Minimum power draw for 1000W PS

R

rtangwai

[H]ard|Gawd
Joined
Jul 26, 2007
Messages
1,369
I have a OCZ 1000W power supply in my media server, which currently has a [email protected],8GB DDR2, HD 4870 1GB, and 4 hard drives. It had 6 hard drives at one point, but I've been deleting old crap that I don't need anymore (my granddaughter hasn't watched the Barbie movies ever since she discovered boys) and removing my high spindle-time drives as I know they will start failing soon. I will soon be swapping the Q9400 for a dual-core and removing the HD 4870 in favour of a HD 5450, which uses about a tenth of the wattage as I want to save money on electricity especially when summer comes and the server becomes a space heater the central AC has to contend with.

What I want to know is this: is there a minimal draw for a 1000W power supply to keep it healthy? I am uncertain what my media server pulls from the wall right now, but based on eXtreme Power Supply Calculator it claims I should have a 562W power supply. That's peak, which I almost never do as the media server doesn't play games (it records TV and compresses video). Once I make the changes the calculator says I need only a 385W PS. Is it a problem if the power supply is *TOO* underutilized? I am concerned that the server itself will average only 200W or less, which means I'm using only 20% of the capacity of the current PS. How efficient is a 1000W PS at such low utilization? Will I save more electricity if I toss it out and get a 400W PS?
 
It really depends on the specific power supply.

Look for a review for your specific power supply here or on JonnyGuru.com

Both will have the information you are looking for.
 
You might save $5 a year by switching it out, which is not worth the hassle IMO. Using only 20-30% of its capacity will not harm it in any way, thats what most people with 1000w+ use with small bursts higher when they bench or game.

Heck, my SB-E rig is idling at 95w right now, and I have a 1200w supply.
 
LigTasm said:
You might save $5 a year by switching it out, which is not worth the hassle IMO. Using only 20-30% of its capacity will not harm it in any way, thats what most people with 1000w+ use with small bursts higher when they bench or game.

Heck, my SB-E rig is idling at 95w right now, and I have a 1200w supply.
Click to expand...

Depends on how expensive power is.

Most PSUs have lower efficiency at their low end. The Plus certifications only measure 20%, 50% and 100%. I'm folding on my 950 right now, and my 750W PSU is probably only pulling 250-300W with all my accessories going.
 
Chrisroman said:
If anything it will just run cooler and last longer.
Click to expand...
That's incorrect.

@OP, your system isn't pulling 562W at peak, it peaks well below 400W ;)
What specific 1KW PSU do you have?

ps. the VGA swap you intend to do is very poor, don't do it.
 
Power supply manufacturers will want to market their high-end power supplies, since they make more money off of it. It's not necessarily the truth. It's like the argument of single rail vs multi-rail, how XFX markets single rail as much better, when in reality it is not. Always take marketing with a grain of salt.

A unit that runs more efficiently will run quieter at the same loads. As for longevity, there are so many variables that affect it, from the components used, to maintenance (regular dusting), etc. that there's no real way to measure it. And today's power supplies are too new to get accurate comparisons of their lifespan.

Additionally, there's a world of a difference between a 500 watt Platinum and a 1000 watt 80+ in terms of running cooler. The Platinum will run cooler due to higher efficiency. The Platinum will in general (not true in all cases, that's what [H]OCP and Jonnyguru reviews are for) use better quality components, and will last longer. If you want silence, get one of the Seasonic fanless power supplies.
 
Chrisroman said:
So you believe a unit that is under less load will not run cooler, quiter, and last longer? Please explain. Quick search seems to favor my conculsion, but I'm very interested to hear what you have to say on this.

http://www.batterypoweronline.com/images/PDFs_articles_whitepaper_appros/RockwellAutomation.pdf

http://www.silverstonetek.com/techtalk_cont.php?tid=wh10_005&area=usa
Click to expand...

