skywater-pdk: cell_footprint value in liberty files doesn't match name in verilog or GDS.

The example in the document seems to be;

cell_footprint : 5MIL ;

Use this attribute to assign the same footprint class to all cells that have the same geometric layout characteristics.

Cells with the same footprint class are considered interchangeable and can be swapped during in-place optimization. Cells without cell_footprint attributes are not swapped during in-place optimization.

When you use cell_footprint, you also set the in_place_swap_mode attribute to match_footprint.

The way I read this is that all cells in cell_footprint group will be considered for replacement if their footprint matches.

If two cells have different cell_footprint values even if their footprint otherwise matches they can not be swapped.

Said another way, if you have two and4 gates with the same footprint then they are only swappable if their cell_footprint value also matches.

Does that make sense?

I feel like the doc is conflicting. Under 1.6.11:

The in_place_swap_mode attribute specifies the criteria for cell swapping during in-place optimization. The basic criteria for cell swapping are:

• The cells must have the same function.
• The cells must have the same number of pins, and the pins must have the same pin names.

So nothing is mentioned about layout geometry. But in 2.1.2:

Use the cell_footprint attribute to assign a footprint class to a cell. Use this attribute to assign the same footprint class to all cells that have the same geometric layout characteristics.

So cells are expected to have the same geometry.

Like I said, I saw different commercial PDKs both adhering and not adhering to the “same geometric layout characteristics” criterion. I would suggest using whichever approach results in the fewest warnings / errors in commercial tools. I wouldn’t doubt that this property is simply ignored most of the time.

@rovinski I meant that since the cells are organized in a way that one cell has all its drives and views in one folder, it may be misleading. So I added that for Tim’s reference. And yes you are right.