How much power do I need?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
because I can see it from my G15. its only spawning 1 thread, and 25% is being used. sometimes it will load 1 core only, sometimes it will spread the load.

and dude, the whole point of my argument is overestimating. you always over estimate when purchasing a power supply. ALWAYS. a corsair HX520 would melt in a second with that setup.

the HX620 would be able to do it, but like jonnyGURU pointed out it wouldn't endure the haul. end of thread.
 

John

Moderator Emeritus<br>Elite Member
Oct 9, 1999
33,944
4
81
Originally posted by: JAG87
the HX620 would be able to do it, but like jonnyGURU pointed out it wouldn't endure the haul. end of thread.

There is zero proof to substantiate that claim.

 

PurdueRy

Lifer
Nov 12, 2004
13,837
4
0
Originally posted by: John
Originally posted by: JAG87
the HX620 would be able to do it, but like jonnyGURU pointed out it wouldn't endure the haul. end of thread.

There is zero proof to substantiate that claim.

An interesting point, the Corsair guys say the 620 can do SLI 8800GTX's on their forum. It has a 5 year warranty. Seems like they have confidence it will work ok.
 

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
Originally posted by: PurdueRy
Originally posted by: John
Originally posted by: JAG87
the HX620 would be able to do it, but like jonnyGURU pointed out it wouldn't endure the haul. end of thread.

There is zero proof to substantiate that claim.

An interesting point, the Corsair guys say the 620 can do SLI 8800GTX's on their forum. It has a 5 year warranty. Seems like they have confidence it will work ok.


Yes it can work, just like a 100 hp car can go uphill. But if you had to ride for many miles uphill, would you take that, or a 200 hp SUV? And this is not even a fair comparison, because the SUV would not only cost more but also consume more, while in this case the PSU costs more, but it will not consume more (unless the efficiency rating is worse of course).

The point is, this unit will be powering all of your expensive hardware. You dont want IT to have a hard time doing that, just to save a few dollars.
 

PurdueRy

Lifer
Nov 12, 2004
13,837
4
0
Originally posted by: JAG87
Originally posted by: PurdueRy
Originally posted by: John
Originally posted by: JAG87
the HX620 would be able to do it, but like jonnyGURU pointed out it wouldn't endure the haul. end of thread.

There is zero proof to substantiate that claim.

An interesting point, the Corsair guys say the 620 can do SLI 8800GTX's on their forum. It has a 5 year warranty. Seems like they have confidence it will work ok.


Yes it can work, just like a 100 hp car can go uphill. But if you had to ride for many miles uphill, would you take that, or a 200 hp SUV? And this is not even a fair comparison, because the SUV would not only cost more but also consume more, while in this case the PSU costs more, but it will not consume more (unless the efficiency rating is worse of course).

The point is, this unit will be powering all of your expensive hardware. You dont want IT to have a hard time doing that, just to save a few dollars.

I personally could care less if my electronics are sweating :p

If my computer works for 5 years and I saved $80-$100 in the process that's good by me :)

But I agree that overshooting is fine, if you don't mind spending the money.
 

MrDudeMan

Lifer
Jan 15, 2001
15,069
94
91
Originally posted by: JAG87
Originally posted by: PurdueRy
Originally posted by: John
Originally posted by: JAG87
the HX620 would be able to do it, but like jonnyGURU pointed out it wouldn't endure the haul. end of thread.

There is zero proof to substantiate that claim.

An interesting point, the Corsair guys say the 620 can do SLI 8800GTX's on their forum. It has a 5 year warranty. Seems like they have confidence it will work ok.


Yes it can work, just like a 100 hp car can go uphill. But if you had to ride for many miles uphill, would you take that, or a 200 hp SUV? And this is not even a fair comparison, because the SUV would not only cost more but also consume more, while in this case the PSU costs more, but it will not consume more (unless the efficiency rating is worse of course).

The point is, this unit will be powering all of your expensive hardware. You dont want IT to have a hard time doing that, just to save a few dollars.

You clearly don't understand electricity or any type of complex circuitry if you are making such bogus analogies.

720W is overkill by a long shot. I would spend some time writing a reply, but it is clear you don't want to hear it. During normal usage, your computer would be using 1/3 or less of the PSUs potential, and PSUs become grossly inefficient at less than 50%.
 

jonnyGURU

Moderator <BR> Power Supplies
Moderator
Oct 30, 1999
11,815
104
106
Originally posted by: PurdueRy

I personally could care less if my electronics are sweating :p

If my computer works for 5 years and I saved $80-$100 in the process that's good by me :)

But I agree that overshooting is fine, if you don't mind spending the money.

