Cadence recommend running the UNR analysis when your code coverage metrics are near 70-95% depending on the size of the design.
RTL and code coverage data are inputs to the formal tool for UNR analysis. Provided RTL is frozen and the test suite is maturing I believe that the better the code coverage metrics
are given as input, we get two advantages
1. Faster UNR analysis
2. Accurate UNR analysis.
Some folks say that the UNR output shall not change if RTL is stable. I see i get more accurate UNR (Lower numbers), when I do the analysis when my test suite is mature
enough to give good code coverage result.
Kindly share your experiences if you see more accurate result when code coverage data is better.
Thanks in Advance.