Parallel Execution of Parlog

This monograph concentrates on the implementation ofconcurrent logic programming languages on loosely-coupledparallel computer architectures. Two different abstractmodels are also considered: a graph reduction based modeland a more imperative approach. The design of a specializedinstruction set is presented that is general enough to beused for the whole family of concurrent logic programminglanguages. In particular the language Parlog is concentratedupon because it is the most efficiently implementable of allthe non-flat logic programming languages. Other issuesarising from the implementation of a programming languageare discussed, in particular the design of a load balancingscheme which deals with the large number of short-livedprocesses inherent in a correct language implementation. Detailed descriptions of all the data structures andsynchronisation algorithms are also presented. Resourceexhaustion is also discussed and handled within theimplementations. Detailed performance results for bothmodels of computation are given. From the material it shouldbe possible for the reader to create an implementation.