Sorry it took so long guys. The system has been running cool and quiet enough for me, so undervolting hasn't been a huge priority.
But after a bit more research, I have discovered that RMClock 2.2 does in fact support Brisbane. You have to enable half FIDs in RMClock, and oddly enough actual Vcore seems to be approx 0.1V higher than its RMClock setting, but it still works (and RMClock has settings down to 0.7V, so it is possible to obtain an actual Vcore of 0.800V, although I would get crashes even at idle with such a low voltage).
This chip is an excellent undervolter, though. I haven't done any long term testing, but it didn't seem to have any problem with dual Prime95 for a few hours at 0.975V actual. This is at the full 1.9GHz, which is very impressive IMHO. Will dig out the power meter some time and measure load power consumption.
Until then, however, we can make some assumptions. This undervolt should correspond to a power consumption roughly 56% of the original. Assuming the processor has a full load power consumption of 50w (which is just a number I've pulled out of my rear, but seems reasonable, if anything a bit pessimistic; it's probably a bit lower, 40-45w) you're looking at a sub 35w processor, more like 28w. Not too bad at all for less than $100 now I'd say. If you want the best bang for your buck C2D is probably the best bet, as they can overclock like crazy without needing a lot of volts (although I am not going to bother overclocking the X2 3600+, based on what I've read they have a decent headroom at stock volts, but don't hold a candle to C2D in this department). If you just want dual core performance at a low cost and without a lot of heat, this processor is for you.