i ran my vcore all the way up to 1.62v
however anything above 1.5v seemed to make no difference
and temps would increase quite fast with voltage increases (more so than clock speed increases in my experience with the prescott)
i would say 1.6v would be the safe limit for me, but i havn't seen anyone burn them yet so it's really anyone's guess
the northwoods seemed to enjoy the 1.5-1.7 area (most would stop at around 1.65 or 1.67 around here)
i would expect the .09u prescott's to burn up atleast .1v lower than the .13u northwoods (maybe even .2v+ lower when compared in the right situations)
if you wanna play the safe ground 1.52-1.57 would be a nice range to work with
good luck
edit: also note that the gate size has been shrunk about 30% this means that the actuall guage of the material carrying the current has decreased in size/mass/etc by 30%
i'm no electrical engineer but if i had to guess [knowing that certain technologies like strained silicon, better manufacturing processes, increased effort to decrease dissipated/wasted electical current (thus heat and the ability to transmit more efficiently), etc. have been invested heavily in for this transistion] I would atleast suppose a %15-%35 reduction in the theoretical safe vcore maximum for prescott when compared to northwood [10% decrease by intel on the default voltages (roughly), thus showing the wasted potential when actual is compared to theoretical efficiency].
This would equate to approxiametly 1v being the default when compared to the ~1.55v that most northwoods use [this would assume however that 1.55v is the best/lowest that the .13u could go and that the .09u would increase in efficiency from that by an exact % which correlates to the reduction in diameter of the gates (assuming the metals/chemicals/electrical strengths of the materials useds/etc. remain roughly unchanged)].
However we are overclockers (know what's going on to some extent too) and so is intel/amd so we know that theoretical limits aren't the truth with these things (obviousely prescotts wouldn't hit 3ghz with 1v in intel testlabs with atleast 60-90% consistency, if at all) or they would be making them now.
I would judge that about 1.6v would be the maximum if you considered 1.65-1.67v to be the maximum for your northwood. And possibly 1.64v if you considered 1.7-1.72v your limit with northwood.
I think prescott's will fry up and burn with over 1.7v on air cooling. I just don't see them living much more than a month or two at that voltage (this was default for many .18u cpu's if i remember right).
With the smaller process you also run the risk of hitting the voltage limit much quicker when raising the voltages by .1 or .05 increments with as much as .02-.06v sways in voltages on most motherboards/psu's these days it'll be a larger change in % of difference in chip performance/death for each increase in voltage (higher risk chip to "max-out" imo).
In other words, I believe 1.5v to be safe and 1.6v to be good for atleast a year if you know what yur doing and have non-standard cooling.
Anything higher is like firing a shotgun at a Gas station, either yur gonna miss and everyone feels lucky to be alive or your gonna hit a gas tank get that exilerating thrill then blow up and the funs over
good luck
Facebook
Twitter