Pre-placement of VLSI blocks through learning neural networks

This paper presents a new neural network approach to the pre-placement of VLSI blocks. The authors´ innovation consists in considering the pre-placement problem as a classification problem, and in implementing a neural network learning algorithm, derived from Kohonen maps, to accomplish this task. The neural network has been simulated and the results have been compared with those obtained from a classical min-cut algorithm. The comparison shows as the authors´ technique offers comparable quality of solutions, but presents a more favourable computational complexity