Note that Assura tends to lump together resistors far more than Diva does - Diva often leaves lots of small resistors in a complex mesh, because it does only limited reduction (and even then, it will only reduce if it can make the network smaller with the same overall resistance, and it will only combine resistors which came from the same layer - if my memory is correct).

So you may not be comparing like with like?

Any concerns you have over accuracy of the extraction rules should go to IBM, I'd have thought.

Regards,

Andrew.