多机多任务实时系统云调度策略

The Cloud Scheduler Politics of Multiprocessor Multitask Real Time Systems

采用云方法对多机实时系统任务的重要性和任务处理的适时性进行分析 ,实时计算被激活任务的价值及其在不同资源条件下的作业执行时间 ,提出一种兼顾任务价值和作业执行时间两方面因素的新的调度策略——云调度策略 ,有效地解决了任务优先级的判定和任务处理适时性问题 ,提高了系统资源利用效率、运行质量和系统整体效能 .仿真证明 ,此策略优于现在广泛采用的固定优先级和先到先处理调度策略 .

The scheduler of multiprocessor among tasks is a key technology to massively parallel processing (MPP) and real time systems.This paper analyses the importance and operating timing of tasks with cloud method,evaluates the value and elapsed time of activated tasks,and puts forward a new scheduler policy,cloud scheduler policy, which can take both the value and elapsed time of tasks into account.Cloud scheduler policy may solve the problem of how to define task PRI and task operating timing well,improve the using efficiency of system resources,operating quality,and enhance the system effectiveness.By means of simulation,it is proved that cloud scheduler policy is better than PRI and FIFO scheduler policies that are very widely used today.