A new optimizer for performance optimization of integrated circuits by device sizing

A new optimizer is presented for multiple criterion constrained performance optimization of integrated circuits, based on nonlinear programming and heuristic techniques. Biggs´ dominant degree method, which handles nonquadratic functions, is used for gradient-based optimization. A new guided random search technique is used to perform optimization based only on function evaluations. These two methods are combined to overcome numerical difficulties arising during simulation and gradient-based optimization. Experimental results show performance improvements in the range of 20-200%