Abstract
This paper discusses the implementation of Hopfield neural networks for solving
constraint satisfaction problems using Field Programmable Gate Arrays (FPGAs).
It discusses techniques for formulating such problems as discrete neural
networks, and then it describes the N-Queen problem using this formulation. A
prototype implementation of the a number of different N-Queen problems is
described and results are presented that illustrate that a speedup of up to 3
orders of magnitude is possible using current FPGAs devices.