Reducing power usage, ideas?

[transylvanic]

Underclocks aren't really the solution as you've noted. I'd reiterate the advice in favour of plumping for a 4830, or waiting for the (expected to be) even more efficient RV740. I really don't know how much power those fans use, but 11 is overkill however you think of it

You could go for just the one hard drive. Or as suggested above, laptop drives. Otherwise, your CPU isn't that much of a power slurper, the focus should be squarely on finding a more energy-efficient graphics card.

[rs4847]

Hmm, how many fans you have in your system?. I have 11(including NB/cpu cooler), it could be my motherboard as its an nivida chipset and iirc they are known to be power hungry(and useless overclocking) chips. I really dont see how it could be more than your system though, you have a gtx 260 and a quad core which is alot more power hungry than my chip thats underclocked to 2.5ghz.

Im going to try and take out every fan and see the difference on my system later in the week. It could well be my PSU not being as efficient at lower usage as if mine takes about 140w then an extra 20% is needed(~80/85% effcient) so that would be about 170.

I have 4 120mm fans and a Zalman CNPS9500-AT heatsink/fan.

My PSU is 80 PLUS certified.

[Pob255]

http://archive.atomicmpc.com.au/foru...c=7&t=9354&p=0

9800gtx+ 45w at idle
4850 41w at idle

9800gt 35w at idle
4830 no figures given yet.

Couple of things to do, stop overclocking in seriousness have you got any automatic overclocking in the BIOS or utility so that overclocking only kicks in when under load.
Also check "intel EIST" is turned on in your BIOS

EDIT: look at you fans, they should have an Amp rateing on them, multiple by 12 to get Wattage when running at full speed. common 120mm fan is 0.2-0.3 amp so 2.4-3.6 watts each
Rough order of power comsumption (high to low)
CPU
Graphics card
hard drive / memory (not sure which is most here)
CD/DVDrw
CD/DVDrom
Fan

Hard drives vary but around 20w each (on spinup), so 5-6 hard drives to the graphics card.

[Hicks12]

Hmm, no overclocking and all settings are on manual along with c1 and intel speedstep being on. If i have 12 fans and an average of 3watts a fan(ill check later|) then thats putting alot of un needed wattage.

I cant really see where this power usage is coming from if all i have using power is....

CPU: e8400- Max 65W(underclocked aswell)
Mobo: Asus p5n-d, high ?
GPU: 9800GTX 55w as its not the + version
hard drives: 250AAKS and 640AAKS which are low power 10watts i believe so 20w total
DVD RW drive: only 1 drive so cant use to much
Memory: 4gb underclocked 2 sticks of crucial ballistic(1.95v)
Fans: 11, will cut down tomorrow
Fan controller: cant use much surely
Sound card- Not much again really

Anyone else see a reason here?XD

[Webby]

Easiest way to check what your fans are actually drawing power wise is to monitor turning them on and off, even if they ave a sticker on with power draw numbers that will be an engineering estimate so way over what the fan is likely to draw.

My 2000rpm (not that they run that fast normally 750rpm idle and 1200rpm at full load) Scythe Ultra Kaze 38mm thick fans draw 3-4W each (fluctuations make it hard to get a definite number) from the wall so any standard 120mm 25mm thick fan is realistically going to be less than that. My pump draws 20W which adds a bit to my power consumption though

If saving energy is important then the best thing you can do is turn your PC off (including your monitor) when it is not in use or stick it in suspend or hibernate it. As others have said getting a small laptop or a netbook for surfing will save you a bit as well keeping the main PC for things that actually need the grunt. Most energy saving exercises though will cost you more than you will save in the cost of electricity, if you can't do it for free its probably not worth doing.

[Hicks12]

Yeh alot of these ideas require me spending money that A, i dont have and B, it wont regain its cost within a few years atleast. I tested the fans today and it was switching from 180-190 on idle(from power meter) and when i took the power off the fan controller which powers about 6 fans the wattage was 175/177 so it went down about 3 or 5 watts, im really confused now as this seems way to low in difference from 6 fans. Could the power meter just be crap?.

How can you work out the power usage from Amps again? i think you need the resistance values but if theres another way someone tell me >
Its using 1amp at the time on idle and the meter says max was 1.3(guessing load).

[smargh]

... when i took the power off the fan controller which powers about 6 fans the wattage was 175/177 so it went down about 3 or 5 watts, im really confused now as this seems way to low in difference from 6 fans. Could the power meter just be crap?.

I read a post on usenet 10 days ago which is relevant to this:

CPC sell a similar unit (possibly the same chinese made guts as those
in that B&Q version) where, unusually, the specification, more openly,
admits to the meter's limitations by stating an accuracy tolerance
figure of +/- 5% (quite good by itself) or +/- 1 digit of resolution
(ok, it's a digital meter with a 1 watt resolution) or (and this is the
telling figure) +/- 10 watts.

