平台上的一组实时周期性任务。提出的策略分四个阶段进行,分别是截止期限划分,核心集群,任务分配和能源感知调度。我们的实验分析表明,所提出的策略在节能方面(最先进的平均水平为16%至47%)进行了改进,并且可以显着提高资源利用率。