How much power a PSU is drawing

Orbit · Post by **Orbit** » Fri Feb 04, 2005 5:23 am

Well, I've read the articles on Power supply fundamentals and a few forum threads on similar topics, but this is more just to try and settle something.

On IRC I was chatting with some people, and when one of them said they were about to buy a 520W power supply, I joked at what the hell they were going to be powering with it. I then pointed out the power supply fundamentals article, in which it was stated the highest power draw you've seen was ~180W from a desktop system. After that, I was laughed at, ridiculed, called a complete tosser etc etc because someone there who claims his friend has 'built a power supply' says that its wrong.

They talk about this graphics card drawing 110W, the CPU drawing even more, plus the hard drives drawing 25W balblblalbal, they're adding it all up and telling me I'm wrong. (for the record, I never said I knew anything about PSUs, simply pointed them to the site, and said I believed you guys did).

So, any ammunition you can give me to mock their ignorance? or are they right?

Mats · Post by **Mats** » Fri Feb 04, 2005 5:35 am

Read this. However, I'm not sure that it's a completely good article, the power consumption for some components seems strange.

sthayashi · Post by **sthayashi** » Fri Feb 04, 2005 6:08 am

Tell 'em to read this article (6800 ultra maxes out at 72W), and then find their processor on this page.

So with a theoretical Intel SLI system, you could have a max power draw of: (2*72) + 115W = 260W. Lets add on 10-20 watts for hard drive and other peripherals again. Now tell me why anyone needs a 520W PSU to run their 280W computer again?

SometimesWarrior · Post by **SometimesWarrior** » Fri Feb 04, 2005 9:02 am

The Xbit-labs article sthayashi linked is good. Also, here's one where actual CPU power consumption is measured. But the strongest evidence comes from all the Kill-a-watt measurements scattered throughout this site, the forums, and elsewhere. It says right on the display, "This PC is using 130W at full load," or whatever. If they claim that the watt-meter is lying, don't waste another word on them. They're probably one of those flame warriors documented on A Netizen's Guide to Flame Warriors (great site, by the way).

Sthayashi's example about the Intel SLI system is, IMO, flawed. such a system will draw almost all of those 260W from the 12V line of the power supply. That's 22 amps just for the CPU and video, which is the max continuous draw allowed from a Super Silencer 400W. In such a rediculous setup, a 520W power supply might be necessary. A more realistic setup is an Athlon64 3200+ and a single 6800GT. The processor maxes out at ~55W, as does the video card. A hard drive could only reach 25W at spin-up, and documentation on modern drives says that reading/writing will only use about 10W. The motherboard might use another 10-20W maximum, and nothing else in the system requires active cooling, so you can assume other components don't have significant power draw. The whole system will use 150W maximum, which any decent 300W power supply could handle without problems or voltage irregularities.

monkeh · Post by **monkeh** » Fri Feb 04, 2005 11:09 am

the 12volt rail seems to take most fo the current draw on my system
as i watch it drop to 11.7 volts and then my PC reboot's.

it's fine at moderate overclocks but when i push it harder then the current draw increases

My processor is capable of doing 3.9ghz, but the PSU isnt

My pc's isnt exactly crammed full of kit either

P4 2.8c @3.6ghz
2x512mb
2 diamond maxtor 9's
1 DVDr
1 CDRW
onboard NIC and soundcard
X800 XT (slightly overclocked)
CWT 420W powersupply

So i DO need atleast a 500W PSU.

Post by **MikeC** » Fri Feb 04, 2005 11:24 am

Hmmm, looks like the timing of the new REAL SYSTEM POWER REQUIREMENTS section I just posted on page 3 of the Power Supply Fundamentals & Recommended Unitsarticle is perfect for Orbit.

Post by **MikeC** » Fri Feb 04, 2005 11:31 am

monkeh wrote:CWT 420W powersupply

So i DO need atleast a 500W PSU.

