Archived from groups: alt.comp.hardware.overclocking.amd (
More info?)
"Steve Wolfe" <unt@see-signature.com> wrote in message
news:2t2fluF1qgjidU1@uni-berlin.de...
>> Thanks guys - I am currently undervolting my athlon (using standard
> 133x15)
>> and I was trying to figure out if dropping the volts from 1.65 ro 1.5
> would
>> actually generate significantly less heat or not!
>
> 1.5/1.65=0.909, so you can figure on about an 11% reduction in heat.
I would think it would be more than 11%. For the purposes of heat
dissipation we can consider the CPU to be a resistive load. And since P =
V**2/R, a reduction from 1.65 to 1.5 will yield a 1.5**2/1.65**2 reduction
in heat. Which is 17.4% Of course the CPU is not purely resistive, but
this is still a good approximation.
>> I figure that at 2000Mhz power usage is (as above) 68.3W, but dropping
> from
>> 1.65v to 1.5v reduces power intake to 56.4W - 17.5% less power, therefore
>> 17.5% less heat generated. Does this sound reasonable?
>
> It does. Not that the two numbers don't match up (11% vs. 17.5%) because
> the 17.5% number obtained by comparing the theoretical maximum value to a
> real-world value. Had you measured the real-world usage before and after,
> the difference should be much closer to 11%.
Or 17.4% in fact.
> If you really wanted lower power, you could go with a mobile Athlon XP,
> where the max power draws can be as low as 35 watts, and the "power now"
> featured (I believe there's at least one desktop board that supports it)
> will keep actual draws much, much lower when not under load. However, it
> does sound like you've done an outstanding job with what you've got!
The mobiles are good chips, but they only use less power because they run at
lower voltage, and you can always run a normal XP at lower voltage too. As
I said in another post, I ran a Barton 3200+ at 1500MHz at only 1.2v with no
heatsink fan for a couple of months without problem.
Chip
Facebook
Twitter
Instagram