基于离散粒子群优化的MPSoC节能调度算法  

MPSo C Energy Saving Scheduling Algorithm Based on Discrete Particle Swarm Optimization

作　　者：于国龙[1] 崔忠伟[1] 左羽[1] 

机构地区：[1]贵州师范学院数学与计算机科学学院,贵州省高校工业物联网工程技术研究中心,贵州贵阳550018

出　　处：《江西师范大学学报（自然科学版）》2016年第3期307-311,共5页Journal of Jiangxi Normal University(Natural Science Edition)

基　　金：贵州省2014年省级本科教学工程项目(黔教高发〔2014〕378);2015年省级本科教学工程建设项目(黔教高发〔2015〕337);卓越工程师教育培养计划(黔教高发〔2013〕446)资助项目

摘　　要：为了降低多核片上系统MPSo C在应用中的能耗,在MPSo C上提出了基于优化离散粒子群算法的节能任务调度算法.通过比例选择算子生成初始种群,以任务在MPSo C上不同内核执行的能耗作为解空间,粒子群在整个解空间上搜索最低能耗调度方案,并在算法中优化了粒子群算法的局部早熟问题,使算法性能进一步提升.仿真实验表明:基于优化离散粒子群算法的节能调度算法与常用的3种调度算法相比,能耗得到了降低,且算法的截止期错失率并没有升高,保证了算法的整体性能.In order to reduce the energy consumption of the MPSo C in the application,a energy saving task scheduling algorithm based on the optimization of discrete particle swarm optimization is proposed in MPSo C. The algorithm generates the initial population by the proportional selection operator,and takes the energy consumption of different cores on the MPSo C as the solution space,search for the lowest energy consumption scheduling scheme in the whole solution space and optimize local premature problem of the particle swarm optimization algorithm. Through these improve the performance of the algorithm. The simulation experiments show that the energy of saving scheduling algorithm based on the optimized discrete particle swarm optimization algorithm is lower than that of the common three algorithms,and the deadline miss rate of the algorithm does not increase,that ensure the overall performance of the algorithm.

关 键 词：粒子群算法 多处理器片上系统 节能 调度算法 

分 类 号：TP302.1[自动化与计算机技术—计算机系统结构]