Actually you just need a better power supply. The labelling on PSUs is often atrocious and CWT is definitely one of the culprits. I can guarantee you that your unit doesn't even manage to put out more than 200~250W max before it craps out. If you are really interested, send it to me. It won't take but a few minutes at a real 420W load on my load tester to have your CWT reduced to what it really is -- a pile of stinking metal.

Sorry for the tough language, no offense intended to you monkeh, just all the nasty lying marketers.

monkeh · Post by **monkeh** » Fri Feb 04, 2005 11:35 am

Doesnt CWT make PSU's for antec, so wouldnt that make antec's psu crap as well ?

it may well be a poor psu, but most the reviews i saw when i bought it 1 or 2 years ago said it was identical to the Antec units ! if i rememeber correctly Epox reccomended it to me as well, and i trusted them... then thre their crap motherboard in the bin. so your probably right

i am looking at the Antec phantom 500w psu when it comes out, as that should be quite efficient ?

Post by **MikeC** » Fri Feb 04, 2005 12:02 pm

monkeh wrote:Doesnt CWT make PSU's for antec, so wouldnt that make antec's psu crap as well ?

Could be. But the fact is, I have not paid that much attn to Antec PSUs except for the ones reviewed -- and they met the published claims for power output. But my point is that the total power drawn by your components cannot possibly reach 420W -- or anywhere close! -- and if your PSU is shutting down then it can't possibly provide anywhere close to 420W.

It could also be that maybe the total load is not that high, but all on the 12V line, which could be then be overloaded -- it's not until ATX12V v1.3 and up that the 12V current capacity started getting beefed up. (And if this is the case, my previous comment about your PSU being junk is a dumb jump to conclusions.

)

Post by **Rusty075** » Fri Feb 04, 2005 12:48 pm

The extent of that overclock may have more to do with the rebooting than the PSU. The higher the OC, the more sensitive the system is to the little variations in voltage. And 2.8 to 3.9 is pretty big. At that kind of amperage the failure could be as much the fault of the VR's on the mobo as the PSU.

monkeh · Post by **monkeh** » Fri Feb 04, 2005 1:46 pm

Hey Mike i took no offence, and appreciate your advice.

davidstone28 · Post by **davidstone28** » Fri Feb 04, 2005 5:17 pm

SometimesWarrior wrote:They're probably one of those flame warriors documented on A Netizen's Guide to Flame Warriors (great site, by the way).

That site is hilarious! thanks for the link

Orbit · Post by **Orbit** » Fri Feb 04, 2005 6:26 pm

Nice addition to the article mike.

Its like trying to convince people millennia (or millenniums?) ago the earth is round.

ITS NOT FLAT OK FFS!!!

GlassMan · Post by **GlassMan** » Sun Feb 06, 2005 7:23 am

When you are using current generation hardware the total power draw is not that critical, the available 12 amperage is. Older psu's available power was concentrated on the 3.3-5v rail, as vid card and cpu's were using 5v power. You need an ATX12v1.3 or higher for a modern computer.
Remember during boot up all the drives are being powered up, cpu capacitors are being charged up and the cpu is under 100% load. These are all 12v loads, and can be higher than 100% load under the O/S, as the drives are generally already powered up.
For silent computing it is important to have the power draw be high enough for the psu to approach maximum efficiency, which tends to argue for a lower power psu, but you also must have the proper power available. A psu with 15 amps of available 12v power will have difficulty starting an A64 system with a standard configuration.

Bitter Jitter · Post by **Bitter Jitter** » Sun Feb 06, 2005 9:56 am

GlassMan wrote:A psu with 15 amps of available 12v power will have difficulty starting an A64 system with a standard configuration.

Can you quantify a 'standard configuration'?
I have no problem with only a tiny 15a on the 12v rail running an Athlon 64 2800+, Radeon 9800pro......
The maths: 15 x 12 = 180
180w just for the 12v rail and bare in mind the whole computer still does not use only 12v.
The max peak I have seen at the plug is 190w….

