Genetic Algorithms: The Powerful Driver of the Functional Verification Process

The number of mobile devices is increasingly higher. These devices are powered by microprocessors. In order to fast and reliable verify the functionality of the chips before they are manufactured, the classic methods of verification are sometimes not enough, even only for the reason that at least six-eight months are needed to verify a digital design. Therefore, methods of automating the verification process are more and more approached by almost all important companies in semiconductor industry. Aligning to this trend and taking as case study a smart lamp design, in this paper, several approaches of functional verification based on genetic algorithms have been developed, and their performances have been compared. A means of automatic communication between Python code containing genetic algorithms implementations and the simulation environment incorporating the digital design has been established. It was shown that the automation algorithms, although having a common structure, need to be tailored to the specifics of each digital design in order to achieve the best performance in terms of maximum coverage value reached and processing time. It was demonstrated that many developed versions based on genetic algorithms were able to achieve 100% coverage without any direct input from the programmer.