Placement and Routing in Computer Aided Design of Standard Cell Arrays by Exploiting the Structure of the Interconnection Graph

Standard cell placement and routing is an important open problem in current CAD VLSI research. We present a novel approach to placement and routing in standard cell arrays inspired by geometric constraint usage in traditional CAD systems. Placement is performed by an algorithm that places the standard cells in a spiral topology around the center of the cell array driven by a DFS on the interconnection graph. We provide an improvement of this technique by first detecting dense graphs in the interconnection graph and then placing cells that belong to denser graphs closer to the center of the spiral. By doing so we reduce the wirelength required for routing. Routing is performed by a variation of the maze algorithm enhanced by a set of heuristics that have been tuned to maximize performance. Finally we present a visualization tool and an experimental performance evaluation of our approach.