Replies: 5 comments 4 replies
-
@oharboe |
Beta Was this translation helpful? Give feedback.
-
Hmm.... is the result of detailed routing deterministic across machine configurations? I had the failure on one machine, but not another. A difference between the machines is the number of cores. Also, I see a different number of violations:
Will |
Beta Was this translation helpful? Give feedback.
-
Its a rather internal error so it will be tricky to debug without a test. Can you turn on the gcell grid (under Misc) in the GUI image? I'm guessing the net lies on the gcell edge? The router should be deterministic so the variation across machines is concerning. |
Beta Was this translation helpful? Give feedback.
-
I wonder if you could try https://github.com/The-OpenROAD-Project/OpenROAD/blob/master/etc/deltaDebug.py to reduce this to a simple enough test case that you could share it. You only need to iterate on the drt step. |
Beta Was this translation helpful? Give feedback.
-
This was sorted using deltaDebug.py and communicating the confidential case with Matt directly. |
Beta Was this translation helpful? Give feedback.
-
What does this mean and is there anything I can do about it?
Is AutoTuner designed to adjust parameters to find parameters that work around such problems?
E.g. I thought I might try to increase or decresase PLACE_DENSITY to see if that might be the problem.
Can the GUI tell me something? How would I look?
Looking at the source code,
nextPin->getName()
doesn't have a name(it is an empty string):https://github.com/The-OpenROAD-Project/OpenROAD/blob/d297c5b18d08684b550e74a3115d4c366bd92e01/src/drt/src/dr/FlexDR_maze.cpp#L2914
Beta Was this translation helpful? Give feedback.
All reactions