Speculative Computation and Priorities in Concurrent Logic Languages

Abstract

Speculative computation is a technique which enables us to obtain extra amounts of parallelism from a problem by executing computations which may only possibly be required (we “speculate” that they will be required). We show how speculative computations may be expressed in the concurrent logic languages, but indicate that in the current state of these languages they are unsatisfactory as they do not allow us to express priority between computations. We propose a simple priority operator which overcomes this problem.