摘要 |
Embodiments of the invention provide a method, system and computer program product for scheduling tasks in a computer system. In an embodiment, the method comprises receiving a multitude of sets of tasks, and placing the tasks in one or more task queues. The tasks are taken from the one or more task queues and placed in a priority queue according to a first rule. The tasks in the priority queue are assigned to a multitude of working threads according to a second rule based, in part, on share values given to the tasks. In an embodiment, the tasks of each of the sets are placed in a respective one task queue; and all of the tasks in the priority queue from each of the task queues, are assigned as a group to one of the working threads.
|