PSU at 100% efficiency.

[Watashi]

I have heard reports of PSU's not being used to their maximum potential unless the maximum power a system is pulling is about half of what the PSU can supply. Example: a 750W system using a 1500W PSU. I was curious on if this provides better overclocking or not? Or if you guys have any sources of someone reviewing the topic that would be great. I'm curious if it is something that actually has appreciable gains

[DocSwag]

It's only that the peak of the PSU's efficiency curve is around half the rated wattage. It doesn't improve OCing, it just means your psu is a little more efficient.

[Watashi]

3 hours ago, DocSwag said:

It's only that the peak of the PSU's efficiency curve is around half the rated wattage. It doesn't improve OCing, it just means your psu is a little more efficient.

Could you more clearly define efficient please? 

[AresKrieger]

7 minutes ago, Doomerson said:

I have heard reports of PSU's not being used to their maximum potential unless the maximum power a system is pulling is about half of what the PSU can supply.

Well it depends on the PSU as they have different power efficiency curves, but generally the peak is around 50% as you said, however a system that at peak can only use 750W will not use that wattage the majority of the time it will use a lot closer to 50% or less so it is pointless to get such a big PSU unless you actually need it

 

Oh and it has no effect on OC, the quality of the PSU affects that not the wattage (so long as you have enough in the first place) @Doomerson edit notification

[mariushm]

The peak efficiency of a power supply is somewhere between 50% and 80% but it's completely unrelated to overclocking potential.

 

A power supply can be very efficient but can have lousy voltage regulation, which can affect overclocking by a very tiny amount.

Just the same, a power supply can be fairly inefficient yet could output very smooth ripple free voltages which components will like.

 

Low efficiency just means more heat will be produced in the power supply which has to be pushed out using fans, which means potential for more noise from the fan

 

Example of efficiency curve

 

 

This gold efficiency psu rated for max 650w reaches peak efficiency (92%+) somewhere after 300w but will still be very efficient (> 90%) up to more than its maximum rated power

 

This 600w bronze efficiency psu will reach peak efficiency at higher load percentage (still 300w, but percentage wise, it's closer to 600w) but the efficiency will fall much faster as you move away from the efficiency peak.

 

[79wjd]

5 minutes ago, Doomerson said:

Could you more clearly define efficient please? 

If your computer requires 100w to run, then a PSU with 80% efficiency (let's assume it's a flat 80% regardless of everything) will pull 100/.8=125w from the wall to supply 100w of power to the system. 

 

So, a higher efficiency percentage means that more of the electricity pulled from the wall goes to powering your computer and isn't wasted in the form of heat. So, a PSU with 90% efficiency would only need 111w~ to power your 100w system. 

[DocSwag]

Just now, Doomerson said:

Could you more clearly define efficient please? 

PSUs take AC from the wall and have to convert it to DC that powers your components.

 

However, it can't do this 100% efficiently. AKA not all of the AC power going into the psu actually makes it out as DC. Some of it is converted into heat instead. The ratio of the power coming out/the power going in is the psu's efficiency.

 

It just so happens that PSUs don't always have a constant efficiency, depending on how much power you're drawing from it. It also happens to be that a psu tends to reach max efficiency when the components are pulling about half the PSU's rated wattage.

As you can see above, PSUs tend to reach their max efficiency at around 50% load.

[laminutederire]

12 minutes ago, DocSwag said:

It's only that the peak of the PSU's efficiency curve is around half the rated wattage. It doesn't improve OCing, it just means your psu is a little more efficient.

Better to OC the PSU

 

[huilun02]

Assuming all other operating variable are the same and that the PSU is not at its max output, the PSU load has absolutely no effect on overclocking potential.

How did you get the notion that the load% affects overclocking?

[ARikozuM]

2 minutes ago, DocSwag said:

PSUs take AC from the wall and have to convert it to DC that powers your components.

