Open oharboe opened 4 weeks ago
We do allow some degradation in order to avoid local minima but this seems a bit extreme.
Can you clarify the issue a bit? Are you concerned about setup WNS degradation during hold fixing? Or is this about setup WNS QoR trajectory from floorplanning to CTS?
I dont have any concerns or expectations as I dont know the repair setup algorithm, I just thought the WNS dip during setup fixing might be an interesting test case that merited some further study: do we understand what is happening here?
OK. Understood. We'll look into WNS fluctuation during setup fixing.
I'm getting an incompatible db schema error after re-building OR with the latest version: Error: cts.tcl, 4 incompatible database schema revision 0.96 > 0.90
That would suggest you have an older version rather than a newer one. What commit id does OR show?
OpenROAD v2.0-15833-ge3b4fc0c9
Sorry, I had an issue with git pull which caused the db schema error. Now I can reproduce the issue.
Here is another plot from various runs, it looks like the test case in this issue is indeed a bit peculiar and merits some further study:
Describe the bug
Here WNS gets worse in timing repair in CTS before bouncing back up to roughly where it started.
untar https://drive.google.com/file/d/1ICisYfJdyxyJbMKcgs6JkNTB2IT05afA/view?usp=sharing
Expected Behavior
I don't know what the expected behavior is here, but I tought it looked interesting enough to file a standalone test-case.
Environment
To Reproduce
See above
Relevant log output
No response
Screenshots
No response
Additional Context
No response