So why the need to panic about this 12v rail issue all the time?

burcakb · Post by **burcakb** » Sun Feb 06, 2005 10:21 am

15A equipped 300W Fortron ran my Athlon64 just fine. It did fail however, when I added a second disk and swapped the R9200 for a R9700. 15A is a bit on the low side today if you're going for more than one hard/optical disk

p5 · Post by p5 » Sun Feb 06, 2005 11:00 am

my 15A 300w fortron (with 120mm evercool @5v), ran an athlon64 overclocked (2.0ghz to 2.4ghz), a r9700non pro (had no problems at very high clocks 405/324), 2 hard drives, 2 optical drives, sound card, tv card etc.

All completely fine.

As does my current 'older generation' Seasonic SS-300GS (similar to FS) modded with a practically silent panaflo 80mm L1A running at ~4-5v.

The max i ever saw being pulled from the fortron was 180watts AC, and 168watts AC for the more efficient Seasonic.

Like Bitter Jitter says, 15A equals to 180watts DC, and my entire usage has never reached ~130watts DC, and that's on all rails +3.3v +5v and the +12v.

GlassMan wrote: A psu with 15 amps of available 12v power will have difficulty starting an A64 system with a standard configuration.

15A typical is more than adaquate to power a typical A64 system providing it's from a decent power supplier manufacturer who doesn't lie about it's specs.

Also you mention 'starting an A64 system', isn't this something else, max inrush current at bootup or something? Again I'm lead to believe this is something the better power supply manufacturers provide more of.

Maybe the other power supply 'pros' are more knowledgeable in this area.

Also I've owned two CWT power supplies previously with the whole "CWT MAKE FOR ANTEC" in mind, thinking they made decent power supplies.

The 450watt 'BDP' model blew up on me, fair enough it was being cooled by 1 panaflo 80mm L1A @ full speed rather than it's stock dual high speed fans, maybe these things need their rated high airflow fans to keep them running sweet.

The other was a 400w CWT, which was relegated into someone else's machine because it was unstable in games in an even lesser system that the later purchased 300w fortron and 300w seasonic could, back then it was just an xp1700, r8500le, one maxtor hd, one optical drive. And as mentioned above, the 300w fortron and seasonic can perfectly power an overclocked A64 @ 2.4ghz, r9700np at high clocks, 2 hds, 2 optical drives.

Post by **MikeC** » Sun Feb 06, 2005 11:50 am

Rusty075 wrote:The extent of that overclock may have more to do with the rebooting than the PSU. The higher the OC, the more sensitive the system is to the little variations in voltage. And 2.8 to 3.9 is pretty big. At that kind of amperage the failure could be as much the fault of the VR's on the mobo as the PSU.

Hmm yeah, very true!

Post by **MikeC** » Sun Feb 06, 2005 11:58 am

Also you mention 'starting an A64 system', isn't this something else, max inrush current at bootup or something? Again I'm lead to believe this is something the better power supply manufacturers provide more of.

Yes, this is true of better manufacturers. It's difficult to track what the turnon power surge is w/o a logging device of some kind, but just carefully watching a Seasonic Power Angel or Kill-a-Watt, it never seems as high as the max power I could obtain from a system. It's close to what folding @ home demands of a system, which is about 10-20% less than max.

Bitter Jitter · Post by **Bitter Jitter** » Sun Feb 06, 2005 12:13 pm

MikeC wrote:It's difficult to track what the turnon power surge is w/o a logging device of some kind, but just carefully watching a Seasonic Power Angel or Kill-a-Watt, it never seems as high as the max power I could obtain from a system. It's close to what folding @ home demands of a system, which is about 10-20% less than max.

I have noticed a similar draw on my computer, about 150w which is what folding would be in windows roughly.
The only component which would need more than average power at startup is the hard drive as the motor gets the spindle up to speed.

GlassMan · Post by **GlassMan** » Sun Feb 06, 2005 3:13 pm

Can you quantify a 'standard configuration'?

I would define a standard setup as having a current generation cpu, gpu, and 3 drives, hard or cd-rom, as well as a few fans.

