[mythtv-users] silent and low-power: AMD or Intel?

[John Drescher]

On Sat, Jun 6, 2009 at 4:27 AM, Alex Butcher<mythlist at assursys.co.uk> wrote:
> On Fri, 5 Jun 2009, John Drescher wrote:
>
>>> AFAIK and it was discussed on this list, frequency scaling on Intel is
>>> only useful to limit a processor from over-heating and will not change
>>> the power consumption significantly, if at all
>>
>> I have just tested and verified this on with a brand new core2 quad
>> 9650 3.0GHz here at work.
>
> My observations conflict with this statement.
>
> I have a Core2 Quad Q6600 also on an Asus P5Q board. It has 4GB of memory, 2
> 1TB SATA HDDs, 2 200GB PATA HDDS, a Promise PATA card, a Hauppauge Nova-T
> DVB card, a Lite-On DVD-Rom drive and a Sony Optiarc writer. At idle on a
> GNOME desktop, my no-name power meter shows the system drawing about
> 107-116W. The system runs FC10, and uses the ondemand governor module
> (acpi-cpufreq) in the 2.6.27.24-170.2.68.fc10.x86_64 kernel.
>
>

A core2 quad Q6600 lacks some of the energy efficiencies that the Q9650

>
> I started up four terminal windows and in each ran:
>
>        while [ true ]; do true; done
>
> Each instance causes one core to jump from 1.6GHz to 2.4GHz. I then killed
> them off one by one and recorded the power consumption.
>
> Instances       Power
> 4               174W
> 3               154W
> 2               138W
> 1               127W
> 0               116W
>
> The TDP of the Q6600 is 105W. (105W/2400MHz)*1600MHz=70W,

You can not do that calculation because the power is not linear with
frequency.

John