一种负载感知的异构MPSoC任务调度算法

发布时间：2018-04-06 03:40

本文选题：异构MPSoC　切入点：负载感知　出处：《电子科技大学学报》2017年06期

【摘要】：处理器核的异构性、运行时负载和任务间依赖关系,是影响异构MPSoC任务调度算法性能的关键因素。该文提出了一种负载感知的异构MPSoC任务调度算法,在满足任务间依赖关系的前提下,根据计算开销和通信负载将待调度任务集划分为任务子集。在考虑处理器核负载状态的基础上,通过赋权二部图最大权匹配,将任务子集调度到适载的处理器核上运行,提高了待调度任务集总执行效率。仿真实验结果表明,该算法有效降低了任务集的调度长度,提高了处理器核的利用率。
[Abstract]:The heterogeneity of processor core and the dependency between runtime load and task are the key factors that affect the performance of heterogeneous MPSoC task scheduling algorithm.In this paper, a load-aware heterogeneous MPSoC task scheduling algorithm is proposed. On the premise of satisfying inter-task dependencies, the task set to be scheduled is divided into a subset of tasks according to computational overhead and communication load.On the basis of considering the load state of the processor core, the task subset is scheduled to run on a suitable processor core by weighted bipartite graph maximum weight matching, which improves the total execution efficiency of the task set to be scheduled.The simulation results show that the algorithm can effectively reduce the scheduling length of the task set and improve the utilization of the processor core.
【作者单位】：中国科学院成都计算机应用研究所;中国科学院大学;西南民族大学计算机科学与技术学院;广西民族大学广西混杂计算与集成电路设计分析重点实验室;电子科技大学计算机科学与工程学院;
【基金】：国家自然科学基金(11371003,11461006) 广西自然科学基金(2012GXNSFGA060003) 中央高校基本科研业务费(2015NZYQN28)
【分类号】：TP301.6;TP332

【相似文献】