I'd phrase it this way.... Only 5% of consumers would push that PSU to 50% of it's capability (SLI and Crossfire users are probably less than 1% of computer users over all.) And Corsair is counting on that. But it has a 5 year warranty so if it blows up in 5 years, I can get another for free. ;)

Originally posted by: MrDudeMan
You clearly don't understand electricity or any type of complex circuitry if you are making such bogus analogies.

720W is overkill by a long shot. I would spend some time writing a reply, but it is clear you don't want to hear it. During normal usage, your computer would be using 1/3 or less of the PSUs potential, and PSUs become grossly inefficient at less than 50%.

No offense, but you're not necessarily one to talk. You don't know exactly how much DC power the OP needs. And PSU's are only typically "grossly inefficient" at 20% or less, but that can vary based on how the load is distributed.

Components in PSU's, specifically recitifiers, transistors, etc. are subject to hot and cold cycles. And their life span is shortened by prolong exposure to high heat, even if it's within operating temperature.
 

MrDudeMan

Lifer
Jan 15, 2001
15,069
94
91
Originally posted by: jonnyGURU
Originally posted by: PurdueRy

I personally could care less if my electronics are sweating :p

If my computer works for 5 years and I saved $80-$100 in the process that's good by me :)

But I agree that overshooting is fine, if you don't mind spending the money.

I'd phrase it this way.... Only 5% of consumers would push that PSU to 50% of it's capability (SLI and Crossfire users are probably less than 1% of computer users over all.) And Corsair is counting on that. But it has a 5 year warranty so if it blows up in 5 years, I can get another for free. ;)

Originally posted by: MrDudeMan
You clearly don't understand electricity or any type of complex circuitry if you are making such bogus analogies.

720W is overkill by a long shot. I would spend some time writing a reply, but it is clear you don't want to hear it. During normal usage, your computer would be using 1/3 or less of the PSUs potential, and PSUs become grossly inefficient at less than 50%.

No offense, but you're not necessarily one to talk. You don't know exactly how much DC power the OP needs. And PSU's are only typically "grossly inefficient" at 20% or less, but that can vary based on how the load is distributed.

Components in PSU's, specifically recitifiers, transistors, etc. are subject to hot and cold cycles. And their life span is shortened by prolong exposure to high heat, even if it's within operating temperature.

Actually, I am one to talk. I am an electrical engineer (edit: this doesn't mean I'm right. It means I'm not a total idiot and I at least know something about what I'm talking about) and I happen to work on stuff like this. 720W is overkill. Thermal cycling in solid state parts is barely worth mentioning in this context. Process variations or manufacturing flaws already introduced in individual components would be much more likely to cause a problem than using the PSU near its rated output.

What seems to be overlooked here is that most, if not all of the power drawn from the PSU is taken in cycles. The very exaggerated power estimates found on sites like pcpower.com don't take into account start-up versus steady state power conditions. Hard drives don't use 15-25W all day. They use half of that if you aren't accessing them, the dvd/cd drive and floppy drive are using next to nothing the vast majority of the time, and current CPUs aren't using more than 80-90W even at max power, barring a quad core.

I have an E6700 system (no video card mind you) that runs on less than 90W at idle with a hard drive, sound card, dvd drive, floppy drive, 2gb of ram, and onboard NIC. it barely peaks 115W at full load (2 instances of prime95 as well as super pi to exercise IAU, IEU, etc.) With essentially maximized context switching and fully saturated bus I/O, that system still doesn't use hardly any power compared to what most people would think.
 

jonnyGURU

Moderator <BR> Power Supplies
Moderator
Oct 30, 1999
11,815
104
106
Originally posted by: MrDudeMan
Actually, I am one to talk. I am an electrical engineer and I happen to work on stuff like this. 720W is overkill. Thermal cycling in solid state parts is barely worth mentioning in this context. Process variations or manufacturing flaws already introduced in individual components would be much more likely to cause a problem than using the PSU near its rated output.

I don't care if you're an EE or not if you don't have personal experience with the components the OP is using. There's a difference between an EE and someone that works on computers for a living and has experience with a lot of different configurations.

I've had PC's not even boot past post with a pair of GTX's with anything less than 700W and you're saying a 720W is overkill?

And as an EE, how much experience do you have with SMPS and the RMA and RTV of computer components and the reasons why they fail? Telling me thermal cycling is barely worth mentioning? I don't like to toot my own horn, but I've seen failures from thermal cycling on down to caps failing on boards due to high ripple. Experience out weighs an engineering degree.

Not saying you're not a great engineer, I just think you're not thinking outside of your EE box, that's all.

Bye all. Egos are getting to big in here. Enjoy!
 

