open-source-silicon.dev

The klayout lyp file colors are :face_vomiting:

<@U03UV5NUV6Z> has also complained about this and wants a standard color scheme implemented for all PDKs.

<@U03UV5NUV6Z> also wants me to fix the color scheme in magic, which yes, could probably use an overhaul, since the metal stack is all based on a pretty ad-hoc system I came up with the first time I had to implement a 9-metal stack for a 90nm process.  (Also, unrelated to gf180mcu, I waffled a bit about whether local interconnect in sky130 should be presented like metal1 or something else.  I chose to present it like metal1 but I think the better choice would have been "something else".)

We should put this out for discussion. I've got some strong biases based on my old experiences but I'll adjust if there is strong input on color scheme for the community

My biases were set by the magic tech files for MOSIS SCMOS in the 1990s.  Consequently, my biases end at around metal 3.

What I'd like to ensure is that we get patterns and overlays and in general try to bias to a black background so that colors can be more readily discerned. The white vs black environment background is like a step one for the color scheme as it determines the right colors for good contrast at lower brightness.

The end goal should be readability.  You should be able to look at a layout with a stack of five metals and be able to see where everything is, including all the components underneath (from about five metals upwards, it gets really messy no matter how you do it).

The old scheme we used was: LI is gouda yellow. MM1 Navy blue, MM2 is close to Barbie Pink but slightly darker. MM3 Sky blue. MM4 Burple, MM5 Yellow. MM6 and above RDL layers I'm open. Metal layers have a sparse fill. Contacts have a higher fill pattern but transparency in a slightly darker version of color above them. Mcon was unique in that is was grey. Licon was   brie yellow with low transparency.

I submitted a PR to the PDK with BEOL colors similar to this. Also:
poly red
diffusion green
nwell tan?
Pwell purple?

<https://github.com/google/globalfoundries-pdk-libs-gf180mcu_fd_pr/pull/85#issuecomment-1378170347|https://github.com/google/globalfoundries-pdk-libs-gf180mcu_fd_pr/pull/85>

<@U03UV5NUV6Z>:  That follows closely to what I was thinking of originally, which is to cast local interconnect as looking more like a 2nd poly layer than a 1st metal layer.  That probably also helps prevent designers from thinking about and using local interconnect as a routing layer.