WorkCrews: An abstraction for controlling parallelism

In implementing parallel programs, identifying the opportunities for concurrency is only part of the problem. In most cases, it is equally important to recognize that unrestricted parallelism can lead to inefficiency. When this occurs, it is important to find strategies for controlling parallelism that make the most effective use of available resources. A dynamic strategy called WorkCrews is introduced for controlling the use of parallelism on small-scale, tightly coupled multiprocessors. In the WorkCrew model, tasks are assigned to a finite set of workers. As in other mechanisms for specifying parallelism, each worker can enqueue subtasks for concurrent evaluation by other workers as they become idle. The WorkCrew paradigm has 2 advantages. First, much of the work associated with task division can be deferred until a new worker actually undertakes the subtask and avoided if the original worker ends up executing the subtask. Second, the ordering of queue requests favors coarse-grained subtasks, which reduces the overhead of task decomposition.