This is related to the previous post. It appears t...
# openramm
Mitch Bailey
12/01/2021, 3:32 AMThis is related to the previous post. It appears that the openram device generators are set to use only one licon per diffusion. Is there a way to change that to connect all the way across as @User suggested?
Also, thin li power rails trouble me. I'm used to seeing wide metal power rails.
Is there a process engineer at skywater willing to answer design questions?
m
Matthew Guthaus
12/01/2021, 4:16 AMI believe there is rough code there to do this but it needs to be debugged. Please submit a PR if you can help fix it.
Matthew Guthaus
12/01/2021, 4:16 AMOtherwise we will get to it soon.
Matthew Guthaus
12/01/2021, 4:17 AMThese li rails are tiny so that is likely not an issue but you are welcome to add m1 straps. That would be great.
Matthew Guthaus
12/01/2021, 4:17 AMThe feedback is great but we also need help!
Matthew Guthaus
12/01/2021, 2:26 PM@User can you fix the licon?
m
Mitch Bailey
12/01/2021, 11:00 PMNot easily. I've only done a smidgen of layout and am not familiar with magic's device generation. You are using with magic for layout, right?
Currently, I'm in the middle of device level LVS and CVC (reliability check) for the mpw-2 shuttle designs. Depending on those results, I'm hoping that efabless will be willing to fund continued verification.
m
Matthew Guthaus
12/02/2021, 12:12 AM@User This does not use magic's device generation. It is Python code in OpenRAM
Matthew Guthaus
12/02/2021, 12:13 AMOk, we'll try to fix it, but this is mostly a reliability issue and not a correctness issue.