MrDudeMan

Lifer
Jan 15, 2001
15,069
94
91
Originally posted by: jonnyGURU
Originally posted by: MrDudeMan
Actually, I am one to talk. I am an electrical engineer and I happen to work on stuff like this. 720W is overkill. Thermal cycling in solid state parts is barely worth mentioning in this context. Process variations or manufacturing flaws already introduced in individual components would be much more likely to cause a problem than using the PSU near its rated output.

I don't care if you're an EE or not if you don't have personal experience with the components the OP is using. There's a difference between an EE and someone that works on computers for a living and has experience with a lot of different configurations.

I've had PC's not even boot past post with a pair of GTX's with anything less than 700W and you're saying a 720W is overkill?

And as an EE, how much experience do you have with SMPS and the RMA and RTV of computer components and the reasons why they fail? Telling me thermal cycling is barely worth mentioning? I don't like to toot my own horn, but I've seen failures from thermal cycling on down to caps failing on boards due to high ripple. Experience out weighs an engineering degree.

Not saying you're not a great engineer, I just think you're not thinking outside of your EE box, that's all.

Bye all. Egos are getting to big in here. Enjoy!

I would understand your point about disassociating with being an EE except I literally work on stuff like this. Literally.

Experience outweighs an engineering degree for sure depending on what you are doing. That's obviously true. However, thinking something failed due to thermal cycling when it probably didn't would be where experience can't help you. I wasn't trying to toot my own horn either - just trying to show you I wasn't talking out of my ass. No ego involved at all - if you can prove me wrong, more power to you.

Edit: To answer your question, I worked on SMPS last summer both designing, building, and repairing them. I've also designed one in my spare time.
 

MrDudeMan

Lifer
Jan 15, 2001
15,069
94
91
A great link - two 8800GTXs on a 500W PSU

This aligns pretty closely to what I was saying about my E6700 system. The systems use way less power than most people think.


Edit: Also, he notes what the theme of this thread has been:

Of course, the argument still stands that there should be a reserve, for unexpected power usage spikes - and I agree. The only thing is, I believe that the heaviest runs of multithreaded, hi-res 3DMark06 are a kind of power usage spike, naturally - they stress pretty much everything, from the CPU cores, their caches, FSB, memory, PCI-E, also 25% overclocked via auto LinkBoost, graphics cards, the power-consuming water cooler, HDD, just not touching the DVD. Yes, a 600 W PSU would give me a more peaceful sleep here, but it is obviously not a necessity.

This is what I meant when I said 720W is overkill. It simply is. It's fine if you want to buy it, but its not gaining you anything.


The OP wants to know how much power is needed for:

2x 8800GTX
DVD Drive
680i motherboard
Q6600
SATA x 2

and the guy in the link was using:

2x 8800GTX
DVD Drive
680i motherboard
QX6700
2gb 1066 memory
water cooling


He also OCed up to 3.33GHz with a QX6700, all on a 500W PSU. He was using more power than the OP would.

So now your argument will be "but it won't last long term." SMPS are designed to be used at 60-80% load. He was using a MAX of 429/500W, which is 85%. Since you will never actually hit that type of usage on a system unless you are specifically trying to max out power, it will fall well below the 80% line. He even shows that typical desktop use hovered around 330/500W, which is 66%. Running a game would certainly increase power usage on the video cards, but the cores would stay relatively unused considering the system.
 

Puffnstuff

Lifer
Mar 9, 2005
16,203
4,884
136
Amps on the rails are the most important item in determining whether or not a ps can meet your needs. If a system will not boot initially but will after you disconnect several peripherals then it doesn't have enough amp capacity to meet the hardwares' cold boot inrush current requirement. All electrical devices surge when energized and not all ps's can meet that surge rating. A higher amp ps with a high cold boot inrush rating will boot the system whereas an inferiour unit will not. I've had this happen to me before and it can be frustrating if you don't understand what is going on.
 

jonnyGURU

Moderator <BR> Power Supplies
Moderator
Oct 30, 1999
11,815
104
106

Are you joking or just grasping at straws? He just ran 3D Mark for God's sake. I can run 3DMark in addition to four instances of K7 Burn (one for each core of a quad father) and still have enough overhead to run other apps. That is hardly the example to use when trying to prove that a PC doesn't use "that much power."

Mind you, I'm not saying the OP needs a 1kW PSU, but I still don't think a 620W is adequate.

And you seem to ignore the "quality" point I've made in other posts. For example, there's more emphasis in over all quality in a Thermaltake Toughpower 1kW or 1200W than in any of the models below it. Same with a Cooler Master Real Power Pro 1kW. A Turbo-Cool 1kW is still a better build quality than a Silencer.

For some people, a bigger PSU is just e-penis, but sometimes it just means you really do get a better quality unit.
 