 

However, it can't do this 100% efficiently. AKA not all of the AC power going into the psu actually makes it out as DC. Some of it is converted into heat instead. The ratio of the power coming out/the power going in is the psu's efficiency.

Seriously, Seasonic? Only 85% efficient at 40W? How am I going to power my Intel Atom with your 850W? Better to say that some is "lost" as heat since all of the pulled power will be converted to heat at some point.

[79wjd]

8 minutes ago, ARikozuM said:

Seriously, Seasonic? Only 85% efficient at 40W? How am I going to power my Intel Atom with your 850W?

RAID three Atoms in one case and run all three off the same PSU.

 

Or just don't acknowledge Atoms because they're crap.

[ARikozuM]

Just now, djdwosk97 said:

Throw three Atoms into the case and run all three off the same PSU.

Oh, sh#t! That'll bring it to 120W! Great idea! 

 

I'm kind of not kidding though. I've got an Asrock C2750 that only draws like 50W by itself. 

[79wjd]

5 minutes ago, ARikozuM said:

Oh, sh#t! That'll bring it to 120W! Great idea! 

 

I'm kind of not kidding though. I've got an Asrock C2750 that only draws like 50W by itself. 

Meh, I wouldn't worry too much about it, it'll die soon enough. 

[ARikozuM]

Just now, djdwosk97 said:

Meh, I wouldn't worry too much about it, it'll die soon enough. 

Killing me with those words.

[79wjd]

Just now, ARikozuM said:

Killing me with those words.

On the bright side, LPC Clock will have killed your Atom before it kills you

[ARikozuM]

Just now, djdwosk97 said:

On the bright side, LPC Clock will have killed your Atom before it kills you

I've heard of that. Hopefully, my Atom server continues monitoring my house for a little while longer.

[79wjd]

17 minutes ago, ARikozuM said:

I've heard of that. Hopefully, my Atom server continues monitoring my house for a little while longer.

It's made it this far....so hopefully it'll last. 

 

I'm just happy I decided not to get a C2750 myself. I was considering it, but I decided against it because at idle loads, total system power consumption isn't that much lower but top end performance would be on the possible future cases where I needed more peak performance. (Plus I also found a great deal on a 1230v2+board+ram, although I had made my decision against the c2750 at that point)

[ARikozuM]

9 minutes ago, djdwosk97 said:

It's made it this far....so hopefully it'll last. 

 

I'm just happy I decided not to get a C2750 myself. I was considering it, but I decided against it because at idle loads, total system power consumption isn't that much lower but top end performance would be on the possible future cases where I needed more peak performance. (Plus I also found a great deal on a 1230v2+board+ram, although I had made my decision against the c2750 at that point)

I'm trying to do the same, but server parts in my area are pretty scarce at the moment. Probably going to have to spend on the new Xeons. 

[79wjd]

5 minutes ago, ARikozuM said:

I'm trying to do the same, but server parts in my area are pretty scarce at the moment. Probably going to have to spend on the new Xeons. 

Check out the classifieds on ServeTheHome, I've seen good deals there occasionally -- I got my 1230v2+Supermicro board+32gb ecc for $350 (in December 2015) from there.

[System Error Message]

what efficiency means is how much power is lost from the wall as your PSU converts AC to the DC voltages you need.

Efficiency is affected by many things, temperature being one of the main things as well and component ratings. In another thread i argued about higher wattage PSUs for better efficiency stating that despite theory, practically higher wattage is more efficient at lower use because of thermal designs and better components

 

However a crap high wattage PSU is still crap even compared to a good lower wattage PSUs.

 

Despite part ratings, practical power use is very different. An atom despite being lower power is actually less efficient than a full CPU core, mainly because the longer you take to do a task the more power other parts of the system consumes. its not about performance per watt, the faster you complete a task the less power you will use even if using a CPU using 2x more power but finishing 2x faster not only uses the same amount of power but other components are used 2x less too. Intel atoms only make sense in situation where the CPU is never the bottleneck.

