• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Silly power requirements, an over hyped myth?

Soldato
Joined
12 Mar 2006
Posts
23,011
Location
N.E England
Inspired by this thread http://forums.overclockers.co.uk/showthread.php?t=17736615, I am going to do a few experiments to settle this GPU power myth on these boards, this should also give people a good idea as to what PSU they should consider if buying an HD2900XT. Hopefully this should be helpful to some people.

I purchased one of these :-

l61aqir5.jpg


This can Measure voltage (V), amps (A), watts (W), volt-amps(VA), hertz (Hz) and power factor (PF).

Ok so the first test I did was to get into Windows, In windows my pc idles using around 250-260w. This is my power usage at idle when you take into account I have an R600 which is supposed to be some sort of power monster even in 2d I found this pretty impressive. So i fired up Orthos and stressed for 10 minutes, throughout the whole ten minutes the power stayed at 300-308w and didn't budge. I will be doing some 3D tests when i get back on my other PC, does anyone want to take a guess as to the power my PC will use in 3d mode based on the 2d power results?, What do you think?, post up your guesses.
 
Ok the 3D Mark results are in, the power usage remained between 370-400w the highest it peaked was 400w and it only stayed there for a split second this was on Canyon Flight and in the cpu tests the power usage remained around 350w. Im fairly suprised and I will do more tests with various different games to see how it affects the power usage. All in all 400w is impressive, just shows you can most likely get away with a 2900xt on a 450w with decent railage and still have a nicely oced C2D.


cimg0317pe0.jpg
 
Cyber-Mav said:
same here, over 400w if the cpu and gpu are both stressed at 100%

also
can you do some cpu testing.

use orthos and see how much power it consumes, and use snm to see how much power is then used. for all the systems i have ever used snm generated the highest heat load so i guess it will be the one that pulls more power.


Stressing both cores to 100% on the CPU test in Orthos @ priority 9, the power usage didnt go above 308w. It hovered between 300-308w.
 
Cyber-Mav said:
interesting, so does that mean any psu with 400w on the 12v rails is good enough?

would be good to know how much power is sucked up from the 5v rails these days.

but yea as you say, something like a corsair 520 is more than enough for anyones needs.

Precisely, it's a shame Corsair dont do a cheaper 450w model or do they?, because even thats more than plenty :D. Very impressed tbh its put my mind at rest and opened my eyes anyway.
 
Dureth said:
To factor in efficieny yiou need at least a 500w supply then for that power consumption. Maybe a bit more.

Yea but if you get an effieicent enough 450w or 500w then your spot on. If you got a totally efficient 700w it would last you forever :D , this does not mean any old 450w psu could run my setup, it would have to be fairly efficient.
 
Sir Random said:
Can you try running 3DMark while the Orthos CPU test is running? It still won't use the sound card, but will give us a better idea of the max game consumption.

OK, I will run Orthos at priority 1, this won't change power usage I dont think but will at least allow 3D Mark to run :D . Will post results soon.
 
Cyber-Mav said:
can you run the ati tool cube spinning thingy, that does the most power woundage for the gfx card.

Cant even run ATi tool atm just bsods as its not compatible, OK have redone the test this time I let Orthos stress in the background, both cores at 100%, this did not increase the power usage massively, the power usage this time peaked at 428w and once again this was in canyon flight and only stayed there for a few seconds, canyon flight seems to be the most intensive, on all other tests the power usage stayed aroud 410-420w. Once again pretty impressive. Will be trying other games to get a better idea. Either way I now know I have a nice Oc'ing headroom for the XT for when I decide to do so (Whenever ATi tool is compatible)
 
Boogle said:
While these tests are great, and show what power is really used - this doesn't mean we can all go out and get cheap lower powered PSUs. Q-Tecs were known to work fine for a while, then spontaneously die, sometimes taking hardware with it.

Also the closer you get to a PSU's maximum, the worse the ripple and voltage regulation becomes (not to mention efficiency). Best bet is to check out places like http://www.jonnyguru.com and http://www.hardocp.com who both test PSUs properly at various loads. Just check your projected load, and see how well the PSU does.

Ripple is arguably one of the most important values, because high ripple on the 3.3v rail kills memory, and high ripple on the 12v rail kills HDs.

Which is what we have all said above :confused: , obviously low efficiency 450w is pointless thats not what I set out to prove, I've showed that with a good efficient PSU theres no need for silly wattage PSU's. A Corsair 520w would be ideal in this situation and is some what cheaper than the Type R (I think?)
MikeHunt79 said:
I think this pretty much proves that 1kw PSU's are pretty pointless... They'll probably be bought by people who think that more watts = better PSU :)

Spot on. a good efficient 750w-800w is more than enough ATM (SLI/Crossfire) and would last eons.
 
Boogle said:
Not disagreeing with you at all :) But usually when someone shows how much power a PC uses, some people wander out and get cheap low powered PSUs and then come and cry about it later.

Aye but im sure the majority of posters on here would have more sense than that :D .
 
chaparral said:
If i remember right My system reads 229watts idle and about 360watts load (3dmark06)

C2D 6600 running 3500mhz
8800gtx running 630/950
18watt waterpump
6x 120mm fans
1x 92mm fan
x3 harddrives
1x dvd writer

I check again this evening after work...

Idle wattage....i get a load wattage picture up later
DSC00127.jpg


My idle wattage is identical to yours, I think mine is a tad lower though :eek:
 
cells said:
WOOOOOOW


that is a LOT for idle

£263 of electricity per year assuming 12p per KWh and that your computer is on all the time at idle. if it is on half that time then £131.5 per year!!


you might be better off building a rig to use for general idle stuff. could probably build a duel core rig which idles at less than 70 watts for under £150 (humm reading around, seems even energy effient builds use 125+ watts)

I rechecked and had read it wrong my PC idles around just under 230w. Still fairly high though, may enable speedstep and see how that changes things. Only problem with speedstep is it lowers the clock and not the Vcore (that stays at whatever I had set in the bios) :(
 
chaparral said:
What are all your PC specs ??.E.G how many hardrives , Fans , DVD writer , CPU & R600 overclocks etc

2 HDD'S 7200RPM
2 DVD/RW/CDRW Writers
4 Case fans 2x 90mm 2x 120mm
CPU @ 3.4 1.38v (Actual volts after the droop)
R600 @ Stock.
 
fornowagain said:
Do you have separate options for C1E and EIST in the bios? I've never looked, but maybe the Vcore needs to be set to Auto? That would be a pain in the proverbial.

The Vcore has to be set to Auto afaik no way around it, and EIST is automatically enabled I think with my borad and I had enabled C1E yes, then I changed the power profile in Windows and it worked fine (lowered clocks on idle) but the Vcore stayed the same deeming it pretty useless. :(
 
Lanz said:
I have one of these plug things and my 6300 C2D HTPC idles around 90W. This I dont mind having on 24/7, it uses less energy than my old P4 2.4ghz Northwood HTPC.

But it's just throwing money away having your main gaming PC on 24/7 - And we're meant to be using less carbon aint we??

Meh. I dont leave my PC on 24/7 :confused:
 
Back
Top Bottom