A CLAUSE INDEXING UNIT FOR PROLOG

The Clause Indexing Unit provides hardware support for the depth first search operation of Prolog. It is implemented as a specialized memory unit tightly coupled to a general purpose or Prolog specific processor, which executes the Prolog program. When a goal is passed to the Clause Indexing Unit, it performs a parallel search for all clause heads (which are stored in the unit) that match the goal. Then it selects the first matching clause head and returns a pointer to the selected clause. Features of the Clause Indexing Unit include: binding-consistency check, fault-tolerant design and elimination of the traditional garbage collection activities for dynamic clauses. In this paper we present a description of the Clause Indexing Unit, an analysis of its benefits and an estimate of its performance.