The problem with netgen is that it was running for hours on digital standard cell layouts, and it has to do with the symmetry breaking routine. The fast symmetry breaking solves the problem of running for hours and hours, but it causes LVS to fail (with rather cryptic errors) on designs with more complicated symmetries than just a lot of antenna cells in the layout. The "-full" switch returns to the slow but exhaustive symmetry breaking method. Sometime soon I will go looking for a way to automatically determine which method to use.