Bringing support for ICSprout 55 (ics55)

[ckdur]

Greetings.

I did enough modifications to build and enable the ics55 PDK here. I would be happy if you make this available even in the repository.
Hopefully everything is okay. Let me know if you need more modifications.

[RTimothyEdwards]

This is not really in the form of an open PDKs in the form defined by https://github.com/fossi-foundation/open-pdks/. It is also questionable how "open" the ICSprout 55 process really is: There are no device models, no layout other than abstract views, and so no way to characterize a layout other than by blindly trusting the delay values in the liberty files. There are no DRC rules and no way to do any validation beyond what the tools generate for digital-only layouts. It is a long, long way from an open PDK.

[ckdur]

Totally agree @RTimothyEdwards

It was just an attempt to make it available, but I agree that the openness of this PDK is quite questionable.

If they update, I will be ready to update in the open-pdk format the best I can.

I am not associated with ICSprout at all. Just a hobby.

[RTimothyEdwards]

@ckdur : My comments on the openness of the process were mainly to check that you're aware of how much is actually missing from the PDK.

For the other comment, though, I would like to see the PDK reorganized slightly to align with the other open PDKs offered under the "ciel" umbrella. That would involve adding the PDK to open-pdks, which would then handle pulling the sources from https://github.com/openecos-projects and re-organizing the two libraries under IP/IO and IP/STD_cell to be under libs.ref/ instead, and normalizing the syntax (e.g., for the I/O library the verilog has icsIOA_N55_3P3``, the liberty has ICSIOA_N55_3P3, and the LEF has ICSIOA_N55_3P3_1P6M1TM`). The open-pdks project was designed specifically to address such issues and normalize the PDKs so that all the tools know where to find things. The ICSPROUT libraries are close to the preferred organization and syntax, so the handling in open-pdks can be straightforward.

[donn]

@RTimothyEdwards Assuming upstream doesn't want to completely conform to the Open PDKs format, would you be open to the idea of a contribution in https://github.com/fossi-foundation/open_pdks, or should the tweaks live in ciel?

[RTimothyEdwards]

@donn : I'd prefer it as a contribution to open-pdks. It is what I designed open-pdks to do.

[ckdur]

Will try then to make the changes in open-pdks in that case. In the end, is a matter of re-creating the copy using Makefile.in and the json file right?