@ OP, my ES is a G0, not a B3. And her MAX stable on WCG for 2 weeks was 4ghz. This was all under water, on a Gigabyte P35-DQ6 running Crucial Tracer PC8500 on a Antec Quattro 850 PSU.
My regular B3 Q6600 is at 3.6ghz under water. And she has been doing WCG for 1.5 month straight with very little downtime. The last downtime was because i burnt my board out. EVGA replaced it with another broken board, and thats when i got fed up, and replaced that board with a Asus P5K-Deluxe.
My X3220 is on a Gigabyte P35-DS3P, and is currently on WCG crunching @ 3.375 for 2 months.
My X3210 is at 3.2ghz 400fsb x 8 and she's been doing some major photoshop editing for a friend of mine, as well as pulled some WCG time on it.
My clocks are stable. There not benchmark settings, there full WCG loading settings. But mark i have to agree with you on some parts.
1.5Vcore+ <Reported in ASUS Probe> which tiramisu is pushing on his quad for 3.8ghz on AIR, has to be the stupidist thing you can do for a 24/7 loaded quad. Hell, the one thing that hasent stopped me from posting my 4ghz full loaded WCG screenie is because ive gotten lazy to copy and paste it, and then load it up on my Photobucket.
Also my chip is an ES, so theres no point in posting it for people to compare off of. But i have to agree with mark on this case. I would like to see a quadcore higher then 3.6ghz pull a Prime95 longer then 6 hours for it to be deemed a stable overclock.
I dont think tiramisu can pull off a smallFFT prime95 longer then 45 min before he starts to smell his TIM burning.