An actual power supply measurement

[KF]

I measured the input to a computer at the power cord with an ammeter. It came out 0.65 amps at idle/normal. That's 78 watts input total. 0.97 amps under prime95 (blend) load. That's 117 watts input total.

The mobo is an MSI K8N Neo2 Platinum, with a Winchester core Athlon 64, at 1.42 Volts, OCed to 2250 MHz.

It makes you wonder if a 500 watt PS is really necessary.

Cool and Quiet is disabled. The HD is spinning, not powered down, and there is just a CPU fan and the PS fan, no case fan running, because the case is currently open while I'm in testing/experimental mode.

On a second computer, it came out somwhat differently. It has a DFI NFII Lanparty mobo, with a Tbred Athlon XP at 1.58 V, OCed to 1870 MHz. It comes out 0.93 to 0.98 at idle. It wanders a bit depending on what is running. That's about 120 watts total. If I fire up prime 95, it goes to 1.16 amps, which is 139 watts.

I tried burning a CD to see what the power-delta is. That would give an estimate of the extra power used to burn. With a blank in the drive, and Roxio Easy CD Creator running, it starts out at 1.08 A and wanders down to 1.03 before the burn starts. I think the wandeing is the CD drive spinning up to identify the CD and spinning down to a lower idle rpm. While burning it goes to 1.14 at burn speed of around 30x fot that blank. So the difference is 0.06 to 0.11. So burning a CD draws about 13 watts extra.

Just putting a CD in the drive makes the amps go to 1.20 for a second, then it backs down to 1.04 or so. When you play a CD it mostly near 1.04, while the green light goes on about once every 1-2 seconds. I believe the OS reads breifly at some multiple of the speed (10x?) and buffers ahead enough to play continuously.

Just to do the math, if all the power were to the 12 volt rail (which it isn't) , 120 watts would be 10 amps. Since powers supplies are estimated at 80% efficient, the PS would be putting out 8 amps. In other word the PS couldn't possibly be supplying over about 8 amps on the 12V rail in actual use. Is 28 amps on the 12 volt rail really necessary?

The numbers are not a surprise to me. Ballpark numbers give something like this.

 

[Navid]

Originally posted by: KF
I measured the input to a computer at the power cord with an ammeter. It came out 0.65 amps at idle/normal. That's 78 watts input total.

No, it's not! Power is V I Cos(Theta), where Theta is the phase difference between the voltage and current. So, the power is even less!

I am curious how you measured the current!

 

[KF]

> am curious how you measured the current!
Somebody's interested?

I considered inserting low value resistors in the PS leads and measuing the voltages. If you use a resistor estimated to drop around 0.1 to 0.2 V it shouldn't interfere, but I really didn't want to butcher a nice PS like that. The high current rails have multiple wires, to complicate it. I could have made up some extension leads with plugs for the purpose, but obtaining some of the "other-end" plugs is problematic.

Electricians have something to measure AC current without breaking into the wiring: a clamp-on ammeter. I'm way behind the times, or I was. I got a very nice, expensive analog multimeter when I was young because "it would last a lifetime." Electricity doesn't change, right? IAC, the meter I got recently was also a decent digital multimeter, besides the clamp-on loop.

The specs for AC current are +/- 1.5% reading + 4 counts. On the 20 amp range, it counts down to 0.01 amps

So then I bought a 6 foot extension cord (for air conditioners?) that had the three leads visible, like a 3 wire lamp cord. I separated one lead so I could clamp the ammeter loop around it. No exposed wires. It is very easy, and somewhat informative, to now be able to measure the current of anything that plugs ito an electric outlet. Since watts = Current X 120 volts, you know the wattage.

Yeah, you are right about my negelectig the power factor. I am under the impression that it is not significant for power supplies, but I have no firm knowledge. I thought current and voltage only got significantly out of phase for motors heavily loaded. As you say. the calculation is for a worst case. Actual power would be less if out-of-phase.

 

[Navid]

Interesting!

You are right on all accounts!
After posting, I realized that the error would be negligible specially if your PSU has power factor correction (PFC). Sorry about that.

But, something doesn't add up!
You don't need all the power that a power supply can provide all the time. You need margins. When the power supply warms up, its maximum output power capability drops. That's one reason people get higher power PSUs.

But, still I did not expect such a low power.

The meter you describe with a loop around the wire; is it for AC or DC only?

