基于粒子群优化算法的多核处理器任务调度研究  

Research on multi-core processor task scheduling based on particle swarm optimization algorithm

作　　者：田佳 胡威[1] 李琳[1] 柯鹏[1] 张凯[1] 

机构地区：[1]武汉科技大学计算机科学与技术学院湖北省智能信息处理与实时工业系统重点实验室,武汉430065

出　　处：《计算机应用研究》2017年第12期3698-3700,共3页Application Research of Computers

摘　　要：针对多核处理器在调度多个任务时效率不高的问题,提出了一种基于粒子群优化算法的嵌入式多核多线程系统任务调度算法,用来找寻任务调度过程中的最优解,以求取任务的最短完成时间。在算法中通过针对多核多线程任务模型而选择粒子群算法的适应度函数,综合利用局部最优极值和全局最优极值的优势,优化了粒子群算法中存在的过早收敛问题,使算法具有较高的收敛效率。实验结果表明,与基于遗传算法的多核多线程任务调度算法相比,该算法能更快地找到最优解。In allusion to the problem that the efficiency of multi-core processors in the process of scheduling multiple tasks is not high,on the basis of particle swarm optimization algorithm,this paper put forward a kind of embedded multi-core and multi-thread scheduling algorithm to be used to seek the optimal solution in the process of the task. In the algorithm,through the multi-core and multithreading task model,it chose fitness function of particle swarm algorithm and utilized advantages of partial optimal extremum and global optimal extremum optimized the problem of the premature convergence in the particle swarm algorithm to make the algorithm's convergence efficiency high. The experimental results show that compared with the traditional algorithm of multi-core and multithreading task scheduling,the algorithm can find the optimal solution faster.

关 键 词：粒子群优化算法 多核处理器 多核多线程 任务调度 

分 类 号：TP303[自动化与计算机技术—计算机系统结构]