I am working on verifying a parameterized IP that has N design parameters. The goal is to verify the IP in a generic way so that when different business units use the IP with their own configurations, we don’t need to re-run full verification for each specific set of parameters.
Currently, each business unit requires 2–3 different configurations for their own use case, and we might have M customers in parallel. This leads to:
• Running regressions separately for each configuration.
• Collecting code coverage independently for each configuration.
• Facing difficulties in merging coverage results, since parameter changes alter the RTL structure and the coverage merge fails.
My questions are:
1. What is the best methodology to verify a parameterized IP in a generic way (so we don’t need to fully re-verify each customer’s configuration)?
2. How should we handle functional/code coverage across multiple parameter configurations? Specifically, is there a recommended way to merge or abstract coverage so that we can demonstrate overall verification completeness without duplicating effort?