Performance of Sparse Binding Arrays for Or-Parallelism
Vı́tor Santos Costa, Manuel E. Correia, Fernando Silva · 1996
One important problem in the design of novel logic programming systems is the support of several forms of implicit parallelism. A new binding model, the Sparse Binding Array (SBA), has been proposed for the efficient and simplified integration of Independent-And, Determinate-And and Or-parallelism. In this paper we report on the use of this model for pure Or-parallelism. The work discusses the major implementation issues in supporting this binding model for pure Or-parallelism. We show that an implementation based on this Binding model is more efficient then the original Aurora using tbe traditional Binding Array model [16]. Moreover, we explain how the notion of a variable level can be used to reduce overheads of the Orparallel system. Our results in supporting pure or-parallelism show that the approach is very promissing for combined paralell systems.