[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Where to go to learn about reducing electricity use?

Douglas Tutty wrote:
On Tue, Jan 02, 2007 at 09:19:34AM -0600, Hugo Vanwoerkom wrote:
Douglas Tutty wrote:
On Mon, Jan 01, 2007 at 07:30:40PM -0600, Reid Priedhorsky wrote:
On Tue, 02 Jan 2007 01:40:11 +0100, Douglas Tutty wrote:
On Mon, Jan 01, 2007 at 05:06:02PM -0600, Reid Priedhorsky wrote:
I would like to make my Etch box use less power, but I'm having a hard
time consolidating all the information I'm finding. Is there a package I
can install or a web page I can read that is a gentle guide to reducing my
power usage?

A good place to start may be something like the laptop-howto.

What power are you trying to reduce?  Beware spinning down drives to
reduce power; spinning up reduces drive life.
Thanks for your reply. It is a self-built Athlon desktop, though I assume
much of the advice in the laptop-howto will be relevant. I will take a
look. I'm happy to provide more specs on the box if that will be helpful.

I'm not really sure how to answer your question. I know that spinning down
drives is not such a good idea, and that I can reduce processor speed when
idle (though how I don't know), but beyond that I don't have a good idea
of what options are available to me.

Hi Reid,

I'm running Etch amd64 on my box (Athlon 3800+, Asus M2N-SLI Deluxe MB,
1 GB stick of DDR2 800, dual Seagate Barracuda 80 GB drives, DVD burner,
Asus nVidia 7300 GT silent GPU, in a CoolerMaster Stacker case with
cross-flow fan, blow-hole fan, three drive modules with 120 mm fans, CM
iGreen 600W PSU).

I have the MB handle the fan speeds (adjusts for temperature, not for
quiet) and most of the time can't hear anything but a slight directional
high-pitch from the CPU fan.

I just put an ampmeter on it during boot and it uses 1.2 A at 120 V so
that's 100W.

I note that theres a package to throttle the CPU based on load.  I don't
know if this reduces the power consumption for what you're looking for.
It may primarily just reduce the CPU temperature (although mine hovers
at around 40 C).

Why are you worried about power consumption on a desktop?


I measure exactly my system uptime/month and the KWH used where the computer and a desklamp are the only devices using electricity.

The latest data shows:

Mon May 01-05:38:44 04595 == 89  KWH 1 month 303:50 == 205 mins/KWH + Fan
Thu Jun 01-05:52:17 04687 == 92  KWH 1 month 284:30 == 186 mins/KWH + Fan
Sat Jul 01-09:05:37 04769 == 82  KWH 1 month 232:35 == 170 mins/KWH + Fan
Tue Aug 01-09:43:58 04851 == 82  KWH 1 month 271:15 == 198 mins/KWH + Fan
Fri Sep 01-09:27:08 04932 == 81  KWH 1 month ---not running
Sun Oct 01-09:31:20 05001 == 69  KWH 1 month 238:55 == 208 mins/KWH
Wed Nov 01-07:06:51 05069 == 68  KWH 1 month 236:20 == 208 mins/KWH
Fri Dec 01-07:50:31 05127 == 58  KWH 1 month 228:55 == 236 mins/KWH - no Fan
Mon Jan 01-06:37:15 05211 == 84 KWH 1 month 222:35 == 159 mins/KWH ??? no Fan ????

So I get variations between 159 - 236 mins/KWH, which is a huge variation.

E.g. this month the system uptime was almost the same as last month ( - 6 hours) yet the difference in KWH was a whopping 26 KWH!

And the explanation? Can that be accounted for by what the system was running? E.g. a new USB disk?

(geographical location = Mexico)

I think I have the math right: 60000 Watt-Minutes = 1 KWH; 60000 / 159 =
377 Watts.  60000 / 236 = 254 = 254 Watt.

How big is your desk lamp?

The other issue is power-factor.  Hydro meters measure KWH (watt-time,
in effect energy in jouls) whereas power supplies use Volt-Amp.
The last time I looked at this seriously was when I was living in the
bush (in Canada) with unreliable power and had the whole house on an
inverter/battery setup.  The pure sine-wave inverter did the
power-factor correction and my P-75 used 20 W with one hard drive.

Good luck,


Interesting. The lamp is a dual fluorescent desklamp that I got when I started with IBM in 1966. Has 2 OSRAM L 15W/10 in it. But I always have that on.

I just don't get the variation.


Reply to: