• A Great friend to the HardForum with a great kid that he is trying to get a scholorship to continue his schooling. Please give hime a vote! Only 24 hours left! Thanks.
    If you have an VOTE FOR KEENAN!

Excellent Video Card Power Usage Comparison

Zarathustra[H]

Zarathustra[H]

Extremely [H]
2FA
Joined
Oct 29, 2000
Messages
42,423
I cam across this chart today, and my searches didn't show it posted here already:

http://forums.atomicmpc.com.au/index.php?showtopic=264

For amperage values divide by 12.

It looks to me as if we have been over-stating the need for high powered PSU's on here. (Maybe this is because we want to make sure we have plenty of headroom for overclocking)...

Judge for yourselves...
 
That's my point, referring to your statement:

It looks to me as if we have been over-stating the need for high powered PSU's on here.
Click to expand...

The charts themselves are informative and helpful for what they are, thanks for the link. ;)
 
requiem99 said:
The charts themselves are informative and helpful for what they are, thanks for the link. ;)
Click to expand...

NP :p

Now if I could only find something similar for CPU's we could start getting a good way to determine how much power is REALLY needed. For now I guess we could go by manufacturers published TDP...
 
I've know this for a while now. Great find!

Real world gameplay vs Benchmarks also apply to power usage.
 
holy mother of god. 1120w for 4 way sli gtx 480's. compared to 571w for 4 way HD5870's thats just nuts.


thanks for the link OP'er.
 
There are several good reasons we recommend "overkill" PSUs, OP.

Many PSUs are most efficient somewhere around 60% load. They run cooler, they run quieter, and they provide more DC power for a given amount of AC power, which translates to lower running cost, too. Running cooler also means they last longer.

PSUs also de-rate over time. I can't remember the accepted value, but let's say it's somewhere around 5% a year (and it may be more, I can't recall) per year under normal usage. So, in two years a 1000w PSU is really a 900w PSU for all intents and purposes. Obviously this is not a hard-and-fast rule but it does occur and needs to be planned for. Also, the cooler the PSU is running the more slowly its internal components degrade and it de-rates.

OCing can also cause a huge increase in power draw, as you mentioned. OCing a i7 920 from stock 2.66Ghz to 4Ghz results in a more than 2x increase in power draw, or about another 100 watts. GPUs can be even worse about increased power draw. Lots of headroom makes sense.

Still, interesting link. Thanks.
 
Last edited:
silent-circuit said:
There are several good reasons we recommend "overkill" PSUs, OP.

Many PSUs are most efficient somewhere around 60% load. They run cooler, they run quieter, and they provide more DC power for a given amount of AC power, which translates to lower running cost, too. Running cooler also means they last longer.

PSUs also de-rate over time. I can't remember the accepted value, but let's say it's somewhere around 5% a year (and it may be more, I can't recall) per year under normal usage. So, in two years a 1000w PSU is really a 900w PSU for all intents and purposes. Obviously this is not a hard-and-fast rule but it does occur and needs to be planned for. Also, the cooler the PSU is running the more slowly its internal components degrade and it de-rates.

OCing can also cause a huge increase in power draw, as you mentioned. OCing a i7 920 from stock 2.66Ghz to 4Ghz results in a more than 2x increase in power draw, or about another 100 watts. GPUs can be even worse about increased power draw. Lots of headroom makes sense.

Still, interesting link. Thanks.
Click to expand...


Good points all, and worth keeping in mind.

After seeing these charts, it suddenlt makes sense - however - how someone with my 500W PSU (38A on 12v rail) was able to run a HD5970 reliably.

Seeing that real life measured power usage for the GTX580 is lower than for the HD5970, I may actually try getting one now.

(PSU is non-standard form factor, and no more powerful replacements exist)

If It is insufficient, I'll likely get an old Dell Optiplex power brick on ebay, and mod it to provide power for the video card.
 
You're really serious, aren't you? There have been websites that do true measures of video card ONLY power for years now (Xbit Labs and Tech Power Up). Some examples of their articles:

http://www.xbitlabs.com/articles/video/display/radeon-hd6870-hd6850_9.html#sect0

http://www.techpowerup.com/reviews/ASUS/GeForce_GTX_580/25.html

Now, I know it's nice to have it all there in one place, but the information is not new. In fact, after checking the links, I would not trust that information to be accurate because many of their numbers are based off full system power consumption, where they've guesstimated the system power and the PSU losses.
 
