Functional verification consumes more than 70% of the labor invested in today’s SoC designs. Yet, even with such a large investment in verification, there’s more risk of functional failure at tape out than ever before. The primary reason is that the design team does not know where they are, in terms of functional correctness, relative to the tape-out goal. Coverage Closure in SoC verification is like chasing a mirage.
Achieving true functional verification closure, and therefore taping out with full confidence, never happens. The decision to tape out is always a judgment call. Successful engineers achieve a sufficient level of closure and confidence based on a combination of verification thoroughness metrics, the rate, and the complexity of functional bugs being found.
All DV engineers running zillions of nightly regressions, struggling to achieve coverage goals faster, are looking for a plug-and-play solution, something automated that can help close coverage faster using minimal resources.
The Xcelium Machine Learning App helps to reduce regression times by learning from previous regression runs and guides the Xcelium randomization kernel to achieve coverage convergence faster with significantly fewer simulation cycles and catch more bugs around specific coverage points of interest.
The below graph is an example of Machine Learning App success with a leading semiconductor company. As you can see, the ML regression is about 1.4x faster than the baseline (ML hits 96% at ~62k runs, and the baseline hits that mark at about 80k runs).
This clearly represents around a 2–3-week reduction on customers' 2-month closure cycle. It’s clear that Machine Learning (ML) is making the regression notably more efficient for users’ goal of closing the coverage faster.