 

[Navid]

Do you have a link for the meter that you used?

 

[KF]

>The meter you describe with a loop around the wire; is it for AC or DC only?

AC only. Because it works by induction. It is really a one loop transformer. The loop is a large diameter, and it is "air core", so the coupling is very low. Under those conditions, the placement of the current carrying wire is not critical, just so long as it is inside the pickup loop. That is my understanding of the theory of operation.

A clamp-on ammeter
They call it a snap-around ammeter. It looks like mine.

another one
They call it a clamp ammeter.

> You need margins. When the power supply warms up, its maximum output power capability drops. That's one reason people get higher power PSUs.

Somebody said they do the measurments on the label at 0 degrees C. (I don't know, but I doubt it.) That would definitely be misleading. To me, it SHOULD mean, the PS can do the rating 24/7 at room temperature. Who knows what it really means.

My problem is that people are under the impression their computer really uses 400 watts, or that it uses 400 watts under some circumstances. BS.

Here is the how they put it:

" I have 4 case fans, 2 HDs, a CD burner, and a DVD burner. I know I need at least a 500 watt PS for all that. What is best one for a reasonable price. I found this one, a Garbagetron 700 for $15. Rated at 700 watts. Do you think that would do the trick?"

The problem is that any PS that actually delivered what the label promised would work. There is no way to know if the Garbagetron will.

 

[mechBgon]

A typical modern passenger car can cruise at 65mph on probably <20hp. Want a 20hp engine in your next car...?

~ didn't think so.

I do agree about the GarbageTron brands out there, there ought to be some accountability for marketing junk.

 

[boshuter]

My problem is that people are under the impression their computer really uses 400 watts, or that it uses 400 watts under some circumstances. BS.

Why is that a problem for you?

 

[steamnputer]

This is interesting,
But ,
Did you measure the neutral or the hot lead, this will give you different readings after that....
A clamp on Ampeter (comonly called a amprobe by electricans) are only acurate when used on certain types of loads. Most PSUs are switching these tend "fool" these types of meters, and infact fool many digital meters cheap and exepensive, as far as figuring power factor that would be neglible on the high side of a an inductor in this manner and would be better expressed on the low side. Truthfully the analog meter is the best choice if it is capable to be put into the line(series). If I were to take a guess it would be more along the lines of 1 amp 120VAC60Hz. 2 things tend to make the reading off on amprobes PWM and capicators,which power supplies have both.

Clamp-on meters work on either AC or DC they are not normally specific, the old analog ones are best for both. The meter section of the probe might be current specfic but the clamp is not. Many better power supplies rate in RMS not peak, RMS is the square root of 3 or .73 or about 75% of your top end number. If you rate a 500W PSU and its RMS then it would actual peak around 600W wereas a 500W that is rated peak is more likely to run best at around 350W.- I did not do the numbers on those just guessed-
Any CE or CSA certified device is measured from standard room temps, and during there rated loading cycle(i used to test electrical motors) so they would start at room temp and rate it based on certain defined procedures, if they have those ratings(same thing applies to UL).
The PSU changes AC to DC, the DC rating for input(usa) is 120 volts or so. In reality its much higher closer 200(you are up to about 130 watts now @.65 amps) volts peak on the low side this is converted to 12 or so volts on multiple rails on better supplies, in order to make these rails work with each other you need to make one work as well as the other so 2 250W rails(lets say) work well together. If you include the other output voltages, the filters, capicators,diodes, loses etc etc you start talking about 600W.

So why do we need so much wattage on out new machines? mainly startup, the typicalinrush on any inductor, is 8 times the rated load of the device, so all those inductor loops, fan motors, hdd's CD's as pointed out require incredible momentary power just like a car driving down the road getting up to speed. They are also more sensitve to burps in the power supply that a smaller rated PSU might not be able to absorb if they are stressed out. One thing about the analogy of the car is you require about 20hp, that really is wrong, HP is determined by torque and rpm/ a constant.
the faster your engine is turning, the more HP its putting out, granted the torque may be less. Todays engines have high horse power mainly becasue the spin faster and don't have to put out more torque.

 

[Ketchup]

Hey, thanks for the interesting read. Just FYI, I work at Staples and opened up the case on a new Compaq we just started carrying. I was pretty surprised to find that the power supply used is 380 WATTS! For a Celeron with integrated video! I guess the days of the 200 watt power supplies are gone.

