多处理器系统的实时调度算法研究

调度算法是实时系统的关键技术，选取何种算法调度实时任务，这将直接影响着系统的实时响应能力。多处理器系统有局部调度和全局调度两类实时调度方法，以PFair公平调度为代表的全局调度是当前研究的热点。研究了典型局部调度EDF-FF算法和典型PFair公平调度PD^2算法，比较了多处理器系统采用PD^2算法相对于采用EDF-FF算法实现任务调度的优点，分析了由于任务频繁抢占和迁移，PD^2算法引起的时间消耗，估计并比较了PD2算法和EDF-FF算法的时间消耗，最后得出结论：在共享内存的多处理器系统中，公平调度算法是实时任务调度的比较理想的选择。

Study of fair scheduling algorithm on multiprocessors

The scheduling algorithm is an important technology for real-time systems, and the algorithm effects the performance of realtime systems. Partitioning and global scheduling are two approaches for scheduling real-time tasks on multiprocessors. PFair global scheduling is studied by many researches. An EDF-FF partitioning scheduling algorithm and a PD~2 PFair scheduling algorithm are introduced. Compared to EDF-FF, the benefit of PD~2 is presented and the issue by PD~2 PFair scheduling is investigated. In a conclusion, PFair global scheduling is an optimal scheduling methond on multiprocessors.