In this paper we present a distributed multiprocessor model for
OR parallel execution of Prolog. It is based on multiple sequential
Prolog engines connected by real or virtual channels.
In order to minimize the messages traffic, each processor has its own
environment. Two approaches to construct the parent
processor environment on an idle processor have been presented: stack copying
and recomputation. It also presents and compares performance results of both
approaches, implemented on a transputer-based system.