Which PSU is suitable for my system?

[ackrite26]

Currently have a 580w Hiper Type-R but in the middle of upgrading my entire system and would like your opinion on which PSU to buy.

The Brand does not really matter as I will definately be buying top of the range, what I do need help with is how much power I will actually need.
When fully upgraded it we be something like this-

X38 MB (currently P35)
Intel Penryn (currently C2D E6600)
2Gb DDR2
9800 G92 (currently Radeon X800XT PE)
36Gb WD Raptor
2* 150Gb WD Raptor
Creative X-Fi Fatal1ty
SATA DVD-RW
Multicard Reader
Fan Controller, 6 120mm fans

How much Wattage do you think it would take to run that?
Would 750w be sufficient or am I better off going for 850w??

[peterb]

Thyere is a power calculator somewhjere that can help (try seartching these boards or googling, but I would have thought that 750W is more than enough, and you could probably mange with less than that.

[Fraz]

I sorely doubt you're going to need a new one. People really don't realise how undertaxed their PSUs are most of the time. My system (see <---- ) never manages to draw more than 300 Watts at the wall socket (i.e. ~250 Watts actually used by my components), and you're going to be using maybe 50 Watts more than that tops, what with your extra disks and a couple more fans than me.

If you buy a 750W or 850W psu, it'll be running way below the efficiency of what you've got, as PSUs are normally most efficient at about half their max power output.

[peterb]

Originally Posted by Fraz

If you buy a 750W or 850W psu, it'll be running way below the efficiency of what you've got, as PSUs are normally most efficient at about half their max power output.

Generally most efficient at 75-80% of rated output. (although the increase in efficiency from 55% to 80% isn't a lot)

[Fraz]

Originally Posted by peterb

Generally most efficient at 75-80% of rated output. (although the increase in efficiency from 55% to 80% isn't a lot)

Well, I quote you my Corsair 520W PSU efficiency curve, which is pretty typical in shape, AFAIK.

[vicar]

No need to change, stick with what you have got.

[peterb]

Originally Posted by Fraz

Well, I quote you my Corsair 520W PSU efficiency curve, which is pretty typical in shape, AFAIK.

Interesting - thank you for posting that - what is interesting is that there is only about 3% difference across the loading (I did say it was only a small difference ) It would be equally interesting to see efficiency curves from other mfrs. (Or curves for differently rated PSUs) But we are going OT...!

[iranu]

Originally Posted by vicar

No need to change, stick with what you have got.

QFT.

[Pob255]

Most PSU efficiency graphs look like that, they often miss out under 30% load because the efficiency really drops off at low loads.
Also as you point out the efficiency differnce is normally within a 5% range (it's minimum efficiency in the 30%-100% load range that's generally important.)

However as afaik none of the following bits are currently available it's going to be rather hard to say if that PSU will be ok.
X38 MB
Intel Penryn
9800 G92

[Saracen]

Originally Posted by Fraz

Well, I quote you my Corsair 520W PSU efficiency curve, which is pretty typical in shape, AFAIK.

Actually, I would say that an efficiency curve that flat (from 30% to 100%) is far from typical. In fact, it's one of the reasons to buy Corsair (or the Seasonic S12 series on which they're based).

Most half-decent PSU's will have a curve that's broadly that shape, but not over that range. They're often much less efficient under lower loads, and that is why buying a PSU that is 'rated' too high may be worse than a lower-rated unit.

A good bet is to try to work out the idle load, and the maximum load, and look for a PSU that has a curve such that both idle and max load fall in the high efficiency bit. If you don't, then specifying a high capacity PSU may mean that you hit 80% while under max load ..... but what percentage of the PCs time is spent under max load? Even those running DC projects (folding, etc) may be pushing the CPU hard, but hard drives are typically used minimally while these DC projects are running.

Some games will push power draw to the max, and so will some benchmarks, but what percentage of a PCs use is under these conditions? It'll vary from user to user. I play the occasional game, but my PC (or PCs) are often switched on during the daytime, and usually one is still during the evening. Another user may leave the machine on 24/7, but not actually be there for most of that time. Another may only turn it on when needed, and sped most of that time playing power-hungry games. The power needs of these three types of user are all different.

ackrite26, I'm not familiar with that Hiper Type-R PSU, but I would suggest that any decent 580w PSU is going to be able to supply the power draw of all but the most highly specified and power-hungry systems. If you had a 6-disc RAID array in there, together with a couple of juice-guzzling SLI/Crossfire video boards and a cupboard-load of other hardware, then maybe you'd need more power than a good 500w will supply. But from a quick look (and PLEASE get a second opinion on this, because I'm not guaranteeing it) I'd be surprised if that system spec actually drew more than 300w or so.

