Monte Carlo Algorithm for Compressing Corner Tests

Authors: Thomas Nirmaier, Infineon Technologies AG, DE; Georg Pelz, Infineon Technologies AG, DE


Corner tests are used in pre- and post-Si verification to ensure device functionality at the outer limits of AC and DC operating conditions under all constraints given by the specification, process variations or other deviations from nominal conditions. The number of exhaustive tests to test all possible corner conditions grows exponentially with the number of corner conditions, so does simulation time and test time. We propose a compression algorithm for corner tests to overcome this exponential growth, by introducing the assumption that only a limited maximum correlation between corners must be covered for. An appropriate exhaustive solver for finding required combinations of corners that satisfy the correlation condition can minimize the test time then, but reaches a computational complexity that exceeds current computational power of PCs. We propose a Monte Carlo approach in a second step to still find a close-to-optimum solution for large number of corners.


1-4244-4203-0/08/$20.00 ©2011 IEEE

Publication Date: 2011/09/22

Location of Publication: International Test Conference (ITC), Anaheim, California, USA

Keywords: Automotive; Verification