Blogs How Much Does Your PC Cost To Run?

There is not really that much reason for working on efficiency for such low power consumption. Why, you might ask.

Well, just look at simple math. The minimum efficiency allowed by ATX spec is 65%. The current efficiency maximum is around 90% and most power supplies still deliver 80% at such low power consumption. Let's calculate the power consumption difference between these two values at 30, 50, 70 and 100W :
30 - 46W at worst, 37.5W on average, 33.3W at best.
50 - 76W at worst, 62.5W on average, 55.5W at best
70 - 107W at worst, 87.5W on average, 77.7W at best
100 - 153W at worst, 125W on average, 111.11W at best.

Now while the first values look horrible, the second values are more typical representation of what you buy when you are considering power supply efficiency at all (that means you don't buy the cheapest PSU, but some quality stuff). And even at 100W then we talk about 14W difference at best between 80% and 90% efficiency. If i calculate correctly. even if it is running 24/7/365 that difference equals to 122.64 kW used in whole year (365*24*0.014), which in my case equals to insane 9.28 euros per year (@0,0757€/kWh), which pretty much equals to one and half good meal in restaurant here in Slovakia. And that is with the PSU running all year - if we talk about generic PC which runs 8 hour per day then we are down at 3 euro difference per year. And as you can imagine, pretty much no one cares about that small difference, especially if it means buying a much more expensive PSU.

By the way, if you want such high efficiency PSU even at low power consumption, just get yourself a Seasonic X-400, which is 80Plus Gold even at 80W (the minimum tested load 20% according to requirements of 80Plus testing, from the total 400W).

 

the topic of discussion was really geared more toward total power draw of our systems. the draw of individual components doesnt matter since you have to pay to power all of it.

thats very impressive for that system. even at stock speeds my fiancees power draw was never that low with her 2500k.

again, you have to pay money for the inefficiencies, so for the sake of accuracy it should be included.

it depends on what the system is used for. my server hosts a few small services, is my NAS, and plays movies through the projector. about the most stressful thing it does is host a minecraft server on occasion, all well within the capabilities of the system.

i totally agree. my file server is my NAS and HTPC....it would be silly to run multiple systems.

so this whole discussion got me curious. i moved my meter back into the computer room and plugged my main box into it, and i was wrong when i said it was idling at 200 watts. its sitting at exactly 150 watts right now. the only thing i can think of was maybe i was thinking of idle before i re-enabled c-states, so i appologize for my inaccuracy. at any rate, my 4.0ghz OC idles at 1.8ghz with a 1.01v vcore. Just for fun i loaded the default settings in my BIOS and it idled down to 1.2ghz with a 0.88v vcore and was still pulling exactly 150 watts from the socket. overclocking with power saving features is great

your sarcasm is unnecessary. if we cant remain civil im fine with ending the discussion now.

i am aware of the effects of overclocking on power consumption. the goal of my two main rigs was not power efficiency, and i never said that it was. although if you see the above, it turns out it hasnt effected my idle power draw at all (loaded is a different story ).

and my calling my system modest was meant as a comparison to yakybs i7 server. but the whole point of comparing it to my i5-750 was to say that running an i7 server is going to cost considerably more than running a low power system. even if his total system draw is on-par with my i5-750, it would still cost almost $20 a month more to run vs the atom im running.

 

It wasn't meant to be taken nearly as harsh as you seem to have made it out to be.

As for the I7, it depends on which one. Sandy Bridge processors idle down within a couple watts of a typical I3 or Atom.

http://www.anandtech.com/show/4083/...core-i7-2600k-i5-2500k-core-i3-2100-tested/21

 

150W is still way too much. As i said, i owned the exactly same combo as you, the only difference i can think of is that i have just one hard drive. Still it was only 117W and not 150W.

 

Not quite Faugustin, PSUs have an efficiency power curve which doesn't always follow the 80Plus linear relationship. To make purchasing a PSU easier I'd like to see efficiency figures published for loads that are less than 20%. Especially as components are getting better in terms of energy performance and the fact the PC spends most of its time idling at less than 20% of power load.

 

according to anandtechs charts the i7 is still pulling 74 watts at idle. thats twice what an atom will pull.....heck of a lot better than a 1366 though. i wonder if thats with the on board GPU or a discreet card?

ive got 7 case fans that pull 3.75 watts each at full speed (no idea what they pull at idle speed), a water pump (the 5870 is under water) that pulls 6.5 watts, and the 2 extra hard drives. i suppose that could add up to 20-30 watts.

