At a high level: pycells are parameterized cells that are written in the Python language, these require technology from Ciranova (now Synopsys) and are not supported by Cadence, whereas PCells are written in SKILL (or object oriented SKILL, known as SKILL++) and are supported by Cadence.  PCells are well tested in the Cadence tools and tool flows, but due to the nature of pycells being third-party software, the same cannot be said there.

Hopefully this answers your question without being too 'political' (today of all days), so hopefully you can 'cast your vote' accordingly :-)

Lawrence.
Q:

Thanks for your reply Lawrence.

For the end customer or tapeouts , we need to give gdsII database. Does this make any difference in final GDSII using pcells or pycells.

If the GDSII database is same for Pcell or Pycell. Then we can use pcells without any confusion as we have enough experience earlier.

Thanks.

Srinivas.
A:

Srinivas,

Whether you use pycells or SKILL pcells, most foundry PDKs would aim to have the same layout regardless of which was used.

However, using non-SKILL pcells means that:

1. If there is a problem with the tools, we cannot provide support since we can't test and debug the issue (we don't have the source code). So there is greater risk if you are using non-SKILL pcells, because we may not be able to solve a problem because of inability to debug/test with your data.
2. Some functionality is not available when using non-SKILL pcells (e.g. modgens, placement, PVS running inside Virtuoso).

Regards,

Andrew.

from:https://community.cadence.com/cadence_technology_forums/f/custom-ic-skill/24084/pcell-vs-pycell