Solutions to the Module Orientation and Rotation Problems by Neural Computation Networks

In this paper we study two strategies for modifying a given placement of modules in order to improve the quality of the routing results in the next stage of design. We assume that the modules have already been placed. The first strategy seeks to minimize the total wire length by flipping each module about its vertical and/or horizontal axes of symmetry. The second strategy seeks to minimize the total wire length by rotating each module by a multiple of 90 degrees. We introduce a new algorithm based on the Hopfield-Tank neuralnet model to solve these problems. Our algorithm performs better than the best algorithms known for these problems. Both problems are shown to be NP-Complete.