 

No one talks about linear, but the fact is that pretty much no 80Plus certified power supplies drop bellow 70-75% until you hit very low double digit power consumptions.

Don't believe me ? Right. Then believe the reviews :
Corsair AX850 : http://www.silentpcreview.com/article1073-page3.html - 850W PSU, at 22W (2.5%) load it is 65.5% efficient, at 43W (5%) you are already at 77.3% efficiency.
Antec TP-750 : http://www.silentpcreview.com/article1067-page4.html - 750W PSU, at 22W (3%) load it is 62.9% efficient, at 43W (6%) you are again already at 73.8% efficiency.
Seasonic X-400 : http://www.silentpcreview.com/article1062-page5.html - 400W PSU, at 22W (5.5%) you have 69.8% efficiency, at 43W (10.75%) you are at 80.4%.
Corsair X400, which is not even 80Plus : at 22W (5.5%) you are at 63.2%, at 43W (10.75%) you are at 73.2%.

Long story short, for better PSU you are looking at 75% or higher efficiency at 43W, for a bit worse ones at 70% or higher efficiency at 43W. At 22W, we are looking at power consumption from socket of difference between 31.51W vs 34,97W between the worst and best of above. That is 3.5W difference, which is about 2€ when you run the PC at that power consumption through whole year. At 43W, we are looking at difference between 53.48W vs 58,26W. That is 3.2€ difference through whole year.

And at this low power consumption, we are back at the point when you should ask yourself if you really need to care for something which makes a single digit euro difference in a year when running 24/7/365. You can care about efficiency when you start hitting 3 digit power consumption values, but up until that point, it is really pointless unless you need to save every cent.

And if you think for some reason i choose only recent power supplies, feel free to read more reviews at SPCR, even old power supplies like Corsair HX520 from 2006 had 67.7% efficiency at 43W or Enermax Liberty 500W from 2005 had 69.6% efficiency at 40W.

Long story short, we are talking about cents or few euros at best at double digit power consumption figures, when running it 24/7/365.

 

I knew about the efficiency curve for PSUs as this was started earlier. When you started to give calculated efficiency figures I knew they were more linear in nature. But now that you done a bit more research then perhaps you can see things at the same level as me, I hope. So now I refer you to my original statement at post #60 where I said the PC spends more of its time in idle below 100 watts and yes efficiency figure at single digits is just as important even if you're saving a few pounds a year.

On one of Tomshardware's articles they gave efficiency figures for PSUs where the loads were at 50watts and if my memory serves me right it the coolermaster m500 was 74% efficient.

 

The first test (Toms) was just a board with Intel graphics.
The second test (Anand) was a complete system with dedicated graphics.

The first shows an I3 is as efficient as an Atom, the second shows the I7 is as efficient as an I3.

Each drive can pull 15-25 watts, even an LP.
It wouldn't surprise me if your water pump pulls more than 6 watts. Especially if it's cheap or "leaks" electricity (more common than people realize).

 

The issue with that is he talks abou idle power usage. For example my desktop is idling at 94W, but can go up as high as 400W when CPU & GPU is fully loaded. Same for the fileserver - it idles at 65W, but at boot it goes as high as 150W and with high CPU load it can easily go over 200W.

Of course for some use cases <150W PSU is fine, but those use cases are rather limited.

 

I don't bother personally, if I need it, its got to stay on I guess, so it will!

 

No, I am not talking about idle use only, this is why I said if you can get your TOTAL below 150watts.

If your file server is peaking at 400, you either have a TON of hard disks (at 15-25 watts each, that equals a lot of drives!), or your file server is ridiculously over powered. It actually pays off in the long run to replace many small drives with fewer, larger ones.

Your desktop probably has a dual or quad core processor capable of 100-125 watts, plus a video system capable of 250-300 watts on it's own. Then you have a board with tons of features, lots of ram, a DVD drive, a few fans, maybe some lights. Then you power it with a power supply that needs to cover a wide range of power which adds inefficiency.


Here is a 100 watt (or less) PEAK server/desktop.
Atom, AMD 350, Sempron (35-45watt), FX (65watt), Old X2's (45-65watt).
LP drive (15 watts)
ITX board, single stick of ram and as few fans as possible (20 watts est.).
Use a USB stick or external DVD to install your OS and skip the DVD rom (18-24watts!).

