Distributed OR Scheduling with Granularity Information

Logic Programming has some implicit sources of parallelism as OR parallelism that facilitates the automatic parallelization. There are several parallel logic systems (PLPS) that exploit only OR parallelism but most part consider shared memory in its design. Few distributed implementations exist. PLoSys is an example of this kind of distributed implementation of parallel Prolog. This paper presents a proposal of granularity control for parallel logic system and the possible enhancement in PLoSys OR parallel system concerning scheduling using granularity information. Scheduling is one of the most important part in a distributed system since it decides the task attribution to processing elements and the execution order of each task. Several performance measures could be optimized in this allocation process