 

[mechBgon]

My point about the horsepower of the car engine is the same point you made about momentary power demands in a computer... sure it can cruise on 20hp, but you still want the 200hp V6 when you're trying to hammer a 30-to-70mph acceleration to merge onto the freeway with four passengers and six stuffed suitcases in the car.

When the computer is swapping levels in HalfLife 2, that's my concept of "merging onto the freeway." 1GB of RAM, a northbridge, a southbridge, a sound card, a NIC, one or two hard drives, a 70W+ video card and an 80-100W CPU are all screaming for power at once.

While there are people who say that 300W is plenty, I noted that both Dell's midline systems (that can take an add-in video card) and some of Shuttle's SFF systems have now moved up to 350W PSUs. Hmmmm.

 

[jamesbond007]

Well, I know next to nothing against you guys in terms of electricity and power, but I just got an APC UPS unit a few days ago. To my amazement, EVERYTHING on my desk uses 199W of power, according to my APC PowerChute software. This includes a 17" LCD, Dell XPS Gen4 tower with 480 or a 460W PSU (I can't remember), speakers, cable modem, switch, HP PSC unit, the works. I thought it was going to be a very heavy load, but I was wrong! BIG TIME. This is a 1500VA 865W max APC battery backup unit.

I know I offer very little to this thread, but I thought I'd share my personal results.

 

[statik213]

Originally posted by: jamesbond007
Well, I know next to nothing against you guys in terms of electricity and power, but I just got an APC UPS unit a few days ago. To my amazement, EVERYTHING on my desk uses 199W of power, according to my APC PowerChute software. This includes a 17" LCD, Dell XPS Gen4 tower with 480 or a 460W PSU (I can't remember), speakers, cable modem, switch, HP PSC unit, the works. I thought it was going to be a very heavy load, but I was wrong! BIG TIME. This is a 1500VA 865W max APC battery backup unit.

I know I offer very little to this thread, but I thought I'd share my personal results.

Wow, that sounds like a nice UPS, did not know they provided that sort of info.....

 

[jamesbond007]

Yeah, it's *REALLY* nice. It's an APC BX1500 XS series.

http://myweb.cableone.net/twiggy9560/powerchute.jpg

 

[Slaimus]

Also, the power supply itself is only about 75% efficient, so the actual DC power draw (for which the PSU is rated by) is even less.

 

[Amplifier]

Glad someone finally put some numbers to this . Has anyone experienced insufficient power before?

 

[KF]

Of course I agree that you need some reserve for peaks, Mechbegon. How much?


I'm going to argue that the peaks are not much. Take the starting current of the motors in HDs and CDs, or the head actuators. You can project numbers to get you higher than any supply will actually give. It has nothing much to do with the era of P4s or A64s. 250W supplies worked fine with HDs that took about the same power (or maybe more) in the 486 era. If these starting currents were a problem, then you would expect computer crashes when the HD powers up from standby, but in reality that is not the case.

>Did you measure the neutral or the hot lead, this will give you different
>readings after that....
The readings are identical. And according to electrical theory, they have to be. It is a circuit. There is no other path. Current enters on one lead and exits on the other. The only difference between hot and neutral is that neutral is connect to earth.

>A clamp on Ampeter (comonly called a amprobe by electricans) are only
> acurate when used on certain types of loads. Most PSUs are switching these
> tend "fool" these types of meters, and in fact fool many digital meters
> cheap and expensive
Maybe. They generally use switching supplies in anything electronic nowadays because its cheaper. I do not believe they distort the waveform of the power line, so they should not fool an AC ammeter. They swiich at an ultrasonic rate (so they can use tiny, cheap inductors) and filter the transient junk of the power line.

>Clamp-on meters work on either AC or DC they are not normally
> specific, the old analog ones are best for both.
None of the clamp on ammeters work on DC. They can't There is no induced current with DC. You need a changing current to have induction. If you put a wire next to a wire carrying pure DC, there is no induced current.

> Many better power supplies rate in RMS not peak
You have something crossed here. RMS only has meaning for AC. The output of a computer PS is 100% DC.

> RMS is the square root of 3 or .73 or about 75% of your top end number.
A memory burp. RMS is the square root of 2 times the peak of an AC sine waveform.

