• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Realistic power supply need

D

Digital Viper-X-

[H]F Junkie
2FA
Joined
Dec 9, 2000
Messages
16,218
So I borrowed a kill-a-watt to play with, and test to see just how much power my system is eating up by leaving it on all day.

I have 2 UPSs(don't ask :p)
the UPSs + 1 router + access point + 1 voip router, take about 80W with nothing else on.

The monitor, 30" pulls about, 120W with brightness set at a decent #
The system, i7 920 @ 3.2 (almost stock v)
12GB ram
5970
XFI Prelude
3SSD
1 VR
2 7200rpm 3.5" hdS
7 fans total

the # at the wall, from UPS #1 is 390W currently(writing this post), it can go up to 420w or so depending on the brightness of the background. the cpu is sitting @ 2400mhz(speed step?)
under furmark load, the wall # is about 770W - 120W or so for the monitor, 80W for the various other crap that's about 580W with the gpu @ 800/1150 stock volts(both of them :p)
@ 850 1.1v vs 1.05v stock, the power jumps up nearly 50W to 825W at the wall, again - 200W for monitor+other = 625W * .85 for the power supply efficiency.

625(gpu: 850mhz 1.1v)=531W
580(gpu: 800mhz 1.05v)=493W

I guess an 850W psu is slight overkill ^^, a 700W or even 650W might have been enough, since furmark sort of stresses the card more then most games =). Goes to show how much power we really need. and holy crap I never knew the monitor sucks up so much power
 
Oh so glad I have a copy paste for this:

Don't take those KAW figures as accurate:
Paul_Johnson said:
Yes and a quick search would turn up this topic a million times over. Here is the recap:

1) APFC can fool Kill-A-Watts into giving you abnormally low readings (some times giving better than 100% efficiency)

2) Power supplies derate with temperature anywhere from 2w/c above a nominal rated at value to 10w/c.

3) Kill-A-Watt's and most power meters sample too slowly to catch transient loads (the Transient load from our tests is 117w and is COMPLETELY missed by Kill-A-Watts).

4) Power supplies last longer if you stay in the 40% to 60% range of their output.

5) power supplies are quieter if you stay in the 40% to 60% range of their output.

6) Power supplies are cooler if you stay in the 40% to 60% range of their output.

The power meters in UPS software are just as bad. You have to spend some change before you get anywhere near an accurate power meter when your PSU has APFC.
Click to expand...

Here's Paul Johnson's post, PSU reviewer of HardOCP.com, about the inaccuracy of the Kill-A-Watt:
http://hardforum.com/showpost.php?p=1032190998&postcount=7

In addition, three other PSU experts backs up Paul Johnson's statement:
Oklahoma Wolf of JonnyGuru.com:
http://hardforum.com/showpost.php?p=1034843536&postcount=21

JonnyGuru of JonnyGuru.com himself as well as the senior PSU engineer over at BFG acknowledge the inaccuracies of a KAW (Post #7 in regards to Post #2):
http://www.jonnyguru.com/forums/showthread.php?t=5977

Redbeard of Corsair also acknowledges the inaccuracies:
http://hardforum.com/showpost.php?p=1032811067&postcount=22
 
Danny Bui said:
Oh so glad I have a copy paste for this:

Don't take those KAW figures as accurate:


Here's Paul Johnson's post, PSU reviewer of HardOCP.com, about the inaccuracy of the Kill-A-Watt:
http://hardforum.com/showpost.php?p=1032190998&postcount=7

In addition, three other PSU experts backs up Paul Johnson's statement:
Oklahoma Wolf of JonnyGuru.com:
http://hardforum.com/showpost.php?p=1034843536&postcount=21

JonnyGuru of JonnyGuru.com himself as well as the senior PSU engineer over at BFG acknowledge the inaccuracies of a KAW (Post #7 in regards to Post #2):
http://www.jonnyguru.com/forums/showthread.php?t=5977

Redbeard of Corsair also acknowledges the inaccuracies:
http://hardforum.com/showpost.php?p=1032811067&postcount=22
Click to expand...

mmm, does it make a difference that I was measuring the power that the UPS was drawing rather then straight from the power supply :p?
 
If the killawatt is so innaccurate, what can we do to get a real power reading?
 
Spare-Flair said:
If the killawatt is so innaccurate, what can we do to get a real power reading?
Click to expand...

Buy a better, and much more expensive power meter, ($150-$500), like the Brand unit referenced in a couple of those links.

One thing I didn't see after following a bunch of the links was a more detailed explanation of why the cheap meters are so easily fooled by APFC.
 
Metaluna said:
Buy a better, and much more expensive power meter, ($150-$500), like the Brand unit referenced in a couple of those links.

One thing I didn't see after following a bunch of the links was a more detailed explanation of why the cheap meters are so easily fooled by APFC.
Click to expand...

either way though, wouldn't a UPS negate the fact that your power supply has APFC,(if you measure the power drawn by the UPS?)
 
Metaluna said:
One thing I didn't see after following a bunch of the links was a more detailed explanation of why the cheap meters are so easily fooled by APFC.
Click to expand...
That's because nobody has figured it out yet.
 
Digital Viper-X- said:
either way though, wouldn't a UPS negate the fact that your power supply has APFC,(if you measure the power drawn by the UPS?)
Click to expand...

That's part of the problem: we don't have enough information. Without understanding what makes the power meter innaccurate in the first place, it's hard to judge if the UPS will filter out whatever badness the APFC circuit is causing in the AC waveform.

Even better would be if someone actually tested the effect of the UPS, but I didn't see any test like that.

Another option would be to insert probes directly into the DC power rails coming out of the power supply inside the case. Unfortunately this would also require some fairly expensive equipment unless someone already mass-produces a widget like that.
 
It would be nice to know for certain. I'd much rather spend the $250 on a good UPS and a $30 kill-a-watt than $300 on a supposed high end power meter and have no UPS.
 
To estimate power supply needs, you would also need to load the CPU along with the GPU. Try running 8 threads of Prime95 along with Furmark and then see what the power numbers are.
 
Metaluna said:
Another option would be to insert probes directly into the DC power rails coming out of the power supply inside the case. Unfortunately this would also require some fairly expensive equipment unless someone already mass-produces a widget like that.
Click to expand...
Xbit Labs has done just that, although it was some time ago.
Their test results hold true: Just about nobody will need a PSU rated for more than 800W.

While the comment on how a PSU works best at ~50% load is true, one must also realise that the typical load for most computers (as seen over time) is very near idle.
I don't recommend picking a PSU rated so that the idle load equals 50% of the rated load, but at least the rated load shouldn't be twice the actual peak load.
Another reason for not going over the top with PSU rated power is that most PSUs have crappy efficiency once you get into the 5-15% range of their rated maximum.

Cheers
Olle
 
siliconnerd said:
To estimate power supply needs, you would also need to load the CPU along with the GPU. Try running 8 threads of Prime95 along with Furmark and then see what the power numbers are.
Click to expand...

I don't think that's true, there really is no game that will pin both your cpu and gpu currently
 
never played gta4 :p so i guess I was wrong! but anyway, does GTA4 put as much stress as furmark on the GPU?
 
You must log in or register to reply here.
Back
Top