Verification is a labour-intensive work. Machine Learning (ML) algorithms can be used to eliminate the manual work. This paper shows ML techniques used to reduce or eliminate manual verification work which includes validating waveforms for the written design checks and manual regression in the context of verifying complex hardware designs. These techniques are based on the fundamental problem of looking for similar “difference patterns” and grouping them. When the differences are grouped together, it reduces or eliminates the need to manually verify them. These techniques are not specific to design verification but can be used for any similar problem. Download Now

By Comments off April 1, 2021