Evaluating the effectiveness of D-chains in SAT-based ATPG

Jan Burchard,Felix Neubauer,Pascal Raiola,D. Erb,B. Becker
DOI: https://doi.org/10.1109/LATW.2017.7906752
2017-03-13
Abstract:With the ever increasing size of today's Very-Large-Scale-Integration (VLSI) designs new approaches for test pattern generation become more and more popular. One of the best known methods is SAT-based automatic test pattern generation (ATPG) which, in contrast to classical structural ATPG, first generates a mathematical representation of the problem in form of a Boolean formula. A specialized solver evaluates this representation to determine the testability of faults and extracts a test pattern in case a satisfying assignment was found. In order to increase the solving speed [1] introduced the concept of D-chains which add additional information to the mathematical model. In return, this forces the solver to only consider assignments that might lead to a valid test pattern and thus reduce the search space. With the advent of incremental solving new concepts like the backward D-chain or even more recently an indirect D-chain were introduced. However, none of the previous publications tried to analyze and evaluate which of these methods is the most beneficial. In this paper we present a thorough investigation of the different D-chain concepts to evaluate which is the best method for different problems. In addition, we propose a new indirect D-chain algorithm with two extensions. Our experimental results show that depending on the incorporated D-chain the runtime can be reduced tremendously.
What problem does this paper attempt to address?