I think I already explained it a few dozen times on this forum.. anyway, here we go again. :)

run cooler -> directly related to the unit's efficiency, fan and fan profile.
ie. a 1KW Platinum unit running a 100% load will be cooler than a 1.5KW Bronze unit(for example) at any load.. provided similar fans and fan profiles.

run quieter -> same as above, countless examples can be given.

last longer -> why would it?.. there are too many variables, most of them far more important than load(inside spec load that is), cap quality, cooling efficiency, ambient temps etc

There are far too many variables to take into consideration when you're talking about quality units, and load isn't the most important one.. on the contrary, for example, a unit with a "Seasonic X" type fan profile, will run a lot warmer at very low loads than it would at loads around 50% and above

ps. linking a Silverstone marketing page does not qualify as an argument, quite the opposite.
 
Tsumi said:
Additionally, there's a world of a difference between a 500 watt Platinum and a 1000 watt 80+ in terms of running cooler.
Click to expand...

a 500 watt 80+ platinum at 100% load is 89% efficent.
a 1000 wat 80+ gold at 50% load is 90% efficient.

Basically the same. In fact theres only a 2-3% efficiency diference between gold/platinum. at 20%, 50%, 100% loads. http://en.wikipedia.org/wiki/80_PLUS

I find in hard to believe that even with the same efficiency ratings that the 500 watt platinum PSU running at 100% isn't going to be hotter and fail sooner than the 1000watt gold running at 50% capacity. I suppose it's possible, and I'm no electrical engineer, but I'd honestly like to see some valid tests that demonstrate that to be true.


Profumo said:
I think I already explained it a few dozen times on this forum.. anyway, here we go again. :)

run cooler -> directly related to the unit's efficiency, fan and fan profile.
ie. a 1KW Platinum unit running a 100% load will be cooler than a 1.5KW Bronze unit(for example) at any load.. provided similar fans and fan profiles.

run quieter -> same as above, countless examples can be given.

last longer -> why would it?.. there are too many variables, most of them far more important than load(inside spec load that is), cap quality, cooling efficiency, ambient temps etc

There are far too many variables to take into consideration when you're talking about quality units, and load isn't the most important one.. on the contrary, for example, a unit with a "Seasonic X" type fan profile, will run a lot warmer at very low loads than it would at loads around 50% and above

ps. linking a Silverstone marketing page does not qualify as an argument, quite the opposite.
Click to expand...

All valid points, thanks for sharing. And after looking at the 80+ specs I can see PSU's seem to hit peak efficiency at about 50%, so I suppose it's totally possible to not run cooler and quiter with a lower load since it is, in fact, running less efficient. But thats only the spec.. like you said build quality and design come into play as well... very interesting ...
 
Chrisroman said:
a 500 watt 80+ platinum at 100% load is 89% efficent.
a 1000 wat 80+ gold at 50% load is 90% efficient.

Basically the same. In fact theres only a 2-3% efficiency diference between gold/platinum. at 20%, 50%, 100% loads. http://en.wikipedia.org/wiki/80_PLUS

I find in hard to believe that even with the same efficiency ratings that the 500 watt platinum PSU running at 100% isn't going to be hotter and fail sooner than the 1000watt gold running at 50% capacity. I suppose it's possible, and I'm no electrical engineer, but I'd honestly like to see some valid tests that demonstrate that to be true.




All valid points, thanks for sharing. And after looking at the 80+ specs I can see PSU's seem to hit peak efficiency at about 50%, so I suppose it's totally possible to not run cooler and quiter with a lower load since it is, in fact, running less efficient. But thats only the spec.. like you said build quality and design come into play as well... very interesting ...
Click to expand...

I said regular 80+, which has 80% efficiency at 50% load. Also, just because it has that level of certification does not mean that it's efficiency are at those levels. It can be higher or lower, depending on the model.

Let's take your hypothetical model, and say they're the same efficiency at the same load levels. That means the same amount of electricity is being converted to heat, which means the heat output of both power supplies will be the same. Now, if the 1000w power supply runs cooler, that will ONLY be because it has larger heatsinks to dissipate the 1000w load. And the difference will be very slight, maybe 2-3 C at the most, which in power supplies, means nothing.

