Very informative article on power consumption.

Fair play on testing what goes where on the rails and what's pulling what current/wattage... but they miss the point entirely in terms of efficiency.

When I get an 85% efficient PSU I'm bothered about how much power it pulls, not what it delivers.

Certainly interesting to a degree, but I think they play it up a bit.
 
Fair play on testing what goes where on the rails and what's pulling what current/wattage... but they miss the point entirely in terms of efficiency.

When I get an 85% efficient PSU I'm bothered about how much power it pulls, not what it delivers.

Certainly interesting to a degree, but I think they play it up a bit.

To find efficiency numbers which vary wildly between different psu's its easier to just head to a review of that specific PSU.
The point of this article is to help to choose what power rating to go for methinks... and they have demonstrated fairly well that nothing over 750W is ever needed for a top-end crossfire/sli system.
 
To find efficiency numbers which vary wildly between different psu's its easier to just head to a review of that specific PSU.
The point of this article is to help to choose what power rating to go for methinks... and they have demonstrated fairly well that nothing over 750W is ever needed for a top-end crossfire/sli system.

Hardly groundbreaking stuff though - a lot of the more sensible users on this forum have been saying the same thing... although you'll note that even though the last rig only JUST skims 500W, it pulls a lot of current making it unsuitable for a good number of 500W PSUs (how many supply 40A on the 12v rail(s)?) . Something they fail to mention entirely and which WILL cause problems if you don't have a PSU which can supply enough current.

Don't get me wrong, I think it's a very useful article - I just think they need to consider their well collected data for longer before giving any kind of conclusion, especially when it proves very little and misses out a lot of the important points they do hit on.
 
Good article but as said they missed a few tricks which could have made it the definitive article:

1. As Audigex says, just cause 500W is the max used doesn't mean a 500W or even 600W psu would be able to handle the ampage on the 12v rails.

2. No mention of the fact that all psus pretty much have their highest efficiency between 50 -70% load which means a 800W+ psu would be cheaper to run than a 500/600W psu on that last gaming rig.

3. Okay, they at least picked highish end systems but people with rigs like that tend to have more optical, hard drives, fans, water cooling loop(s) etc. I know with mine I have 2 optical drives, 4 hard drives, water pump, 7 120mm fans, fan controller etc. This would probably take that final rig to a max of 600W+

4. No mention that in decent psu reviews that as loads increase the voltages on the rails drop and the stability and ripple of the voltages deteriate as well which affects the whole stability of the system especially if overclocking. In fact if you look at extreme overclocking forums, the super high end overclocks have come from people who have modded their psu's to give 12.5v at idle and hence still 12.2v ish under load whereas most psus start above 12v and drop below 12v under high loads.

5. Although I am not convinced yet, it has been widely reported that a psu will lose up to 10% of its power per year. I want my psu to last 3 years minimum, ideally 4 - 5 years and hence cope not just with my current cpu and gpu but future ones over the next 5 years without having to buy another psu.

6. And finally for those of us after quiet systems, the fan in a psu is at it's quietest up to 50-70% load (varies from psu to psu)


Therefore, although I agree with their results, I slightly disagree with their conclusion when you take the above into account.

A 750W psu would indeed be "sufficient" for their gaming rig but as the "minimum" psu you should go for to avoid all the problems I have listed above.
 
Last edited:
Interesting article. Power consumption has been a real focus of mine lately and having seen my own PC's consumption it's clear we've all been mislead into buying rediculously powerful PSU's when for the average PC 550w would easily be enough. The only time this really changes is when SLI/CrossFire configurations come into play.

Even stacks of hard disks don't make that much difference. Today's models only use about 10w each.
 
A long and interesting read but the more I read about PSUs, whats needed etc the more confusing it gets due to so many conflicting stories and info. :(
 
Interesting article. Power consumption has been a real focus of mine lately and having seen my own PC's consumption it's clear we've all been mislead into buying rediculously powerful PSU's when for the average PC 550w would easily be enough. The only time this really changes is when SLI/CrossFire configurations come into play.

Think you're spot on there mate. I was VERY surprised when I bought a wall power meter from a shop to test what my PSU draws, given that its about 4 years old and rated at 350W.

On idle my rig draws ~100W, and on gaming ~170W - maxed stress testing can pull the draw up to ~220W.

According to all the online calculators etc I should barely even be able to run this rig in my sig off my PSU, let alone max it out and still laugh it off!

*Smug* < Never has bought excessive PSU :cool:
 
Good article but as said they missed a few tricks which could have made it the definitive article:

<snip>

5. Although I am not convinced yet, it has been widely reported that a psu will lose up to 10% of its power per year. I want my psu to last 3 years minimum, ideally 4 - 5 years and hence cope not just with my current cpu and gpu but future ones over the next 5 years without having to buy another psu.

10% would be the most extreme example, like a very cheap PSU run 24/7 @ load.

More likely would be 5% per year for a cheap PSU used 24/7 mostly idle or heavy load 8h per day.

A PSU using decent quality parts you'd expect a range of 1-3% loss. And most quality PSUs can go 5-10% above their quoted spec. So you'd expect them to run to quoted max spec for at least 3 years. Given you shouldn't build a PC that uses over 80-85% of the PSUs output, aging should only be an issue for cheap budget PSUs or running the PSU at 100% load.
 
Back
Top Bottom