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
LMAO jonny, here come the inquirer links.

MrDudeMan, DID I NOT SPECIFICALLY ASK YOU NOT TO LINK THE INQUIRER RUNNING ON A 550W PSU

Originally posted by: JAG87

and please don't link me to the inquirer article where they run everything with a 550w psu, I dont care. I like my system to be 100% reliable, and with 620W corsair or not you are PUSHING IT. its very easy to do the math.


Yes I believe I did. MrDudeMan, get some hardware, run some tests, make a website which thousands of people are devoted to, and then come back with something to say.


Originally posted by: jonnyGURU

And you seem to ignore the "quality" point I've made in other posts. For example, there's more emphasis in over all quality in a Thermaltake Toughpower 1kW or 1200W than in any of the models below it. Same with a Cooler Master Real Power Pro 1kW. A Turbo-Cool 1kW is still a better build quality than a Silencer.

For some people, a bigger PSU is just e-penis, but sometimes it just means you really do get a better quality unit.


Godly words. You mind if a siggy this jonny? Too bad, I'm doing it.

edit
crap it doesn't fit in my sig
 

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76


taken straight from his post:

a GTX draws 160W
a QX6850 draws 130W

his system:

QX6850 overclocked to 3.3G = 130W + OC = 150W (and I'm being conservative)
GTX x 2 = 160W x 2 = 320W

150+320 = 470

470/620W = 76% load on the PSU

and thats WITHOUT anything else. There is still a motherboard, and all the other "assorted devices" as he likes to calls them. Yet he claims 70% load on the PSU. Yes, maybe when he is playing cs source or supreme commander thats true, but thats not the theoretical maximum load. Its a practical load. There is a difference between the two. You design around the theoretical load not the practical load.
 

MarcVenice

Moderator Emeritus <br>
Apr 2, 2007
5,664
0
0
Really jag ? I stress test my PC to be sure it's stable, because if it is stable under such a heavy load, it will be stable under lesser loads. That doesn't mean I expect my PC to run under such a load 24/7 though. If you do, then yeah you should adapt your expectation pattern, but if not, then you shouldn't. Don't really want to mix into this discussion though. I reckon if he can spend 1000$ on 2 videocards he can pay the extra 40$ or so that a pcp 750w costs over the 620hx.
 

John

Moderator Emeritus<br>Elite Member
Oct 9, 1999
33,944
4
81
Originally posted by: JAG87


taken straight from his post:

a GTX draws 160W
a QX6850 draws 130W

his system:

QX6850 overclocked to 3.3G = 130W + OC = 150W (and I'm being conservative)
GTX x 2 = 160W x 2 = 320W

150+320 = 470

470/620W = 76% load on the PSU

and thats WITHOUT anything else. There is still a motherboard, and all the other "assorted devices" as he likes to calls them. Yet he claims 70% load on the PSU. Yes, maybe when he is playing cs source or supreme commander thats true, but thats not the theoretical maximum load. Its a practical load. There is a difference between the two. You design around the theoretical load not the practical load.

Until you can show us your rig pulling 700W from the wall, like you claim it will, I think it's safe to say you do a lot of overestimating. :)
 

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
Originally posted by: John
Originally posted by: JAG87


taken straight from his post:

a GTX draws 160W
a QX6850 draws 130W

his system:

QX6850 overclocked to 3.3G = 130W + OC = 150W (and I'm being conservative)
GTX x 2 = 160W x 2 = 320W

150+320 = 470

470/620W = 76% load on the PSU

and thats WITHOUT anything else. There is still a motherboard, and all the other "assorted devices" as he likes to calls them. Yet he claims 70% load on the PSU. Yes, maybe when he is playing cs source or supreme commander thats true, but thats not the theoretical maximum load. Its a practical load. There is a difference between the two. You design around the theoretical load not the practical load.

Until you can show us your rig pulling 700W from the wall, like you claim it will, I think it's safe to say you do a lot of overestimating. :)


Dude, thats what you gotta do with electricity, ALWAYS. Just ask an electrician, when you wire a house, the outlet is 15A, the breaker is 15A, but the wire is 20A. Its called over dimensioning. Have you ever had one of those 12v AC adapters for a utensil blow on you or overheat? Yes, it happens all the time, and yet the manufacturer gives you that AC adapter. How could they be wrong? Well, lets say the utensil absorbs 500mA. A 600mA adapter costs $1.20, while a 1000mA one costs 1.50. multiply that by thousands of units, and you can see why they chose to include the 600mA. The chances that it will fail? Scarce. Maybe 1 in 1000 units. But it happens. And when it happens it can cost 10 bucks to replace the adapter, or the adapter can take your utensil to heaven with him. I wouldn't wish that on you.