I should also note that DF runs slower if it's switching between cores. Significantly slowly. IT IS single-core, after all.
I set it to the second core strictly, and use Process Explorer to set EVERYTHING else to the other two cores, so DF gets a core entirely of it's own.
I also set it to high priority. (Realtime works too but makes it hang on exiting)
It's important to turn OFF Cool'n'Quiet or other power-saver features, and if you overclock, overclock ALL the cores- using my trick above, DF does seem to outsource some of the CPU load (probably for sound and graphics) and so those cores matter quite a lot.
Also, memory with tighter timings seem to work better than just faster memory. I can clock up to 1320 MHz on the memory but it actually makes the performance drop, not improve. Suggesting that DF is dependant on an ungodly amount of very, very small reads and writes.
It should also be noted that even my FPS drops to nil once there's a few dozen dwarves. 7 is a piece of cake, but once I get to 100 dwarves, i'll be lucky to hit 30.
The disparancy between my performance and everyone else's is likely just an issue of data throughput. So anyone else should be able to recreate it. Someone with a larger cache and similar IPC and frequency should perform better. But once you get a lot of dorfs, that's when the great equalizer strikes.