Evaluation of test generation algorithms

Many ATPG algorithms are proposed every year. Evaluation of ATPG algorithms not only provides possibility to compare algorithms, but also predicts required computation resources for design and test of extra large circuits. The evaluation includes aspects of fault coverage, computation efficiency, and test set size. ISCAS 85 and ISCAS 89 benchmark circuits are available common examples for the evaluation. This paper presents a general methodology for the evaluation in spite of the difference of the computing environments that the algorithms run in. Eleven ATPG algorithms are evaluated, as a case study, based on the data of their experimental results published in the literature to show the feasibility and validation of the methodology presented.<>