Also, most of the time a computer will be idling. Idling for most single GPU computers is under 150 watts. A 1000w Gold power supply can have an efficiency as low as 70% at these low load levels. The 500w Platinum will be at least 90% efficient.
 
Profumo said:
That's incorrect.

@OP, your system isn't pulling 562W at peak, it peaks well below 400W ;)
What specific 1KW PSU do you have?

ps. the VGA swap you intend to do is very poor, don't do it.
Click to expand...

I'm using the eXtreme PSU calculator, that's what it says. Short of buying a kill-a-watt that's the best I can come up with for power consumption calculations.

I have a OCZ ELITEXSTREAM 1000W power supply. Already blew it up once, although that was due to bad input power - I have a 1200W UPS to prevent *THAT* from happening again.

The VGA swap is to save power - the server is my phone/fax server, my BeyondTV recording server, my media server, and my file server. I'm almost never on it directly because I have a very nice workstation with six monitors to use and a HTPC for watching movies on a 46" LCD TV. Besides, I sold the HD 4870 for $100 which these days is pretty good. If I decide I need more power on the server I'll throw in one of my HD 6950 2GB cards (I have Crossfire) which would effectively end the minimum power consumption question as well. I'm trying to figure out the maximum performance for what it does do for the minimum electrical power *WITHOUT* blowing up the PS.
 
There really is no minimum safe level that you can reach in a normal computer setup. Maybe below 20 watt pull for extended periods of time can damage it, but your computer will never drop below 100 watt usage.

Those PSU calculators are never accurate, and cannot be trusted.
 
Profumo said:
I think I already explained it a few dozen times on this forum.. anyway, here we go again. :)

run cooler -> directly related to the unit's efficiency, fan and fan profile.
ie. a 1KW Platinum unit running a 100% load will be cooler than a 1.5KW Bronze unit(for example) at any load.. provided similar fans and fan profiles.
Click to expand...

For example, say we have:

1000w PSU with 89, 92, 90% efficiency at 20, 50, 100%
and
1200w PSU with 83, 87, 85% efficiency at 20, 50, 100%

Now say we are drawing 240w. That is 24% of the 1,000w, and 20% of the 1,200w.
For the 1,000w psu we will say it gets 89.5% efficiency at this load. (This fits the above efficiency curve pretty well)

The waste heat is as follows:

1000w unit
240 / 0.895 = 268.16w drawn from the wall, so 240 - 268.16 = 28.16w is wasted, and turned into heat.

1200w unit
240 / 0.830 = 289.16w drawn from the wall, so 240 - 289.16 = 49.16w is wasted, and turned into heat.

It would be fairly easy to calculate the waste heat with a different load and different efficiency.

load / efficiency = power drawn from wall
power drawn from wall - load = waste
 
rtangwai said:
I'm using the eXtreme PSU calculator, that's what it says. Short of buying a kill-a-watt that's the best I can come up with for power consumption calculations.

I have a OCZ ELITEXSTREAM 1000W power supply. Already blew it up once, although that was due to bad input power - I have a 1200W UPS to prevent *THAT* from happening again.

The VGA swap is to save power - the server is my phone/fax server, my BeyondTV recording server, my media server, and my file server. I'm almost never on it directly because I have a very nice workstation with six monitors to use and a HTPC for watching movies on a 46" LCD TV. Besides, I sold the HD 4870 for $100 which these days is pretty good. If I decide I need more power on the server I'll throw in one of my HD 6950 2GB cards (I have Crossfire) which would effectively end the minimum power consumption question as well. I'm trying to figure out the maximum performance for what it does do for the minimum electrical power *WITHOUT* blowing up the PS.
Click to expand...

Your system would run off a PICO PSU without any issue whatsoever.

The EliteXStream 1000W is a very good Impervio built unit, but it's a very poor choice for the system you listed considering it's probably hardly ever seeing loads above 100W now that you've changed your VGA, and that means <70% efficiency.

Stop using power draw calculators, they're trash.. but if you absolutely must, use this.
 
You must log in or register to reply here.
Back
Top