Last edited:
Zarathustra[H];1036438033 said:
NP :p

Now if I could only find something similar for CPU's we could start getting a good way to determine how much power is REALLY needed. For now I guess we could go by manufacturers published TDP...
Click to expand...

Or, you could just go and check out Xbit labs, which actually takes the time to test processors:

http://www.xbitlabs.com/articles/cpu/display/cpu-benchmark-value_10.html#sect0

And then there's Tech Report. When they get a hold of new processors, they even calculate total power consumed during the course of a test:

http://techreport.com/articles.x/16044/7

Sure, it's not isolated CPU-only power, but the reality is that you don't have much choice for chipsets these days, so you might as well throw that into the platform measurement. Every other component is the same, so it's a good comparison.
 
silent-circuit said:
Many PSUs are most efficient somewhere around 60% load. They run cooler, they run quieter, and they provide more DC power for a given amount of AC power, which translates to lower running cost, too. Running cooler also means they last longer.
Click to expand...

Is this really a big deal with modern PSUs? I know that older PSUs can dip down into ~70% efficiency range or lower when pushed near their limits but part of the 80PLUS certification is that the powersupply maintain at least 80% efficiency at 20%, 50%, and 100%.

PSUs also de-rate over time. I can't remember the accepted value, but let's say it's somewhere around 5% a year (and it may be more, I can't recall) per year under normal usage. So, in two years a 1000w PSU is really a 900w PSU for all intents and purposes. Obviously this is not a hard-and-fast rule but it does occur and needs to be planned for. Also, the cooler the PSU is running the more slowly its internal components degrade and it de-rates.
Click to expand...

I'd be curious to see the extent to which this is actually true. To me it sounds like something that was created out of the computer tech mindset where things become obsolete unusually fast. Even if the powersupply did "de-rate", it would probably only do so to a small extent after which it would likely plateau out and remain fairly constant in power output for quite a large number of years.

I have a large collection of Vintage audio gear, much of which dates back to the mid 70's and almost all of which are still using all original parts. I recently took one of my best units, a 240wpc Yamaha stereo amp from 1979 into the shop. It was still able to deliver it's rated power output on a bench test despite 31 years of age. In my server I'm still using the same 550w EPS12v powersupply I originally got back in 2003 for a dual-xeon rig. It had 24+8, plenty of power, still worked great, so I used it. I think the concept of a powersupply getting significantly weaker over such a short amount of time is comically silly.
 
GotNoRice said:
Is this really a big deal with modern PSUs? I know that older PSUs can dip down into ~70% efficiency range or lower when pushed near their limits but part of the 80PLUS certification is that the powersupply maintain at least 80% efficiency at 20%, 50%, and 100%.



I'd be curious to see the extent to which this is actually true. To me it sounds like something that was created out of the computer tech mindset where things become obsolete unusually fast. Even if the powersupply did "de-rate", it would probably only do so to a small extent after which it would likely plateau out and remain fairly constant in power output for quite a large number of years.

I have a large collection of Vintage audio gear, much of which dates back to the mid 70's and almost all of which are still using all original parts. I recently took one of my best units, a 240wpc Yamaha stereo amp from 1979 into the shop. It was still able to deliver it's rated power output on a bench test despite 31 years of age. In my server I'm still using the same 550w EPS12v powersupply I originally got back in 2003 for a dual-xeon rig. It had 24+8, plenty of power, still worked great, so I used it. I think the concept of a powersupply getting significantly weaker over such a short amount of time is comically silly.
Click to expand...


newer stuff isn't made to the same standards of quality that that older stuff was.....PSU should maintain their power delivery until a component fails within a few % of what it was new. Now a days there is a lot of crap that is seriously over rated and that is why it fails when you try to run it at rated load for long times (unlike that old Yamaha that was engineered with a connsiderable safety margin- I'd be willing to be it can repeatidly go significantly past it's rated power)
 
You must log in or register to reply here.
Back
Top