On Sunday 10 April 2005 03:50, John R Pierce wrote:
> steve elias wrote:
> > it's possible that running your
> > PC in a very cold garage or basement would result in less power
> > consumption since the ambient temp would require less cpu fan
> > operation.

Interesting idea - but is the effect going to be great enough to offset the 
extra maintainance costs due to rapid aging of mechanical components operated 
below their rated temperature range? I very much doubt it!

Beware of high humidity, which often goes with cold garage/basement type 
environments. Condensation can cause fires as well as malfunctions.

Desktop PC systems - and components like PSUs & HDDs - are usually rated for 
oeration at ambient temperatures no lower than 5C and humidity not exceeding 
90%.
>
> CPU fans tend to be 5-10 watts.

Where do you get yours from? Typical ratings are 0.05 - 0.25 amps at 12V; the 
beefiest in my posession are rated at 0.25A = 3W.
>
> I'd bet its equal parts the CPU and the memory (which is getting cycled
> pretty much constantly at bus speed during the LL tests)

Dynamic RAM depends on being accessed every few milliseconds in order to 
maintain its contents - refresh cycles - and has no standby mode. Power 
consumption doesn't vary significantly however hard the processor drives it. 
If you don't believe me, tack a thermal sensor to the surface of a DRAM chip 
and observe the (lack of) difference between Prime95 running and "idle". In 
any case, except for P-1 stage 2, the proportion of system memory used by 
Prime95 will be small to insignificant.

On the other hand, it is well known that modern CPUs can and do "turn off" 
unused execution units, with a very significant saving in current. They can 
do this because they are based on bipolar logic, which (unlike DRAM) can be 
run at low voltage with a slow or zero clock speed without losing data 
contents and can therefore be "turned on" again instantaneously when required.

CPUs with "speed step" technology - which includes _all_ Pentium 4s - can 
even sleep between clock cycles when the clock speed is reduced, with a 
proportionate reduction in power consumption. So much so that a P4 system 
will run without _any_ heatsink - just very, very slowly as thermal 
throttling reduces the clock speed to kilohertz rather than gigahertz! Try 
this on a processor without "speed step" technology and you will have a fried 
CPU, maybe even a hole burned in the mobo where the CPU socket used to be.

Prescott P4 CPUs will dissipate up to (I think) 95W (maybe more for the EEs 
with 2MB cache) but very, very much less when idle and/or throttled.

40W sounds the sort of saving which I would expect from an older P4 system, 
without having to resort to explanations involving second- or third- order 
effects involving components other than the CPU. In fact, bearing in mind the 
multiplier effect of the inefficiency of the CPU, the measured utility power 
consumption could be changed by very much more than this - I suspect up to 
around 150W on a system based on current CPUs.

Anyway this has no effect on the argument that, if any extra power 
consumption due to Prime95 is offset by reducing room heating, Prime95 is 
energy neutral - the extra KWh will be transferred from the heating budget to 
the utility power budget, though probably at a different unit cost.

Regards
Brian Beesley
_______________________________________________
Prime mailing list
[email protected]
http://hogranch.com/mailman/listinfo/prime

Reply via email to