@User, @User: There are a number of possible solutions here. (1) For one, this problem happens only when the design has already essentially passed LVS. That leaves the possibility that the program could just exit and not do the final symmetry breaking, other than the matching against property and against pin names. (2) There is an unoptimized loop in the pin matching part that can cause a slowdown. It is easily fixed by string hashing. I just didn't think it was the longest delay in this case, but I could be wrong. (3) If there is any evidence that there are significant delays before the symmetry breaking starts, then I'd like to know, because I have been concentrating on the latter since that is where I'm seeing the delays. If someone can post the two netlists from an example with a long run-time, that would help me debug.