Fast search algorithms for reconfiguration problems

Abstract

A number of reconfiguration strategies have been proposed for increasing the yield of VLSI chips. In most cases the associated reconfiguration problems are NP-complete. Therefore, exhaustive search algorithms are generally used in order to find a solution when one exists. In this paper we present the notion of admissible sets and show how such sets can be used to significantly reduce the running time of many exhaustive search algorithms for reconfiguration problems. As an example, the authors find a class of admissible sets called excess-k critical sets that can be used in the design of fast search algorithms for the problem of reconfiguring redundant random access memories (RRAMs). They also consider applications to the problems of reconfiguring RRAMs with shared spares and reconfiguring redundant programmable logic arrays (RPLAs). Experimental results indicate that this approach is very powerful.<>