Resource-Efficient Execution of Conditional Parallel Real-Time Tasks

Under the federated paradigm of multiprocessor scheduling, a set of processors is reserved for the exclusive use of each task. We consider the federated scheduling of parallel real-time tasks containing conditional (if-then-else) constructs, in which different executions of the task may result in workloads of substantially different magnitude and different character (e.g., degree of parallelism and critical path length). If the task is hard-real-time, then processors must be reserved for it under worst-case assumptions. However, it may be the case that most invocations of the task will have computational demand far below the worst-case characterization, and could have been scheduled correctly upon far fewer processors than had been assigned to it based upon the worst-case characterization of its run-time behavior. Provided we could safely determine during run-time if the worst-case characterization is likely to be realized during some execution and all the processors are therefore going to be needed, for the rest of the time the unneeded processors could be idled in low-energy "sleep" mode, or used for executing non-real time work in the background. In this paper we propose an algorithm for scheduling parallel conditional tasks that permits us to do so.