> If you rate a 500W PSU and its RMS then it would actual peak around 600W
> wereas a 500W that is rated peak is more likely to run best at around 350W.
>- I did not do the numbers on those just guessed-
Like I said, you have two things crossed. If the PS supplied AC, it would make sense.

>The PSU changes AC to DC, the DC rating for input (usa) is 120 volts or so.
>In reality its much higher closer 200 (you are up to about 130 watts now
> @.65 amps) volts peak
The peak voltage of 120 V AC is 169. By convention 120 is used because all the arithmetic works out simple then. When you calculate power, it works out right. The true power actually is Volts times Amps even for AC volts and AC amps. If you used the peak voltage and amperage, you would get twice the correct power.

>So why do we need so much wattage on out new machines? mainly startup, the
>typical inrush on any inductor, is 8 times the rated load of the device, so all those
>inductor loops, fan motors, hdd's CD's as pointed out require incredible momentary
>power just like a car driving down the road getting up to speed.

I think the reason is this: The old 230Watt supplies couldn't really supply 230Watts, but it was plenty enough. Today you do need more power, the extra mainly for the CPU, but to get one to really deliver even 200 Watts, it has to be rated at 350.

 

[tallman45]

Get a $25 KILL-A-WATT Meter and you will know exactly waht you current draw is. Most likely in the 120w-14w range at idle

http://www.p3international.com/products/special/P4400/P4400-CE.html

 

[Insidious]

Go ahead and put that 200W PSU in a modern system.....

I bet it won't even get hot.....












of course that will be because your system didn't boot

-Sid


It's called specmanship. You have to use very specific current values on each of the unique voltages to reach the "max" output advertised. In the mix of a real system.... especially at startup (stationary fan rotors, disk drive motors, uncharged capacitors, etc.)
you need a commercial rating far above the power comsumption you will measure at steady state. (prove it to yourself with an oscilloscope. Watch what your PSU supply voltages do at startup and during evolutions on your computer. The quality of the power signal degrades as you load the supply... significantly! To maintain the quality DC regulation required for stable operation, you will find you need your system requirements to be far, far below the rating of your power supply.

And don't quote me some mindless article you read once..... get a scope and see for yourself.

 

[mechBgon]

I think that we should try to separate these two questions, because they're not equivalent:

1) how much power does the computer use at the wall outlet

2) what wattage of decent-quality power supply will make the computer run stable


Frankly, I don't care what #1 is. It could be 50W, it could be 450W. What matters in the end, really, is that the computer runs HalfLife2 and Doom3 stable until the owner's eyeballs fall out, or whatever it was bought for, and has enough reserve for the likely upgrade path.

My experience has been that high-wattage, good-quality PSUs are the answer to a lot of systems that are getting low EyeballFalloutMark 2005 scores, so that's what I'll be recommending. If the worst thing that ever happens is people spend an extra $20-$40 on their PSU, instead of a bunch of LED fans or cold-cathode lights, then I can live with that.

 

[steamnputer]

>>Did you measure the neutral or the hot lead, this will give you different
>>readings after that....
>The readings are identical. And according to electrical theory, they have to be. It is a ?>circuit. There is no other path. Current enters on one lead and exits on the other. The >only difference between hot and neutral is that neutral is connect to earth.

Yes on paper they should be the same, typically the neutral is within 5 volts of the hot lead across an inductive load operating in its normal range. But you must remeber you are checking current not voltage in a perfect electrical system on paper the current used will all be accounted for. In an ideal setting they(hot and N) are always the same, but electrical distrubtion systems esp in offices or industrial areas may have significant backfeed on the neutral. The neutral does not have to be tied to ground~but it should be. Some are tied to the center of the Y on a transformer it really depends on what your local pwer service does. 99.99%of the time in a house the neutral is tied to ground otherwise bad things tend to happen. Really this is nothing major, I have found in the past in idustrial settings you get odd or false readings esp on flurorescent lighting loads.

>>A clamp on Ampeter (comonly called a amprobe by electricans) are only
>> acurate when used on certain types of loads. Most PSUs are switching these
>> tend "fool" these types of meters, and in fact fool many digital meters
> cheap and expensive
>Maybe. They generally use switching supplies in anything electronic nowadays because >its cheaper. I do not believe they distort the waveform of the power line, so they >should not fool an AC ammeter. They swiich at an ultrasonic rate (so they can use tiny, >cheap inductors) and filter the transient junk of the power line.