[79wjd]

5 minutes ago, System Error Message said:

Intel atoms only make sense in situation where the CPU is never the bottleneck.

If you're CPU bottlenecked you probably aren't even considering something as weak as an Atom. 

[Stefan Payne]

 

1 hour ago, mariushm said:

The peak efficiency of a power supply is somewhere between 50% and 80% but it's completely unrelated to overclocking potential.

No, it's worse than that.

Some PSU (AFAIR be quiet Pure Power 10, Cooler Master V-Series) have their peak efficiency way under 50%.

You can extend that range to at least 30%

 

1 hour ago, Doomerson said:

I have heard reports of PSU's not being used to their maximum potential unless the maximum power a system is pulling is about half of what the PSU can supply. Example: a 750W system using a 1500W PSU. I was curious on if this provides better overclocking or not? Or if you guys have any sources of someone reviewing the topic that would be great. I'm curious if it is something that actually has appreciable gains

What you've heard is an old fairy tail that's just wrong for at least 5 Years, more like 10.

And even then it wasn't really true, depends on the PSU in question.

 

WARNING, LINKS WILL BE IN GERMAN!

 

For example:
http://www.planet3dnow.de/artikel/hardware/netzteile/netzteil2006/netzteile_unter_50_euro/5.shtml

Peak at the beginning.

 

Or that Antec.

http://www.planet3dnow.de/artikel/hardware/netzteile/netzteil2006/mittelklasse/3.shtml

Peak at the End, ner 100% Load.

 

The next one (Antec TP3) has Peak at 80% Load as do most of the upcoming PSUs as well.

Like the CM IGreen, peak Efficiency at around 80-100% Load.

Like this Nesteq:

http://www.planet3dnow.de/artikel/hardware/netzteile/netzteil2006/mittelklasse/25.shtml

 

Buttom Line:
Most PSUs in that range have their peak Efficiency at around 80%; sometimes 100% not 50%.

 

That continues in the 700W and more range:

http://www.planet3dnow.de/artikel/hardware/netzteile/netzteil2006/highend/3.shtml

 

Even Enermax Galaxy 1000W:
http://www.planet3dnow.de/artikel/hardware/netzteile/netzteil2006/highend/11.shtml

 

And that was the state 11 Years ago.

Today it's worse and the other way around.

Peak efficiency is somewhere between 30-80% load, depending on the Design, Layout and so on.

You must NOT say that the peak efficiency is around 50%!

That's not true!

It is somewhere between 20% and full Load. 

[STRMfrmXMN]

3 hours ago, Doomerson said:

I have heard reports of PSU's not being used to their maximum potential unless the maximum power a system is pulling is about half of what the PSU can supply. Example: a 750W system using a 1500W PSU. I was curious on if this provides better overclocking or not? Or if you guys have any sources of someone reviewing the topic that would be great. I'm curious if it is something that actually has appreciable gains

We have a thread all about 80 PLUS efficiency pinned on this subforum. It's also in my signature. There's no real correlation these days to "A PSU is most efficient at X% load" because all PSUs are built differently. 

[Watashi]

13 hours ago, huilun02 said:

Assuming all other operating variable are the same and that the PSU is not at its max output, the PSU load has absolutely no effect on overclocking potential.

How did you get the notion that the load% affects overclocking?

No one gave me the notion, I've just seen individuals buying these high wattage power supplies, trying to match the efficiency curve. I assumed there must of been a reason, and now I know why. 

[Stefan Payne]

3 hours ago, Doomerson said:

No one gave me the notion, I've just seen individuals buying these high wattage power supplies, trying to match the efficiency curve. I assumed there must of been a reason, and now I know why. 

No, it's not.

They just believe something they heard somewhere from somewone who didn't really know what he was talking about...

 

A PSU can have an effect on OC, but that is not relevant for us 'air cooling' guys, only relevant for LN2 and max/suicide overclocking.

 

For air cooling it doesn't matter as long as you have a decent PSU.