嵌入式多核多任务实时DVFS调度方法仿真  

Simulation of Embedded Multi-Core Multi-task Real-Time DVFS Scheduling Method

作　　者：王晓莹[1] 张仲雯 何海生[1] WANG Xiao-ying;ZHANG Zhong-wen;HE Hai-sheng(Guilin University of Electronic Technology,School of Computer Engineering,Beihai Guangxi 536000,China)

机构地区：[1]桂林电子科技大学计算机工程学院,广西北海536000

出　　处：《计算机仿真》2023年第8期500-504,共5页Computer Simulation

摘　　要：嵌入式设备在调度多核多任务时,由于引入的处理器数量增多,导致调度效率低,以及能耗大等问题。基于此,提出基于实时动态电压频率调整(Dynamic voltage and frequency scaling,DVFS)的嵌入式多核多任务调度算法。构建嵌入式多核多任务能耗模型,得到不同情况下设备的能量消耗情况;建立多核多任务调度数学模型,在策略集合中找到合适的调度策略,保持时间值始终为最小;利用多核多任务调度算法机制,通过计算嵌入式设备中各个处理核的潜在负载能力,找出与预期负载最接近的处理核,并将任务分配到该任务核中,实现多核多任务的合理调度。实验结果表明,所提方法的调度效率高,且能量消耗低于3.5mW。In order to solve low scheduling efficiency and high energy consumption of multi-core and multi-task scheduling,this article presented an embedded multi-core multi-task scheduling algorithm based on real-time dynamic voltage and frequency scaling(DVFS).Firstly,we built a model of energy consumption of embedded multi-core and multi-task,thus getting the energy consumption of equipment under different conditions.Then,we constructed a mathematical model for multi-core and multi-task scheduling to find the appropriate scheduling policy from the policy set,so that the time value could be always kept to the minimum.Moreover,we used the multi-core and multi-task scheduling algorithm to calculate the potential load capacity of each processing core in the embedded device and find out the core closest to the expected load,and then assigned the tasks to the task core.Finally,we achieved reasonable scheduling.Experimental results show that the scheduling efficiency of the proposed method is high,and the energy consumption is less than 3.5 mW.

关 键 词：实时动态电压频率调整算法 多核多任务调度 能量消耗 任务核 潜在负载处理能力 

分 类 号：TP384.6[自动化与计算机技术—计算机系统结构]