Implementation of a knowledge base for interpreting and driving integrated circuit floorplanning algorithms

An experimental integrated circuit floorplanning system is presented. The system uses a combined knowledge-based/algorithmic approach. The knowledge-based system drives the algorithms and interprets their results. This approach permits the automation of tasks that have not been automated before and permits the definition of a floorplanning strategy that successfully manages the complexity of the problem. In addition, the combined approach permits an early pruning of the solution space of the combinatorial floorplanning problem. The implementation of the approach needs a careful choice of knowledge and context representation. The nature of the knowledge domain dictates the use of design quality factors which play an important role in rule conflict resolution. Extracts of the system operation are given with floorplanning examples that show the advantages of the approach.