open-source-silicon.dev

So on TT09 we have a user that has a customized OpenRAM macro using the `sky130_fd_bd_sram`  cells. The klayout `sky130_mr` FEOL / BEOL checks are OK but `magic` DRC is throwing errors. Is that good enough to be accepted ? ( cc <@U016EM8L91B> )

Yes.  We still do not fail a tapeout submission on DRC errors from magic.  We are in the process of defining the set of DRC rules that we expect to be followed (more relaxed than the entire set of rules, but much more than the "manufacturing rules"), and fixing the old klayout DRC deck to make sure we have a robust system for checking them.  But that's still in the future.

And is it expected that magic fails on those cells ? I thought it had built-in exception for them ?

There are exceptions for the existing SRAM macros by name;  the entire macro will be replaced by its abstract (LEF) view before running DRC.  I have advocated for having the base cell `sky130_fd_bd_sram` similarly abstracted, but I don't think anybody has ever taken action on that (attn:<@U06FKCF4E5T>).