A Run-Time Algorithm for Managing the Granularity of Parallel Functional Programs

We present an on-line (run-time) algorithm that manages the granularity of parallelfunctional programs. The algorithm exploits useful parallelism when it exists, andignores ineective parallelism in programs that produce many small tasks. The ideais to balance the amount of local work with the cost of distributing the work. Thisis achieved by ensuring that, for every parallel task spawned, an amount of work thatequals the cost of the spawn is performed locally. We analyze several...