Is my PSU powerful enough?

[kalniel]

Nice to know its the other way. i.e. the 650w PSU draws about 800w
Next question... Why are cheap power supplies always said to have untrue figures? E.g. a cheap 800w PSU.
Is it just a misunderstanding and really its about the amps that might be low?

A variety of reasons really. Sometimes it's about not providing decent wattage on the rails that are actually useful, sometimes it's that they can only provide the headline figure as a peak, not continuous load, and sometimes it's just plain wrong and they shut down, break or otherwise are horrendously inefficient at loads approaching the headline figure that you wouldn't recommend them for anything close.

By the way I read a review on SLI 670's that mentioned a minimum recommendation of 750w, hence my view point. I am trying to find it now through my history.

The figures from nVidia themselves will be in those kind of ball-parks.. but that's precisely to account for the crap PSU symptoms mentioned above.

edit:
Yup, you found a real world example in your quote:

GeForce GTX 670 SLI provides near-linear performance upscaling over single GTX 670. With a power draw of just 28W in idle, and 250W average load, the GTX 670 SLI, for the first time, makes a dual-GPU flagship card look silly, in terms of performance/Watt. Even an efficient 600W PSU should handle the GTX 670 SLI just fine.

That's the 670, but I think the 760 is even more efficient, the review linked earlier put total system load at 480W for a system with SLI 760s. Toms have the 760 as 10W per GPU more efficient than the 670, at 150W each max.

[Agent]

Is it not prudent to plan for prevention of any issues? I was looking at worst case scenario, e.g. the PSU high usage 24/7, but to be fair I didn't really calculate the quality into the equation. Anyhow....continuing with worst case scenario assuming this is correct

No, it's not correct. Worse case scenario or not - it doesn't matter.

The only way you can "plan for prevention" is to use a unit that has decent components inside and keep it cool. There is no other magic.

80PLUS bronze is 81%ish on 230v efficient so 650w = 526.50w
Using the OP figures of 612w max then the prudent choice is a 750w PSU

This is not how power efficiency works.

I can't add anything to what's been said by Kalniel in that regard.

edit - sorry, post was hanging in browser for some reason. Didn't mean to repeat what's been said.

[Scainer]

By the way I read a review on SLI 670's that mentioned a minimum recommendation of 700w, hence my view point. I am trying to find it now through my history.

Just to contradict myself looking for the review I found this one...

GeForce GTX 670 SLI provides near-linear performance upscaling over single GTX 670. With a power draw of just 28W in idle, and 250W average load, the GTX 670 SLI, for the first time, makes a dual-GPU flagship card look silly, in terms of performance/Watt. Even an efficient 600W PSU should handle the GTX 670 SLI just fine.

Found it but to be honest there is no reasoning behind it as in the tests it drew 433w in full stress

Here is Guru3D's power supply recommendation:

GeForce GTX 670 - On your average system the card requires you to have a 500~550 Watt power supply unit.
GeForce GTX 670 2x SLI - On your average system the cards require you to have a 700 Watt power supply unit as minimum.
GeForce GTX 670 3x SLI - On your average system the cards require you to have a 850 Watt power supply unit as minimum.

670 SLI had 489w with a Intel Core i7 3960X

Just to summarise, it looks like a 500w PSU would be good enough, but why would they recommend a 700w one, or say a 600w one would be good?

[Willzzz]

Because they just mindlessly repeated official Nvidia recommendations.

[Agent]

Just to summarise, it looks like a 500w PSU would be good enough, but why would they recommend a 700w one, or say a 600w one would be good?

A mixture of two things:

1) Because people print things without understanding what they're talking about.
2) Because not all power units are equal.

Watt's on their own is a really bad measurement for a a PC power unit so people over spec, thinking this helps.

[Zak33]

hey guys:

/waves

I've just bought 2 rechargeable drills

Both are 18v. One last a whole day's DIY and will twist the world.

And the other lasts less than an hour and gets red hot and smells.... and is utterly crap.

In all things, quality of components counts.

None more so than in Power Supplies....

if my drill dies... I have to use another one OR plug one into the mains OR delay the job.

