A variable supply model for distributing deductions
Vineet Kumar Singh, Michael Genesereth · International Joint Conference on Artificial Intelligence · 1985
Multiple processors can be used to speed up a backward-chaining deduction by distributing or-parallel deductions. However, the actual speedup obtained is strongly dependent on the amount of communication required for the task allocation strategy. A Variable Supply Model (VSM) is presented for multiple processors with replicated databases on a broadcast network. The term model refers to the set of procedures and messages required to perform the computation. VSM allows an infinite class of strategies with varying amounts of communication. The utility of VSM lies in the easy and powerful way it provides for selecung a strategy that works satisfactorily given certain communication constraints. All strategies in VSM use a dynamic task supply protocol (ESP) that works better than other supply protocols described in the literature.