15A typical is more than adaquate to power a typical A64 system providing it's from a decent power supplier manufacturer who doesn't lie about it's specs.

Adaquate, from an excellent psu, IMO, but if you upgrade the more will likely disappear from your thoughts. The low power draw of fans normal at SPQR helps as well. Will you keep the psu when you upgrade to a new platform?

Yes, this is true of better manufacturers. It's difficult to track what the turnon power surge is w/o a logging device of some kind, but just carefully watching a Seasonic Power Angel or Kill-a-Watt, it never seems as high as the max power I could obtain from a system. It's close to what folding @ home demands of a system, which is about 10-20% less than max.

I yield to your expertise on this point, non the less many 15amp psu's (not as well engineered as the exceptions pointed out) fail to boot an A64 system. Still around 150watts is a lot considering the video card is drawing minimal power as there is no 3d activity.
Instead of "find it difficult" I should have said many (probably most) won't. I participate at the pcper chaintech forum, and regularly a proud owner of an ATX12v1.0 psu upgrading from an Athalon want to know why his new A64 won't start, the fans start to spin, then nothing.
I am surprised at the reported power consumption while folding as it is not that stressful, the video and ide system is not very active. I had to overclock my A64 and 6800GT to the max to blow the 175 watt circuit breaker on my APC ups while running 3dMark and folding. I guess the breaker was sticky.
Super Bowl is getting ready to start, but if you recommend any 1.0 for a new system you are very likely not doing the person you are talking to a favor.

GlassMan · Post by **GlassMan** » Tue Feb 08, 2005 11:23 am

http://www.silentpcreview.com/article217-page3.html

peterson · Post by **peterson** » Sat Feb 12, 2005 7:39 am

MikeC wrote:Hmmm, looks like the timing of the new REAL SYSTEM POWER REQUIREMENTS section I just posted on page 3 of the Power Supply Fundamentals & Recommended Unitsarticle is perfect for Orbit.

Great reading Mike!

One small Q. Will a 350W phantom be enough for a 3.0prescott/6800GT/2 HDDs/1 DVD-ROM/Audigy 2ZS/some fans/t-balancer? If yes, how much headroom. If yes - finally!

sthayashi · Post by **sthayashi** » Sat Feb 12, 2005 11:41 am

peterson wrote:One small Q. Will a 350W phantom be enough for a 3.0prescott/6800GT/2 HDDs/1 DVD-ROM/Audigy 2ZS/some fans/t-balancer? If yes, how much headroom. If yes - finally!

Yes. At least 100-150W of headroom if not more. Your described system isn't all that much more powerful than the high end P4 listed.

GlassMan · Post by **GlassMan** » Sat Feb 12, 2005 6:02 pm

One small Q. Will a 350W phantom be enough for a 3.0prescott/6800GT/2 HDDs/1 DVD-ROM/Audigy 2ZS/some fans/t-balancer? If yes, how much headroom. If yes - finally!

You might want to look at this comparison. (From SPCR front page)

EDIT: PLEASE EMBED long URLS!!!

Mats · Post by **Mats** » Sat Feb 12, 2005 7:06 pm

GlassMan wrote:
One small Q. Will a 350W phantom be enough for a 3.0prescott/6800GT/2 HDDs/1 DVD-ROM/Audigy 2ZS/some fans/t-balancer? If yes, how much headroom. If yes - finally!
You might want to look at this comparison. (From SPCR front page)

ööhhh, why? The P4 is not even in that test!!?

GlassMan · Post by **GlassMan** » Sat Jun 04, 2005 5:41 am

p4 presscot out draws all of them

GlassMan · Post by **GlassMan** » Sat Jun 04, 2005 6:34 am

From http://www.silentpcreview.com/article217-page3.html

Note that modern systems (after 2003) tend to draw most of their power off the 12V line. ATX12V PSU Design Guideline v1.3 and beyond specify much higher current for the 12V line than in the past. As a result, older PSUs that deliver high enough total power may fail to deliver enough 12V current to satisfy new systems.

This is what I've been saying