https://open-source-silicon.dev logo
Title
a

Adrian Freed

09/16/2020, 3:59 AM
@User You are of course free to use anything of my documentation. I plucked the description of each cell from a json file so most of the rest of what I have done could be considered editorial. I sorted the cells by type/use as best as I could (some still have ??? as descriptions). Category information is best perhaps done as tags in the json files. Another editorial thing I did was to flag cells useful to be co-opted for analog circuits. The number of these has grown since I started researching some of the creative uses people have made of digital cells. Another thing to signal in documentation derived from the cell data is the number of outputs a cell has. It’s the sort of thing that “everybody” knows but is useful for new explorers. For example, the openlane tools won’t synthesize with multiple output cells like the adders. It would be useful to flag which cells are actually likely to be used by which toolchains. I have browsed the newly released models. Most of them don’t have easily interpretable human metadata (i.e. comments) but many of spice files for the capacitors have useful comments that could be moved into json files. I didn’t see any json files for the osu cells. This would be handy - again most experts know how to align the nomenclatures developed over the years for the cells but it is arduous for newbies. Once these things are smooth, this PDK availability work (for which I am very grateful) offers up some great learning opportunities. For example, the spice models of the OSU cells have a resistor network along with the transistors that aren’t in the corresponding cells of the skywater libraries. Other fascinating questions that are productive for people to research from the documentation is “why so many cells?“. Why so many libraries? The literature I have been reading is full of contradictions on this as the tradeoffs have changed over the years.
✔️ 3
👏 4