Toward VLSI Complexity : The DA Algorithm Scaling Problem : Can Special DA Hardware Help?

Abstract

With the increasing scale of integration we need to employ DA algorithms to assist us in managing the complexities involved. Many of our current techniques are already costly and slow and yet scale by some power law as the gate count increases. An analysis of the strategies open to us leads to the possibility that in many cases specialised DA hardware is cost effective. This paper then describes a series of trials to employ a 64 x 64 distributed array processor at DA tasks. Some unexpected results were obtained as algorithms evolved in the area of tracking, simulation, placement, test generation, fault simulation, and layout rule checking. The paper concludes with a discussion of the overheads incurred in employing unconventional hardware.