I should have been more specfic I apologize, they do create a harmonic(the ultrasonic rate), you tend to see these in tvs am radio that sort of thing on a branch line. Again you will see something but not always what is there really there, unless you are running are really nice electrical meter like a yokogawa. But none of us has 10K to drop on one so that is pretty much mute. They could really tell us wha the lines are doing and wattage PF that sorta thing. I work with PWM AC invterers on a regular basis if you put an amprobe across one esp digital you get readings that kinda in left field or vary so quickly they are not readable. I think on the digital meters it has to do with the electronics, all though I am not sure why. The analog amprobes tend to always read low, that really depends on the carrier the votlage and few other things. These are extreme cases and would be more in line what you would see on the DC side of PSU as they pretty much work the same way.

>Clamp-on meters work on either AC or DC they are not normally
> specific, the old analog ones are best for both.
None of the clamp on ammeters work on DC. They can't There is no induced current with DC. You need a changing current to have induction. If you put a wire next to a wire carrying pure DC, there is no induced current
NO NO NO
If you go to any car shop you will see a clamp on ammeter on there battery testing rig.
I owned clamp on which reads DC just used on a 90VDC motor like 3 months ago untill someone decided they needed more then me. Here is the one I am hopefully getting

http://www.inotek.com/Catalog/extech2et.html

A clamp-on ammter works on a vary simple priciple. It uses laminations which come together when closed, this intern creates a closed loop. When voltages crosses 0(in either direction) a current is produced in these lams in the opposing direction. This current is transformed in a reading at the meter, it escapes me if they use voltage or amperage, but thats minor. ANY voltage moving through a wire creates inductance, this a bland term. I think you confused an inductor(which is also a bland term normally used in AC) If you with some minor variations take a coil of wire wrap it around a core of iron(lets say) apply AC voltage you have created what is called an inductor or choke becuase of inductance then tend limit or "react" AC current down they are sometimes considered a load sometimes they are considered a control it depends on usage. That same coil is applied DC you have created a electro magnet this is considered stricktly a load, both use inductors and inductance, just in different way.

>> Many better power supplies rate in RMS not peak
>You have something crossed here. RMS only has meaning for AC. The output of a >computer PS is 100% DC.
True but they still rate in RMS not peak, not me, its what they do. I think they do it as that is a recognized standard, audio equipement uses too, but I could not say why they use it. Maybe they take the watt rating from the 120 side?

> RMS is the square root of 3 or .73 or about 75% of your top end number.
>A memory burp. RMS is the square root of 2 times the peak of an AC sine waveform.
No RMS is the standard in the united states at 60hz off the mains is .73 or 1.4 depending on usage. The expressiion RMS does have mathmatical meaning also you can do alot of math and still come up with the same thing. Yes it has everything to do with shape of the waveform but that is what every electrican and electrical meter uses(true RMS). I should have stated it that 120VAC is a RMS reading, this implies a DC equivalent of 120 volts even though 120VAC reaches peaks of 169 volts. It makes figuring of OHMS law easier as you state and thats true, but you cannot discount the peaks when you change to another current type.
RMS is used as standard median for many things electrical as used above.


>The peak voltage of 120 V AC is 169. By convention 120 is used because all the arithmetic works out simple then. When you >calculate power, it works out right. The true power actually is Volts times Amps even for AC volts and AC amps. If you used the >peak voltage and amperage, you would get twice the correct power.

you forgot power factor again....

>think the reason is this: The old 230Watt supplies couldn't really supply 230Watts, but it was plenty enough. Today you do need >more power, the extra mainly for the CPU, but to get one to really deliver even 200 Watts, it has to be rated at 350.
I agree the started there own standard and of instead of making themself look like fools they just keep jumping the rating.
I really scares me when I think that if I was running 480 watts pumping through my system it would be rather hot, not to mention the plug and cord.

heres some specs on atx psu's they used 250W and 300W as examples, after reading this I never thought/realized the low voltage rails use the amperage they do.
http://www.formfactors.org/developer%5Cspecs%5CATX12V_1_3dg.pdf

 

[AnnoyedGrunt]

The RMS for a sine wave is actually 0.707 (2^.5/2) not .73 (but that is a small matter). If you multiply 169 * (2^.5/2) you get approximately 120.

