Difference between revisions of "Whom do you trust?: Validating process parameters for open-source tools"
Jump to navigation
Jump to search
TimEdwards (talk | contribs) (Added link to uploaded slide presentation.) |
|||
| Line 3: | Line 3: | ||
==Downloads== | ==Downloads== | ||
* [[:File:edwards_fsic2022_slides.pdf|Slides]] | * [[:File:edwards_fsic2022_slides.pdf|Slides]] | ||
* [https:peertube.f-si.org/ | * [https://peertube.f-si.org/videos/watch/657e7325-d9fb-461b-88d0-b8ec97213e71 Video recording] | ||
==Abstract== | ==Abstract== | ||
Latest revision as of 23:00, 1 August 2022
- Speaker: R. Timothy Edwards
- email: tim@opencircuitdesign.com or tim@efabless.com
Downloads
Abstract
Foundries often provide important process data in formats that are proprietary, often not human-readable, and restricted in use to specific commercial EDA tools. As a "case study", I will discuss the long journey of obtaining trusted values for parasitic capacitance in the SkyWater sky130 open PDK. I will review how I obtained vendor data, cross-checked against an open-source field equation solver and other sources, and updated the parasitic extraction methods in the Magic VLSI layout tool to incorporate a number of effects that were previously not accounted for, and how Magic's extracted output can be used to generate data for other open-source EDA tools such as OpenROAD.
Software
General information
- Repositories: https://github.com/RTimothyEdwards/magic https://github.com/RTimothyEdwards/open_pdks
- Main documentation websites: http://opencircuitdesign.com/magic http://opencircuitdesign.com/open_pdks
Roadmap
- The project seeks help on: Curve fitting of vendor data to automatically generate coefficients for magic's analytic expressions of parasitic capacitance.