These algorithms require fault simulation to be run for every flip-flop with different workloads, and as such it has not been possible to perform this “selective hardening” process until now. High-performance fault analysis, coupled with other techniques, allow enough iterations to be performed to complete this analysis, generating appropriate metrics and the ideal hardened flip-flop configuration.
By enhancing the core verification methodology, many new safety checks can be performed, as well as other critical issues. Security analysis is also possible using some of these same techniques.
As we are witnessing a new generation of AI devices for automated driverless vehicles, and these powerful, complex semiconductors require a new design methodology. This is not just about maximizing safety, but is necessary to revolutionize the entire integrity of these designs. High performance fault analysis is at the heart of this new approach and will become the center of next-generation automotive design methodologies.
About the authors:
Jamil Mazzawi is CEO of Optima Design Automation - www.optima-da.com
David Kelf is Board Advisor, Marketing at Optima Design Automation.