I'm unable to get a clean DRC result with the pins...
# openramm
Matt Venn
12/22/2021, 9:25 AMI'm unable to get a clean DRC result with the pins on the right side of openram
m
Mitch Bailey
12/22/2021, 10:17 AMYou're viewing magic DRC errors in klayout, correct?
m
Matt Venn
12/22/2021, 10:18 AMyep
Matt Venn
12/22/2021, 10:18 AMwith the .xml file generated by the precheck
Matt Venn
12/22/2021, 10:18 AM(I'm now adding met3 obstruction boxes around the pins to prevent the routing from jogging inside the macro.)
Matt Venn
12/22/2021, 10:19 AMimage.png
Matt Venn
12/22/2021, 10:19 AM😂
Matt Venn
12/22/2021, 10:20 AMimage.png
Matt Venn
12/22/2021, 10:20 AMseems to be working
Matt Venn
12/22/2021, 10:23 AMhmm, maybe not, the obstruction seems to be causing it's own DRC issues
Matt Venn
12/22/2021, 10:34 AMso I can force the lwire to come in straight and that solves the unrelated met3 DRC
Matt Venn
12/22/2021, 10:34 AMimage.png
Matt Venn
12/22/2021, 10:35 AMbut I'm getting metspc DRC between the met3.obs and met3.blk - which seems weird to me
Matt Venn
12/22/2021, 10:49 AMI went back to my old strategy of covering the entire macro with obstruction on met2,3 and 4.
Matt Venn
12/22/2021, 10:50 AMit needs to go almost to the edge of the macro, and that prevents the router from doing it's jogging inside
Matt Venn
12/22/2021, 11:12 AMfixed it with that
👍 1
m
Matthew Guthaus
12/22/2021, 12:20 PMDo you have an example of this? I'm not clear what is going on from the pictures
m
Matt Venn
12/22/2021, 4:33 PMnot any more. The 2 pics below this thread show the issues I was having. If the router does that jog at the macro boundry, it ends up in an unrelated met3 error
Matt Venn
12/22/2021, 4:33 PMno idea why
Matt Venn
12/22/2021, 4:33 PMso I tried a few ways to prevent that from happening
Matt Venn
12/22/2021, 4:34 PMfinally using a large OBS on met 3 & 4 the cover nearly the entire macro
d
Dinesh A
12/22/2021, 4:53 PMThis issue is due to router appears not to honor the wide-spacing rules between the routes and the obstruction layers defined in the macros.
I have raised the issue on Openroad :https://github.com/The-OpenROAD-Project/OpenLane/issues/695
--------comment from "*RTimothyEdwards" --*
I would like to add that the specific problem here is that the OpenROAD router appears not to honor the wide-spacing rules between the routes and the obstruction layers defined in the macros. There happen not to be any actual DRC errors because the obstruction layers extend out further than any of the internal routes, but that does not invalidate the issue, since the router should be maintaining a DRC distance per the information it sees in the macro LEF.
9 Views