An Artificial Intelligence Approach to Integrated Circuit Floorplanning

This work presents an artificial intelligence approach to top-down floorplanning of integrated circuits (IC). The approach divides the floorplanning process into separate knowledge-based and algorithmic tasks. This enables the use of IC specialized knowledge to drive sophisticated floorplanning algorithms and interpret their results. This book describes PIAF, a prototype system that implements the developed approach. Special attention is paid to a detailed description of the architecture of PIAF, its inference engine, knowledge representation schemes, explanation facility, and a mechanism for backtracking with context recovery. This volume also presents in detail new graph theoretical algorithms for floorplan global routing and area and shape minimization. The interaction between these algorithms and the knowledge-based components is described. Rule-based techniques for the selection of a floorplan from a family of solutions according to design tools capabilities and circuit description are also presented. This monograph on CAD, floorplanning, global routing, artificial intelligence and graph theory is intended for graduate students and researchers. cad.