Abstract: In this paper, a management model and an online scheduling algorithm are proposed for the reconfigurable resource management of multitask operating system. The specific implementation is to assign tasks to a block-partitioned reconfigurable device. On the one hand, the reconfigurable device is controlled by a host CPU that runs online scheduler and placer, on the other hand, the reconfigurable device contains fixed size of blocks with the same vertical dimension, but the blocks can have different widths. The motivation of having differently-sized blocks is to achieve a better match between the resources and the tasks. At the same time, the two functions fSPLIT and fSELECT run in the online scheduler and placer to achieve the placement and scheduling of tasks on the reconfigurable device. The simulation results show that the resource management model and scheduling algorithm proposed can not only achieve  the minimization of the average response time and efficient scheduling for a set of tasks, but also can achieve higher resource utilization compared with other scheduling algorithms.

Key words: reconfigurable resource, modelling, partition method, online scheduling, response time, resource utilization

摘要： 针对多任务操作系统的可重构资源管理，提出了一种管理模型和在线调度算法，具体实现了把任务分配给基于块划分的可重构器件。一方面，可重构器件由一个主CPU控制，主CPU运行在线调度器和放置器；另一方面，可重构器件由具有相同垂直尺寸的固定大小的块构成，但块可以有不同的宽度，目的是为了在资源和任务之间实现更好的匹配；同时在在线调度器和放置器运行两个函数fSPLIT和fSELECT来实现任务在可重构器件上的配置和调度。仿真结果表明，提出的资源管理模型和调度算法不仅能够实现任务集平均响应时间的最小化和有效调度，而且相比于其他调度算法，还能获得更高的资源利用率。

关键词: 可重构资源, 建模, 划分方法, 在线调度, 响应时间, 资源利用率