My point is that everything is less stressful than prime pretty much... Rosetta included. Today a guy told me that Prime95 fails on his computer almost instantly, yet it's been running Rosetta for 3 days straight. Only a few stress testers can equal it. CPU failure is due to heat. We know this - think phase change cooling - same voltage gets you more Mhz, more Mhz requires less voltage. If the CPU errors out when it hits 53c, and it only hits 53c after 3 hours of prime on a hot day... then nothing else will make it fail. I especially noticed this on my CCB1E 0550VPMW. The AMD CPUs with coldbugged memory controllers (Bx where X >= 1) tend to be very temperature sensitive. At every voltage setting there is a key CPU temp where anything over that temp fails, and i found that only prime, not my normal activities, would hit those temps.
So say it can prime at 3 Ghz but then 3.02 fails right... load temp is 46c say. I load up dual Super Pi 32M and it only loads at 41c, and it completes at 3.02, even 3.03. I open the window to the room so that Prime95 only loads at 41c - and it doesn't fail at 3.02 anymore. It only fails because it gets hotter than everything else.
So, i could clockgen to an unstable spot, where Prime95 would only run for like 5 minutes. Most people would call that very unstable, but it actually runs all windows stuff fine, i'll leave the computer on overnight at "unstable" settings and it's good. You have to think outside the box, prime isn't everything.
To the OP - sorry for the hijacking. 0608MPMW is a great stepping, 3 Ghz on air is nice and above average.