Closed rodrigowue closed 7 months ago
Looks like the SKY130 LEF file is of a newer format than the parser can handle. Can you make your sky130_fd_io.lef file available somewhere?
Dear trcwm,
I am doing everything via Google Colab https://colab.research.google.com/drive/1a-eBXsQymqJ7C8_wEMNFpTOb9zGQ52WF?usp=sharing
I've uploaded the LEF file that comes with the conda-eda sky130A install and it is available here: https://drive.google.com/file/d/1bsxI78wYewZh71WOuFOwMURsn1DecbAN/view?usp=sharing
Best regards, Rodrigo
Thanks Rodrigo,
The SKY130 can now be read by padring. You still have to provide filler cells, which seem to be absent from the .lef you provided.
Please let me know if it now works for you.
Regards, Niels.
Thanks a lot! Now it seems to be working. But I don't think they released the io-fillers yet for sky130. At least I was not able to find it.
Best, Rodrigo
Actually, I was able to find the proper LEF with corners and fillers by following this pdf: https://docs.siliconcompiler.com/_/downloads/en/noah-examples-cleanup/pdf/
However, unfortunately I am facing another issue now.
[...]
[INFO] 37 cells read
[INFO] Found 37 filler cells
[INFO] Die area : 3000.000000 x 3000.000000 microns
[INFO] Grid : 1.000000 microns
[INFO] Padring cells : 39
[INFO] Smallest filler : 1.000000 microns
[INFO] Writing padring to SVG file: padring.svg
[INFO] Writing padring to DEF file: padring.def
[INFO] Writing padring to GDS2 file: padring.gds
[VERB] GDS2Writer created
[WARN] DEF database units not set! does your imported LEF file specify it?
[WARN] Assuming the value is 100.0
[ERR ] Cannot find filler cell that fits remaining width 0.045000
Maybe should I allow the overlap of the minimum size IO filler? I am not able to open both .svg and .gds. They are broken.
Padring complains that there is a gap of 0.045 it cannot fill. You probably need to make the die area a multiple of the minimum size specified by the PAD SITE. Not sure what that is for SKY130; the LEF file you sent me does not declare it.
I think the correct LEF is the ef_io and not the fd_io I was using before. (or maybe I should use both). both here: https://drive.google.com/drive/folders/14EOiY9i91GHFyLOgfrw_k2vSAPfh5qW1?usp=sharing
But I am keeping getting errors even while using 4500 microns. I tried to change the grid but it does not seem to fix it.
[INFO] 37 cells read
[INFO] Found 37 filler cells
[INFO] Die area : 4500.000000 x 4500.000000 microns
[INFO] Grid : 0.005000 microns
[INFO] Padring cells : 40
[INFO] Smallest filler : 1.000000 microns
[INFO] Writing padring to SVG file: padring.svg
[INFO] Writing padring to DEF file: padring.def
[INFO] Writing padring to GDS2 file: padring.gds
[VERB] GDS2Writer created
[WARN] DEF database units not set! does your imported LEF file specify it?
[WARN] Assuming the value is 100.0
[ERR ] Cannot find filler cell that fits remaining width 0.045000
Nevermind, I was able to work with 3600 x 3600 :)
But I don't have any clue why it looks so wobbly... It looks like some cells have different widths
Yeah, the cells don't have the same height. Perhaps you're mixing two different PAD libraries? Or maybe it's by design - hard to tell without looking at the combined GDS2 in klayout.
Hi!
I am trying to use the padframe_cfg in the openlane flow, but prior to that I am trying to generate the padring manually using the tool of this repo.
I keep getting this error:
My CFG file:
And the run script:
If anyone know how can I use the io cells from sky130 it would be very helpful.
Best Regards, Rodrigo