混合粒子群算法的异构多核处理器间任务调度

针对异构多核处理器间的任务调度问题,为了更好地发挥异构多核处理器间的平台优势,提出一种基于将有关联的且不在同一处理器上的任务进行复制的思想,从而使每个异构多核的处理器能独立执行任务,来减少不同处理器之间的通信开销,并且通过混合粒子群算法(HPSO)来调度异构多核处理器中的任务,避免由于当任意一个异构多核处理器由于任务分配过多而导致计算机不能及时且准确地得出结果.最后实验证明,对比传统的启发式分配方案和常见的遗传算法(GA),基于任务复制思想分配方案和混合粒子群算法(HPSO)具有更好的求解能力,并且可以提供执行时间更少的调度分配方案,具有较好的应用价值.

Task Scheduling for Heterogeneous Multi-core Processors Based on HPSO

In order to solve the problem of task scheduling among heterogeneous multi-core processors and better play to the advantage of heterogeneous multi-core processor platform,a replicating idea based on connected and not on the same processor task is proposed,so that each heterogeneous multi-core processor can independently perform the task to reduce the communication overhead among different processors.The hybrid particle swarm optimization algorithm (HPSO) is used to schedule tasks in heterogeneous multi-core processor,avoids the results can not be timely and accurately shown when an arbitrary heterogeneous multi-core processor has too many tasks.The experiment results show that compared with the traditional heuristic allocation scheme and the common genetic algorithm,the solution has better solving ability,and can provide the implementation scheme of less time scheduling and allocation,and has good application value.