任务驱动的众核资源调度技术研究  

Resources Scheduling Technology by Task-driven in Many-core Architecture

作　　者：罗眉[1,2] 周兴社[1] 吴晓[1] 张凯龙[1] 

机构地区：[1]西北工业大学计算机学院陕西省嵌入式系统技术重点实验室,西安710129 [2]第二炮兵工程大学信息工程系,西安710025

出　　处：《小型微型计算机系统》2014年第6期1212-1215,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61103004)资助;青年科学基金项目(F020308)资助

摘　　要：在采用众核架构的复杂实时系统中,对众核资源进行合理的调度,使得片内资源的分配与系统任务的需求相适应,对提高资源利用率和系统处理能力至关重要.为了使众核运算单元满足不同应用任务的计算需求,本文提出了一种基于众核架构的任务资源调度模型,该模型主要由应用任务集、实时计算管理逻辑和众核处理机三部分组成,基于该模型设计了众核资源调度算法MRSA(Many-core Resource-Schedule Algorithm).最后搭建了PowerPC主处理机和众核处理机的仿真实验环境,测试了系统的核资源利用率及任务调度效率,结果表明,MRSA算法能够有效适应众核架构系统的任务需求,提高系统资源的利用率.In many-core architecture of complex real-time system, it is important that scheduling resources reasonably and making resources allocation adapt to task requirements. In order to raise the utilization ratio of resources and make arithmetic unit of many-core to meet different application task needs, this paper proposes a resource scheduling model based on many-core architecture, the model is mainly composed of three parts ： application task set, real-time computing management and the many-core processor. According to the model, the paper designs a scheduling algorithm called MRSA （Many-core Resource-Schedule Algorithm） to improve resource utilization. Finally, the paper builds the simulation environment consists of PowerPC processor and many-core processor to test the utilization ratio of system＇s resources and efficiency of task scheduling. The result shows that the algorithm can not only well adapt to task requirements, but also get more effective to execute tasks and improve the utilization ratio of system＇s resources.

关 键 词：复杂实时系统 众核架构 资源利用率 任务调度效率 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]