skywater-pdk icon indicating copy to clipboard operation
skywater-pdk copied to clipboard

Missing techfiles

Open kkiningh opened this issue 4 years ago • 12 comments

I'm trying to use the pdk with Synopsys tools (DC/ICC2), but I don't see a vendor or interconnect techfile (*.tf and *.itf respectively) which is needed for place and route and parasitics extraction. Will those be added later or is there another way I can obtain them?

kkiningh avatar Jul 02 '20 20:07 kkiningh

@kkiningh I don't believe anyone has used the PDK with th Synopsys tooling.

Is a .tf / .itf file the same as a "tech lef" file (.tlef)? If so, I will be publishing them shortly too.

mithro avatar Jul 02 '20 20:07 mithro

I think the .tf can be generated from the .tlef but I'll have to look into how to do it.

kkiningh avatar Jul 02 '20 22:07 kkiningh

Initial .tlef files where added in https://github.com/google/skywater-pdk/commit/4cbdbb5a735a8acd53c6247eb2bc23b9245329f8

mithro avatar Jul 06 '20 14:07 mithro

@kkiningh - See also https://github.com/google/skywater-pdk/issues/62

mithro avatar Jul 06 '20 16:07 mithro

I believe by tf you mean Milkyway tech file, information may overlap with TECH tlef. For parasitics you will still be missing the ITF file, needed to generate TLUplus used as parasitic lookup tables. Maybe you can try Wire Load Model to start with.

20Mhz avatar Jul 07 '20 15:07 20Mhz

@20Mhz, yeah that's correct. I believe Milkyway supports a way to create a Milkyway library from a tlef and then exporting a techfile, although I've never tried it. You're right that the itf (from which the tluplus file can be generated) will still be missing.

@mithro I'm super busy for the next couple days, but I'll try to get around to trying to do this and report back soon.

kkiningh avatar Jul 07 '20 15:07 kkiningh

I believe LEF does not have enough info to automatically generate a mw tech. A fair work around could be to get an OpenAccess db and use Custom Compiler exporting capability to produce a mw tech as a starting point. -- EDIT: I was able to read in LEF and output tf in Milkyway, just had to set a dummy lib, haven't tested the file yet

20Mhz avatar Jul 07 '20 21:07 20Mhz

FYI - Full technology LEF files for the standard cells where merged in https://github.com/google/skywater-pdk/pull/68

mithro avatar Jul 09 '20 02:07 mithro

@mithro tech LEF is generally "good enough" for use in place-and-route tools for older nodes, e.g. 130nm. However, generally for timing signoff, designers would want to run a signoff-quality parasitic extraction which includes coupling capacitance. .itf, .qrcTechfile, etc. These checks are important for final power signoff, signal integrity signoff, and timing signoff.

rovinski avatar Jul 09 '20 02:07 rovinski

I'm now blocked by missing tech to GDS mapping.

20Mhz avatar Jul 10 '20 18:07 20Mhz

FYI - There is a spreadsheet at https://docs.google.com/spreadsheets/d/176OcP2h_TT_pDP2-zBLijlgFUP38LVgMQPlDtpLcrsk/edit which includes a lot of information about the layers.

@RTimothyEdwards has a bunch of useful resources at https://github.com/RTimothyEdwards/open_pdks (which will be eventually merged into this repository).

mithro avatar Jul 10 '20 19:07 mithro

@mithro could you please tell me how to get .lib files from .lib.json?

MohamedAliYounis avatar Mar 05 '22 20:03 MohamedAliYounis