A fuzzy-neuro approach for timing-driven system partitioning in VLSI multi-chip design

A contemporary definition of the VLSI system partitioning problem is characterized by multiple contradictory objectives to minimize connection wires while satisfying timing and capacity constraints. Although a number of algorithms have been proposed in recent years, they are unable to efficiently exploit massively parallel architecture. Moreover, the utilization of expert knowledge does not explicitly address the problem of multiple contradictory objectives. In this paper, a fuzzy-neuro approach which models expert knowledge as fuzzy logic rules is proposed to govern the multi-objective optimization process of neural networks. Real circuits are presented to verify the performance and feasibility of the proposed method. Comparisons are made with the Hopfield-type network. Experiments show that the proposed method is better than previous methods.<>