Did you specify the `-rcfile` on the command line ...
# magic
t
Did you specify the
-rcfile
on the command line to loade the sky130 tech ?
d
I did
tech load .../magic/sky130A
instead which seemed to load the sky130 tech. Will try the
-rcfile
method
t
Yeah the
sky130A.magicrc
has a lot more config than just the
tech load
d
magic -rcfile <abspath to sky130A.magicrc
isn't working on my machine, it tries to find
sky130A.tech
using my abspath param as a relative path to pwd, /usr/loca/lib/magic/sys and a few others. That definitely seems wrong
or so the logs say, I'll try to rule out any obvious silly errors on my machine
right think I see, some of the files seem to have abspath baked into them. The machine I copied the /sky130A it to is very resource limited and running the sky130A make target will take forget, I'll have to work around this somehow
got it working after a fresh /sky130A install on the ubuntu machine, although the inserted diode then triggers mismatch error logs, probably from the GDS cells / my recent not being exactly the same I can tell that there's going to be some learning curve here though before I can accomplish anything useful here..
although if the expectation is that these 100+ antenna violations must be patched by hand to have a submittable design, seems that there's no way around it
t
Yeah you can't just copy stuff, they are full of hardcoded path.
But in anycase I don't think you'd be able to fix antenna violations using that 'diode' device because most will happen in the middle of the existing design where there is just no space to insert anything.
d
if there's no config changes to make antenna violation patching more practical, then I'm hoping the upcoming docs will help configure the design to avoid the violations in the first place. One of the videos mentioned that the RC of openlane should resolve the DRC violations at least I might just wait for the RC tools / docs to come out and if I still get violations, well I suppose that's that 🤷‍♂️. I don't have that much time or CPU resources to iterate on different configurations for too long at least not for the Nov 30 deadline