@Tim Edwards Hi Tim, we left io pins that we don't use to be unconnected, then our user_project_wrapper_analog.gds in klayout would have floating metal error basically in every unused pins which are unconnected, which leads to fail

Klayout Metal Minimum Clear Area Density

step when

make run-precheck

. I'm not sure how to deal with those unused pins to avoid this error. Here is our repo https://github.com/YuetingLi666/caravel_user_project_analog_BGR, you could find our

./gds/user_analog_project_wrapper.gds

and

./netgen/user_analog_project_wrapper.spice

there. Looking forward to ur suggestions!

Are you sure that the pins cause the density error? That doesn't make sense to me. To my understanding (and sometimes things change without me knowing about them), the klayout check is done without fill generation to check if any metal is already over-density (which then could not be fixed by fill generation, which can only add more metal and increase density). The fill checks should simply be enumerating all metal shapes in the design and dividing the total area of them by the total user project area to arrive at a density figure.

@Tim Edwards if our design is too small, is it possible to lead this error? If u don't mind, u could have a look at the gds, that it only takes less than 1/10 of the whole caravel area

You can clone the precheck repository from github/efabless and run the klayout script that does density checks on your own design, although in theory, you will get the same result as precheck gives you. You can also just eyeball the layout while displaying one metal layer at a time. Note that it is very difficult to get a metal layer to be over-density; there is only one common case that we know of, which is synthesized digital circuits that take up the entire user space and are dominated by decap cells using the original SkyWater decap cells. I do not know of any other instance in which metal was over density.

@Tim Edwards I passed now. I found for user_project_wrapper_analog repo, the errors from two step ['consistency', 'min density area'] are actually corelated. the consistency step would try to match the

gds/user_analog_project_wrapper.gds

with the

netgen/user_analog_project_wrapper.spice

. If the .spice have many hierarchies, the 'min density area' would take some hierarchies as top hierarchies and try to check the subcell metal density. I tried to flat my whole design firstly into

user_analog_project_wrapper.mag

and then use gds write to get

user_analog_project_wrapper.gds

to avoid this issue, then extract this gds into netgen/.spice, this method works for me.

@jeffdi: Take a look a the comment above. This behavior makes no sense whatsoever!