But what WONT happen is that my drill dying ALSO takes out my torch, bends my screwdrivers, creases and screwsup my tape measure, snaps the elastic on my knee pads and blunts my chisels.

However..if a cheap psu dies.. it MIGHT take out your motherboard, and/or your ram. and/or your graphics cards etc.

Buy a reputable brand, from a reputable supplier, spending the money you have NOT on the alleged power output but on quality.

Agent and Kal talk total sense from experience and knowledge. Heat kills electrical things, and a good PSU has a lot of heatsink, good fans and is stable for a good many years.

[Zak33]

oooooh hold on...

what I could do is answer the question posed in the opening thread and title!



yes.. your PSU is good enough for what you want to do. Stop fretting. It's a great PSU.

[guren]

You won't get anywhere near that level of level of degradation on a PSU. Period. If you do, it's faulty.
Under stress or not doesn't really matter that much - look at PSU efficiency curves. It's the MTBF you need to consider more than anything else.

As kalniel said - that PSU is more than enough for the job.

PSU Manufacturers (and by that, I actually mean the companies that re-brand CWT, FSP and so on...) have done an amazing job at convincing people you need an obscenely large PSU and need to replace it every few years.

A good quality PSU will last for years. The system I'm typing this on has a PSU that's got to be close to 9 years old now and there is simply no reason to change it. A bit of me dies inside when I hear people saying a PSU degrades by about 10% a year under normal use. It's not like there are journals and papers out there with the maths to work these things out....

edit - the other thing that people don't seem to understand is that capacitor ageing doesn't have that much of an impact on the PSUs total power output. This is another myth that has been started because of that god dam extreme power calculator which throws out an unbelievable amount of bad information.

Time and time again, I see people doing something like this:
600W - (2 years * 10% degradation) = 480W, and that you shouldn't expect any more than this from the unit.

This makes me irrationally angry, because it has zero basis from any research done on capacitor ageing.

What capacitor ageing does cause is higher ripple, sudden loss of power, unstable voltages and so on...none of which you want either.

The single, most important thing for keeping a capacitor healthy is temperature. Given that the inside of your PC isn't really that hostile, it's not a concern for the vast majority of people. More so given the huge fans fitted in a PSU these days.

I agree I'm still constantly using a seasonic PSU in my old PC which I started not use since got a new built.
Which I use on average 8 hours a day since 2006!

[guren]

Thx everyone for the ingight
I think I will go for the GTX760 since it is newer!
hopefully its should be better then GTX660 Ti, otherwise I will be
will start looking for a good cooled (as in cold) GTX760!
feel free to offer suggestion?
mainly looking at Asus, EVGA or Giga!
bad experience with MSI, so a big NO!

[Pob255]

Couple of quick things, in part that 10% degrading a year comes from older systems/cases back when you had the psu at the top, a rear 80mm fan and very little in the way of air vents in a case.
The psu actually was a major part of heat exhausting this meant you could be feeding 50-60c air into the psu, which was only 70% efficiency that's when you got accelerated capacitor ageing.

Most current cases have far better venting and fan options greatly reducing trapped heat, also a lower psu design case will separate the psu from the main case air and better efficiency means less heat generated within the psu.

Next you have 12v vs 5v loads, pc components have been moving away from 5v load to 12v load for many years but psu's where lagging behind on output levels, so that while a system could be powered by 300w a 300w psu wouldn't have the 12v output to cope so you needed to go higher.

Next you've got the whole "bigger numb3rs mean moar" attitude of most consumers, two psu's at the same price people tend to pick the one with the higher numbers.
This of course lead to all sorts of number massaging and outright lies, 80+ cert has been a big help in curbing this, although there are some examples of companies claiming 80+ certification when they are not. (if a psu is claimed to be 800w 80+ and is only £20 then it's more than likely not)

Most of the really cheap psu's cannot do their given wattage, standard tricks are "peak" performance, adding up individual rail loads for a total, ignoring atx ripple and voltage range during the "peak" and putting a number followed a w in the model numer (so it's a model number or description not an actual output figure)