RMS stands for "Root Mean Squared" which means that you take some series of data, square the numbers (to get rid of negative values, which usually do not mean negative power but instead power in a different direction, so they still must be counted) then you take the mean of the squared values, and then the square root of the mean in order to factor out the sqauring in step one.

RMS can be used anytime you are trying to find an average of a series of data, and doesn't only work for AC power. Because the load on power supplies is changing, an RMS rating is perfectly valid, but in that case you can't just multiply the peak by .707 (because the load change is dependent on the computer useage and is not going to be sinusoidal).

IMO, a quality 350W supply will be sufficent for any computer these days. SLI users may want to go with a quality 400W for peace of mind.

Also, power supplies typically become more and more efficient as they operate closer and closer to their peak rating, but whether they can remain stable and quiet is another matter. Therefore, some people prefer a higher rated supply in the hope that it will run more reliably @ low load. However, it will typically be running @ a lower efficiency than a lower power supply, so it will use more energy and create more heat (typically) than a lower rated supply would at the same output. I'm guessing that one of the main factors that separates a higher power supply from a low power one is the heatsinking of the transistors, so it's very possible that even though the higher power supply is wasting more energy and heat than a lower power supply, it will still run more reliably and with less temperature due to better heatsinks.

Anyhow, I think that with quality 400W supplies out there, anything more is a waste from a power standpoint (but if you prefer a more powerful one due to features or reputation of the brand - either of which I can see as important factors - I can understand why some might want to buy a higher power/more expensive supply).

-D'oh!

 

[steamnputer]

Originally posted by: AnnoyedGrunt
The RMS for a sine wave is actually 0.707 (2^.5/2) not .73 (but that is a small matter). If you multiply 169 * (2^.5/2) you get approximately 120.

RMS stands for "Root Mean Squared" which means that you take some series of data, square the numbers (to get rid of negative values, which usually do not mean negative power but instead power in a different direction, so they still must be counted) then you take the mean of the squared values, and then the square root of the mean in order to factor out the sqauring in step one.

Like I said it has math implications, but this is how I was taught (on 3 seperate and unrelated occasions) I understand(its quadratic I think...dont remember) and stated its related to the waveform/load.
But this is a standard developed outside of electronics and I would imagine outdates them also. Maybe its luck that its the square root 3, and commonly .7(or 1.4) is used not .707 or .73. I do remeber when I was doing motor testing(I was a engineering tech/dyne tech) that the engineers used .707 on regular basis.
I guess I was going for it be used as average as you state below, and its not just used in ac wave forms, but as an way to convey a median.
[/quote]
RMS can be used anytime you are trying to find an average of a series of data, and doesn't only work for AC power. Because the load on power supplies is changing, an RMS rating is perfectly valid, but in that case you can't just multiply the peak by .707 (because the load change is dependent on the computer useage and is not going to be sinusoidal).
[/quote]
right and thats why I think they use RMS to rate power supplies
[/quote]
IMO, a quality 350W supply will be sufficent for any computer these days. SLI users may want to go with a quality 400W for peace of mind.

Also, power supplies typically become more and more efficient as they operate closer and closer to their peak rating, but whether they can remain stable and quiet is another matter. Therefore, some people prefer a higher rated supply in the hope that it will run more reliably @ low load. However, it will typically be running @ a lower efficiency than a lower power supply, so it will use more energy and create more heat (typically) than a lower rated supply would at the same output. I'm guessing that one of the main factors that separates a higher power supply from a low power one is the heatsinking of the transistors, so it's very possible that even though the higher power supply is wasting more energy and heat than a lower power supply, it will still run more reliably and with less temperature due to better heatsinks.
[/quote]
Yes as I read in the link I poste above they only run 50% eff at light loading, ekkkk
I run an SLI and I would be say that bare minuim of 400W(with a good name PSU)
I run a 480 with active PFC, there have been problems with even 550W psu of lower quality, they are very touchy/fincky and had I not read the specs and forums I would have tried to use a 350W PSU(wonder were that would have lead to).
[/quote]
Anyhow, I think that with quality 400W supplies out there, anything more is a waste from a power standpoint (but if you prefer a more powerful one due to features or reputation of the brand - either of which I can see as important factors - I can understand why some might want to buy a higher power/more expensive supply).

-D'oh!
[/quote]
:thumbsup:
on anything but sli, the amd 64 in general, but its seems to be contained to the nforce4 chipset models right now.
[/quote]