Quick input that might make you consider re-testing F4 with these processors.
It has come to my attention from credible sources, although not in any deep specifics, that the microstutters, macrostutters and other frame rate killers of F4, which were present in 1.07, have been greatly eliminated in 1.08. In particular, certain effects related to sounds were resolved (again, as my source tell sme).
The reason that this may be relevant to your testing is that the more erratic, non computational based, sound file and memory leak problem sare removed, the more likely it is that the tests you did will be more meaningful.
As most will tell you, many of these frame rate issue seen in Single player (and MP) F4 are all but gone now in 1.08, and the memory leak is nearly non existent.
To me, this means that if you repeat your tests bwtween these processors and F4:AF 1.08, you might get far more valid and discernable results. I hope you do. I will be very curious. Why? Becuase LP has clearly hinted that this code base will be the chassis for a new GFX engne and other upgrades in a newer release. So your reults in a cleaner version of F4 will be far more meaningful fo rme, who is currently thinking about building a new gaming system soon.