Given that that is right, you need to look at what the efficiency of that Hiper PSU is like when the load is about 50%. A lot of PSUs will be 10% or more less efficient that that low a load.

However, there are PSUs (like that Corsair and the Seasonic S12s) that have an efficiency curve that is close to flat (for all practical purposes) between about 30% and max load. So a Corsair 520w operates at broadly the same efficiency from around 150W up to max. I would rather doubt that your system is drawing more than 150w while idle, and maybe not even that.

To my mind, there's a LOT of marketing kiddology going on in PSUs, with a tendency to get people to think that a big, beefy PSU must be better. It's not necessarily true, and may well be worse than a lower-rated one. Why? Because a lot of PSUs use rather optimistic power ratings, and there's naff-all regulation or standards to prevent them. Secondly, and most important, because if you buy a high-rated PSU whose efficiency curve is not as flat as that Corsair, it'll be less efficient at your actual load than a decent quality lower-rated unit.

And if you run at lower efficiency, two things will happen. In order to get the power your require out of the PSU, a low efficiency implies that higher power needs to be pumped in to get that power level out. The first inference from that is that your power bills will be higher than they need be. And the second is that the PSU will waste more, which produces more heat (because power going in either ends up as power out or heat loss) so your system will run hotter, more cooling will be needed, noise will go up, etc. Oh, and it'll probably cost more to buy in the first place.


My opinion is that chasing high PSU ratings without a specific requirement is not only unnecessary, but actually counter-productive.

There's one exception to that. If and only if you have a fairly flat efficiency curve, then about the only problem with a PSU rated much higher than your actual load is that it'll have cost a bit more to buy. The upside is that it gives you room to add a substantial amount of further load either by adding peripherals or upgrading components. This is because switching power supplies only draw the power they need, so given a 300w and a 1000w PSU, they'll both only draw 200w if that is the load placed on them (given that they are both of the same efficiency).

To my mind, that Corsair (and the Seasonic equivalents) are typical of a fairly good buy, because they seem to have a high build quality, flat efficiency curve and offer a degree of future-proofing. As for the Hiper, I don't know it, or how it performs. Is it as efficient over a wide range like the Corsair, or is it a manufacturer that indulges in .... erm .... creative ratings?

[peterb]

Originally Posted by Saracen

To my mind, there's a LOT of marketing kiddology going on in PSUs, with a tendency to get people to think that a big, beefy PSU must be better. It's not necessarily true, and may well be worse than a lower-rated one. Why? Because a lot of PSUs use rather optimistic power ratings, and there's naff-all regulation or standards to prevent them. Secondly, and most important, because if you buy a high-rated PSU whose efficiency curve is not as flat as that Corsair, it'll be less efficient at your actual load than a decent quality lower-rated unit.

There certainly is a lot of hype about PSUs - who really needs a !Kw psu (and that has been debated elsewhere on the forum)

Originally Posted by Saracen

And if you run at lower efficiency, two things will happen. In order to get the power your require out of the PSU, a low efficiency implies that higher power needs to be pumped in to get that power level out. The first inference from that is that your power bills will be higher than they need be. And the second is that the PSU will waste more, which produces more heat (because power going in either ends up as power out or heat loss) so your system will run hotter, more cooling will be needed, noise will go up, etc. Oh, and it'll probably cost more to buy in the first place.


My opinion is that chasing high PSU ratings without a specific requirement is not only unnecessary, but actually counter-productive.

Very true

Originally Posted by Saracen

There's one exception to that. If and only if you have a fairly flat efficiency curve, then about the only problem with a PSU rated much higher than your actual load is that it'll have cost a bit more to buy. The upside is that it gives you room to add a substantial amount of further load either by adding peripherals or upgrading components. This is because switching power supplies only draw the power they need, so given a 300w and a 1000w PSU, they'll both only draw 200w if that is the load placed on them (given that they are both of the same efficiency).

I would add that with a system with a lot of drives, simultaneous spin up could cause a power surge - however a good PSU should cope with that, and really if you have that many drives, you should be using a controller that allows for staggered spin up.

But to get back OT - the lower rated PSU should be more than adequate!