Okay, preliminary results are in:
This board(VNF3-250) overclocks/overvolts by default. Stock voltage is 1.42, 1.45 vcore is 1.49 vcore actual, etc. I'll make note of this.
RAM is set to 2:1 ratio(DDR200 @ HTT 200) as per Zebo's suggestions.
HTT 265: Max I could reach while using HT multiplier of 2x. Raising the multiplier to 3x enabled me to press onward.
HTT 290: Max I could reach at 1.42 vcore while still stable.
HTT 300: Required 1.475v vcore(1.52 actual) to be stable. I primed here for 18 hours straight with no warnings or errors.
HTT 310: This remains an unsolved puzzle. I bumped vcore all the way up to 1.525(1.56-1.59 actual, depending on which monitor program you believe), but it failed to stabilize well enough to run Prime95 torture test 1 for more than a few minutes before encountering a rounding error. Bumping chipset voltage up to 1.8v did not correct this problem. I've noticed that, at 310 HTT, my previously-stable voltages are now jittering around a lot. It could be the board is having problems with voltage regulation, or it could be the CPU just doesn't want to go this fast. I'm not really sure which. In any case, I've backed off from this setting, and will return to HTT 290 so I can run the chip at stock vcore.
Notes:
1). Bios misreports chip speed at boot past 2 ghz. CPU-Z is not fooled.
2). 10/19 BIOS for the vnf3-250 does not permit vcore adjustments for my Sempr0n 2800+. The 9/21 BIOS does.
3). As mentioned above, the system will not POST fully(goes into BIOS-induced "safe mode") past 265 HTT with a 2x HT multiplier. 3x is stable. Weird.
I will be playing with the RAM from here on out, but I doubt that is of much interest to anyone here, so that's about that. The final limiting factor is either the CPU itself, or the VNF3-250's voltage-regulation abilities. The mere fact that it posted at 310 HTT was impressive to me, and WinXP was stable. So was Super-Pi. Prime95, however, was not. Boo!