do we expect that magic detect off-grid errors during DRC? my cell has many but magic detected none. are these in the klayout deck instead perhaps?

No, Magic doesn't detect off-grid errors. Its the Klayout which detects off-grid errors

the offgrid in klayout rules are https://github.com/efabless/globalfoundries-pdk-libs-gf180mcu_fd_pr/blob/main/rules/klayout/drc/rule_decks/geom.drc

@Tim Edwards is it correct that magic can't detect off grid violations?

@Matt Liberty: Not exactly. Magic cannot represent off-grid violations in its database except for rare cases of non-manhattan geometry overlapping between subcells. Both cases are detected---the non-manhattan overlap is flagged directly as a DRC error. Off-grid positions that are forced to snap to an integer database value will generate an error during GDS read-in, but is not considered a DRC error per se.

@Tim Edwards I don't mean a non-integer coordinate but off the manufacturing grid which is 10dbu in skywater

great to know, thank you!

@Matt Liberty: What I mean is that magic's database is scaled so that the smallest integer unit (1) is the minimum manufacturing grid (5nm), such that nothing in magic's database can exist on a non-integer point (other than the case I mentioned above).

@aryap did you create this layout in magic? @Tim Edwards that's an odd choice - what is the advantage of that?

@Matt Liberty: Read John Ousterhout's white paper on the corner-stitched database format. It makes sense for a tessellated plane database.

I read it years ago. I know we wrote a tessellated grid-less router at cadence using dbu.

@Matt Liberty no i only used magic to verify

How did you generate the layout?

my tool: https://github.com/growly/bfg