A Heterogeneous Parallel Deduction System

This paper describes the architecture, implementation and performance, of a heterogeneous parallel deduction system (HPDS). The HPDS uses multiple deduction components, each of which attempts to find a refutation of the same input set, but using different deduction formats. The components cooperate by distributing clauses they generate to other components. The HPDS has been implemented in Prolog-D-Linda. Prolog-D-Linda provides appropriate data transfer and synchronisation facilities for implementing parallel deduction systems. The performance of the HPDS has been investigated. Parallel Deduction Systems A parallel deduction system is one in which multiple deduction components run in parallel on separate processors. This is distinct from those deduction systems which run multiple deduction components alternately, such as the unit preference system [Wos, Carlson & RobinsonּG.A.,ּ1964], and those which are only conceptually parallel systems. Parallel deduction systems can be categorised along three axes. Homogeneous or Heterogeneous? In a homogeneous system the multiple deduction components all use the same deduction format. Homogeneous systems do not change the search space of the underlying deduction format. Rather they take advantage of added computing power to distribute the work of searching that space. An example of such a system is ROO [Lusk, Slaney & McCune,ּ1991]. In a heterogeneous system the multiple deduction components use different deduction formats. The advantage of heterogeneity is that each deduction component has a different search space. Further, it is necessary for only one of the components to use a complete deduction format. This feature permits incomplete formats, which are fast on average, to be used as part of complete parallel deduction system. An example of such a system is GLD||UR [Sutcliffe,ּ1991]. Between homogeneous and heterogeneous are the pseudo heterogeneous systems. The components of these systems all use the same deduction format, but each component has different set-up parameters. Each component will thus have a similar search space, but will search the space in a different manner. An example of such a system is Ertel's Random Competition system [Ertel,ּ1991]. Common or Separate input sets? To implement a common input set for multiple deduction components, some kind of shared memory architecture is necessary. On the other hand, if each component maintains its own copy of the input set then more common computer architectures can be used. ROO is an example of a system that uses a common input set, while the Random