Don't know why I never tried this. Anyway, better late than never.
On my quad core I put the server on CPU0 and BF2 on CPU2 (Affinity util recommended). I ran Libisey with 64 bots and x2 anti-aliasing.
Here are the averages from FRAPS.
BF2 ONLYFrames, Time (ms), Min, Max, Avg
6869, 194384, 0, 101,
35.337SERVER & BF2Frames, Time (ms), Min, Max, Avg
24356, 476502, 0, 96,
51.114Looks good but this is the first time I've done this. Num of frames is different between the tests and frame rate changes depending on built up / open area. I'd like to run a few more tests.
This is good advice anyway as I think I have a performance increase running BF2 on it's own (without the server) on CPU2 as I think the system as background tasks on CPU0. Why wasn't I doing this before
.