1009 with a 64MB 9100 | XP1700+ | nF1 | 512MB--figured I'd get a bit more, but hey, I realize I'm not going to be playing anything too taxing too well on this setup.
As for nVidia not "optimizing" their drivers for the Doom III scores Anand posted, did you miss the fact that Anand ran his benches at nVidia's expense on an nVidia system (and almost using an nVidia demo, though iD changed that at the last minute)?
(And not to rain too much on your parade, but I'm still not convinced any FX 3DM03 score is a real one. You never know, the FX architecture may be so convoluted or high-maintenance that it took nV this long to fine-tune the drivers to maximize the hardware--but I doubt it, given the blatant cheats ET exposed and nV tried to litigate into legitimacy. Heck, we've all seen that the reason GF3/4 and 8500 scores jumped so high in 3DM2001SE was because they swapped their own shaders for 3DM's, which is a cheat in the scope of that benchmark. The bench *may* be irrelevent to some, but the cheat certainly makes any score obtained with it irrelevent as far as anyone is concerned. Honestly, I'm convinced first-generation DX-whatever hardware will usually be too slow for games that finally get around to fully exploiting DX-whatever, so it's pointless to even bench it. DX9 hardware may be the exception/breakthrough, though, with HL2 set to use PS2 and run well.)
Wait a sec--are you sure you scored 6000 in 3DM03v330 with the 44.03's? I thought v330 cut off nV's optiizations in that particular driver. You sure you're not using the 44.65's or .67's?
ExtremeTech scored much lower than you with the 44.03's.
Edit - NM, ignore that last link--ET was benching at 12x9, for some reason. Their 4800 at 12x9 can easily turn into your 6000 at a lower res and on OC'ed hardware. I'm still puzzled about the driver and 3DM03 combo, though.