Will it play Crisis? Nope, but I also don't expect it to. While a single core Sempron won't lightning fast, you can't tell me that an FX is going to be terrible for daily use. The LP drive isn't an SSD, but is fine for day to day browsing. You also don't need 50 sata connections or an Nvidia 580. One stick of ram is plenty and when was the last time you really used a DVD drive and couldn't get by with an external in those cases?

 

No, file server is peaking around 200W and yes, it is overpowered (i5-2500K, 10x2TB drives, 12GB RAM - see the setups in the signature). But idling at 65W, which is 95% of the time. So that is fine. The issue is that i need few virtual machines for various testing, so it can't be that less powerfull and on other side i can't those virtual machines run all the time on my desktop.

 

I too bought a similar energy monitor, after which I too wanted to know how much energy certain devices without sockets used. Its simple, use your mains meter, turn everything in your house off until it stops counting, then turn on the device you want to monitor, e.g. your oven, and see how many units it uses in 30min or an hour. Simple!

 

Was this article which inspired me to do a new personal build so soon & the are big cases a dying breed article come to think of it so I can have good performance rig & tiny acceptable performance rig in 1 big rig so I can use electric better & to say no big rigs aren't a dying breed only if you want them to be .

To be honest if I didn't use a PC so much I wouldn't need to but it's on nearly all the time which amounts to a fair bit of money wasted each year on bills.

Good thinking mate , I'll have to do the same for all the unmeasurables the bills at my house are high so doing all I can to slim them down.

 

It will take a very long time to pay for anything done just to save on your electric bill. 1watt running 24/7 is equal to roughly £1 a year. So if you spend £400 upgrading to save 20 watts, it will take 20 years to pay off and that's a best case scenario.

Remember, your system probably pulls under 200watts while doing things other than gaming, and if it goes into standby it likely drops under 20 (possibly ten). So the only you will lower your wattage significantly is while playing actual games.

If you think using an Atom instead will save you a bunch, look at idle numbers on the Core and I series compared to an Atom, it's not much different and once you need to use it, the Atom will run far longer to do what needs to be done before returning to idle, while the faster processors knock it out fast and get back to idle. Meanwhile your drives are still working and you are waiting with the monitor on. If Atoms were truly that much more efficient, batteries in netbooks would last a lot longer.

The only place you will make a significant difference that will actually pay off is the hard drives. If you have a bunch and can replace them for a single drive, you will recover it in a year or so. Almost everything else you will spend more than you save.

 

And the best part is - you can easily kill more than the power consumption of your computer by using a newer kitchen appliances, especialy the fridge, that one is a beast. Current fridges use around 200-300kWh/year (around 0.7kWh). Old ones easily use 500kWh/year or more.

Now when you think about that, and that a highend PC which consumes 400W (0.4kW) while gaming pretty much equals to the difference between the old and new fridge, and you would need to have that high load 24/7 to make up for the difference... I hope you got the idea - it is pretty much pointless to save energy on little devices like your PC or even on rarely used big power consumption devices like your vacuum cleaner. What you need to do is to replace your old refrigerator which has worse than A categtory power consumption, period.

 

What you say there is very true but I should have mentioned about the components I'm going for with my greeny rig for simple pc stuff, already have most components to do it except for mobo & optional sound card so not like I'm buying everything I'll need for it with already having most of it.

Asus E35M1-I DELUXE AMD Hudson M1 Integrated AMD Zacate 18W, 8GB, auzentech x-fi forte sound card, 4 5400rpm 2tb hdd, 1 ssd, 1 optical, 1 slow fan which will be by far the most used out of the 2 rigs overall.

I expect it to use a little less than 50w, my main system is 210w doing light stuff with all the HDD's & light overclock & 350w gaming so estimate to trim £140 off each year with using power better, got to have good sound, me no likey on-board audio & love Auzentech sound cards .

It would be very pointless if it was only 20w less but 160w less is £70 per year vs £210 per year & still be able to have plenty high performance PC sessions when I need the power rig , sure I won't feel the benefit immediately but after 2 years I would & it would begin as soon as I switch .

 

And i was thinking my two S1155 systems (desktop PC and a 11 hard drive system) each using ~100W in idle was using too much power. Your one idling system uses as much power as my desktop, server and HP microserver together .

 

You are of course quite right, although, those RaspberyPi boards might just kick that out the water. 2w draw decoding 1080p 1w as a file sharer and only £35 you could easily pay that back inside a year.