Reducing time by avoiding process serialisations in or-parallel interpretation of logic programs

Two kinds of parallelism arise when interpreting a logic program with a parallel architecture: Or-parallelism and And-parallelism. The former searches in parallel for the solution of one goal-literal by executing one process for each clause whose head unifies with the goal-literal. The latter is applied to solve the clause body activating parallel processes for the literals in the body of a clause. A big drawback of the previous Or-parallel models is that they fall down easily in the combinatorial explosion inherent in the solution of a logic programming problem. In this context, a particular processor in a parallel architecture is constrained to work with many processes which causes huge overheads by change contexts. The limitation in the required number of processes reduces the communication costs and it also avoids to serialise the processes.