I was just thinking about how the interface was streamlined for the PS3 and that got me to thinking. I know someone who knows about the architecture of the PS3 and it can operate very fast under one condition; basically it can do 8 things at once but these things have a very small amount of memory available to them... 256k I think. This however makes the PS3 harder to program (small memory constraint plus you have to manage data handling well). As an OT, this is probably why the PS3 version had to be delayed (trouble working within constraints) and why that version was going to have slightly worse graphics than the PC version on a high end PC... and why PS3 graphics tend to be a hair worse overall despite a more powerful engine; it's more difficult to program for than a 360 and most programmers, at least in the US, are lazy.
Back to the main question, was the code for the PC version originally optimized more for the PS3, its requirements, and a "simple" port in mind or not? If SE optimized how the code is broken up and handled on the PC with a PS3 in mind (like the interface which should have been PC optimized from the beginning wasn't) it would be interesting to see if that is affecting performance of the PC version any.