Why do we consider CPPR in SI timing analysis?

CPPR is the common path pessimism reduction and needs to be considered wherever appropriate to remove the unwanted pessimism you are adding to your analysis.

Now the problem needs to be looked at from a SI timing analysis angle. How will crosstalk delta affect a particular edge? Crosstalk delta will be the sum total of impacts from multiple aggressors and it will be affecting each clock cycle differently. Each aggressor might speed up/ slow down a particular clock edge and it might be different for the next edge. So, derates for the delta cannot be eliminated. On the other hand derates for the common cell delays can be eliminated . If the analysis is for same edge, then you can eliminate all the derates applied to crosstalk delta and cell delays.

To summarize ;

Any checks that involves multiple edges

eg: regular setup checks

Common path pessimism for regular delays has to be removed

Common path pessimism for crosstalk delta should not be removed

2. Any same cycle/ zero cycle check

eg: regular same cycle hold check

Common path pessimism for regular and crosstalk delta can be removed as it is the same edge.

it is an idea how different edges will get different crosstalk deltas

Yes CRPR is considered in SI timing analysis as well. For setup, the delays are considered so as the launch is slow (as late as possible) while the capture is fast (as early as possible). Similarly, for hold check, it’s the reverse. Now following this principle can lead to pessimism on clock paths. And thats where Clock Reconvergence Pessimism Removal (CRPR) comes into the picture