Consolidation performance with empty rule file

Hi,

I am currently reviewing where the consolidation performance of our classic 11.1.2.2:500 application can be improved.  To test how much rules contribute to the consolidation times I have loaded an empty rule file in the system.  What I notice however is that on the first consolidation run (CAWD) that the system is about a factor 3 slower than when I run a CAWD a second time.  As the rule file is fully empty I wondered where this drop in performance could be coming from as all other things are equal.

With the second run I see that depending on which hierarchy I consolidate that the without rule file option is between 2.4 and 3.0 times faster.  Is there any benchmark data available for this?