Processing data streams with hard real-time constraints on heterogeneous systems
Proceedings of the international conference on Supercomputing
Scheduling processing of real-time data streams on heterogeneous multi-GPU systems
Proceedings of the 5th Annual International Systems and Storage Conference
| Hi-index | 0.00 |
Pfair scheduling has usually been applied in the context of scheduling periodic tasks with coinciding periods and deadlines. This paper addresses the problem of scheduling hard real-time tasks characterized by arbitrary deadlines in a pfair manner on multiprocessor systems. We prove some fundamental properties of the weight functions of such tasks and show that when such tasks are scheduled in a pfair manner, every job of every task meets its deadline. We also present RRM (Rational Rate-Monotonic), a static-priority scheme to schedule a set of such tasks, and derive demand-based sufficient conditions for determining the pfair schedulability of hard real-time tasks with arbitrary deadlines on multiprocessors under the RRM scheme. We also address the problem of accounting for interrupt-handling overheads in pfair systems and present a technique for implementing the Priority Ceiling Protocol on pfair systems by modifying task weight functions.