Assuming your meter is as accurate as that CPC one (and the figures of
accuracy tie in with the sort of 'weird readings' I've seen similar
meters give on loads in this power range (as measured with the
Metrawatt), you could expect an 11 watt load to show anywhwere between
zero and 22 watts.

A good way of testing your meter would be to get a specific wattage of old-style bulbs and see if the meter is anywhere near that value My Aldi power meter is spot on with a 5-watt bulb.

[Hicks12]

So since it switches from 170-190(not slow changes either) i guess it could actually be 160? that would seem to put my system down to an acceptable amount since the psu would take 20% of that which makes the pc use in theory 128w. Any reliable way to test power usage? multimeter(if so how) .

[Webby]

Hicks remember that your PSU is not going to be at its most efficient at low or high loads most peak at about 50% load. So if you have a 700W PSU and are drawing 180W that is only 26% load which can potentially put you well down the efficiency scale depending on the PSU. Most of the high efficiency PSUs don't really drop off that much the Corsair HX520 (according to Corsair) achieves ~82% efficiency at 20% load and peaks at ~86% efficiency at 50% load.

As to knowing exactly how much energy you are using well you could always turn off everything else in the house and time the electricity meter to change 1 unit or maybe just a tenth of a unit, 1 unit is 1kWh = 3600000W so that would take a while to time so do it for a tenth of a unit 360000W at 200W should take you 1800 seconds so you will probably need to be timing for 30 minutes to an hour if you can justify having everything turned off (and I mean everything) for an hour then that is probably the most accurate way of confirming how much power is being used.

The more practical method as suggested is to plug in a 60W lamp and see what the power draw is according to your meter if its all over the place you know thats the problem if it is stable at 60W (or there abouts) then your probably reading a reasonable figure.

[Hicks12]

Yeh thats the problem i think, is there a minimum that psu's have to meet at all levels?.


edit: I found it! Mine should be at 80%, according to xclio website it will be 80% at 20% 50% and 100% load, http://www.xclio.com/products/tech_t...lus.htm#80plus So its doubful that its the psu correct?

[Hicks12]

The power meter shouldnt be wrong, i tried an 11w energy light bulb and it was exactly 11w on the reader... confused now on how it can use so much energy when compared to others especially when mines a lower spec!.

woops this was ment to be an edit >

[rs4847]

My meter is fairly accurate as well.

Looking around it seems your motherboard is to blame.

http://www.xbitlabs.com/articles/mai...-p5n-d_12.html onwards.

[Hicks12]

bloody hell, 50w difference from beter boards!. Any idea how much my mobo worth? may sell it and get a more energy efficient mobo(suggestions?), if i sell it for £40-£50, how much would the 50w saving take to cover itself?

[shbris]

50W difference for 50quid would take roughly 6600hours

based on 15p per kWhr

so not worth it

[Webby]

1 unit = 18p ish
1 unit = 1kWh = 3600000W

So 50W works out 0.00025p per second

3600 seconds per hour so 0.9p per hour if you say 8 hours a day 365 days a year that is 2628p per year or £26.28

Do the calculations for your particular usage patterns.

[arthurleung]

Lowering Memory Frequency of a HD4870 -> Saves 22W Idle
Lowering Laptop CPU Voltage (RMClock) -> Saves 14W (T7200) / 15W (T8400)

Reducing your overclock (if any) could save up to 150W depends on how high it was.
I overclock my CPU up to a point where Performace/Watt is highest. My CPU rarely idles though.

With my WD Greenpower drives I see almost no difference in idling and spin-down (not to mention it corrupts my RAID5 occasionally if spun down). 7200rpm drives might see a small difference though.

More stuff you can do is go into BIOS and lower whatever voltage you can find. When not overclocking, most CPUs should handle undervolting a couple notches to somewhere around 1V~1.1V for CPU. (Saves 10~50W)

Slow down memory to DDR2-800 CL5 (or 6 if the memory is crap), and set voltage to 1.8 can also save a couple of watts.

If you're truly into minimal power, get a cheap laptop with integrated graphics and a decent CPU (With better power management / undervolt further). When I was in lecture today and my laptop registered 10:11 battery life remaining at one point when I idle (9Cell 85Wh battery). If you don't have money to spare, Atom-based laptop use about the same power as minimalist laptops with high-end CPU.

And as usual, reducing the computer's power usage doesn't always save you money. I have my computer on, and as a result I don't need to use the boiler to warm my room. But my gas bill is still far more expensive than the electricity bill even though there is only one bedroom (my flatmate) + toliet is heated.