Power Consumption

kazak

Limp Gawd
Joined
Jan 31, 2008
Messages
141
OK now that summer is upon us and the need to conserve electricity, I'm a bit concerned by the amount of energy my boxen will eat up. Not to mention, it's getting hot my in home office. So, has anyone figured out how much a WU cost in watts to fold? How about one of those SMP WUs?

 
OK now that summer is upon us and the need to conserve electricity, I'm a bit concerned by the amount of energy my boxen will eat up. Not to mention, it's getting hot my in home office. So, has anyone figured out how much a WU cost in watts to fold? How about one of those SMP WUs?


I have no idea...at this point, I'm just full speed ahead! :D


 
Tiger and BillR will be the ones to come in with the info.

I do know a few good 80+ PSU will keep the heat and energy bill down.

 
I try not to concern myself with that. The wife handles the utilities, so as long as she doesn't start bitching all of a sudden, I just keep folding on. :D

 
I'd suggest to buy a Kill-a-watt if your interested in knowing exactly how much your using. They are like $15 and will tell you everything.

 
I have a Kill-a-watt. I like it. Aparently they don't read properly. Big discussions in the power supply forum about these meters.
 
OK now that summer is upon us and the need to conserve electricity, I'm a bit concerned by the amount of energy my boxen will eat up. Not to mention, it's getting hot my in home office. So, has anyone figured out how much a WU cost in watts to fold? How about one of those SMP WUs?

It all depends on your hardware and which protien you get to fold.

The best protiens to fold in terms of Points per Day per Watt are the SMP ones.
The GPU-PS3 are in the middle.
The worst is the sandard single core client.

The range for the standard client is only 0.5->2 PpDpW.
The range for the GPU is 2->5 PpDpW.
The range of the PS3 is 5->6 PpDpW.
The range for the SMP is 5->25 PpDpW.

Quads are better than Duals, as they give 100% more folding power for only 50% more power draw.
Low energy CPU's rule.
Get the most energy efficient PSU you can and aim for it to be only 2x bigger than the system power draw.
A simple Quad core system will only pull 150-180 watts max depending on overclock, so a good 300-400 watt PSU if ideal.

I limit the power draw of my farm to ~2 kW/h.
If I was running the standard client on it with a set of bad protiens, I would hope to get ~8,500 PpD.
If I was running the SMP client with good protiens on the same hardware, I would hope to get ~45,000 PpD.

Luck ......... :D
 
My average Quad box is taking about 175 watts.

Based on that the average Quad of mine is pulling in 4400 PPD

so 25.14 PpDpW that is my average for 3 Quads currently running.
-------------------------
On my P4 with 1950 AGP it average about 750 PPD and consumes about 275W.

So that is 2.73 PpDpW
-------------------------
On my PS3 Averages about 900 PPD and consumes 220W

So that is 4.09 PpDpW
-------------------------
My E6600 is good for 2350 PPD and consumes 160W

So that is 14.7 PpDpW
-------------------------
Total Average 17100 PPD and consume 1180W

So that is 14.5 PpDpW

REALLY wish F@H would give proper Points for GPU based work. The Wattage they take for the Points just is way out of wack... and were talking way more work being done than the points demonstrate, thus as Aldamon suggested I keep them folding for that sole reason!




 
My thinking when summertime is upon me is a little bit different.

I worry more about keeping my overclocks as high as possible while keeping temps in a safe range. Right now all six of my machines are in the living room. This happens to be the coldest room in the winter so it makes sense to have them going in there. It's helped keep my furnace from turning on for a long while now and has been nice as far as the power bill has been concerned.

During the summer months, it's a different story. My living room is by far the hottest room in the apartment. Unless I get an upstairs neighbor sometime soon, my a/c is going to be running like crazy most of the summer. If I have an upstairs neighbor, their a/c will run like crazy and mine will only turn on every once in a while. Either way, I'm going to have to move boxen around. I will finally move my son's computer (XP1700+) into his room which is cooler than the living room. I'm planning on getting a couple WCG machines running in my bedroom and moving at least one quad in there as well since it's the coolest room in the apartment during the summer. I might move a machine into my closet since it usually stays cool year round. However, with all the other junk I've been putting in there lately, I'm not sure how fast it will heat up. Just something I'll have to think about.

Basically, just position your machines around your residence to cut down on the heat in one area. Usually the basement is the best place to keep them. I wish I lived in my parents' old house. The basement always stayed pretty cool. There was also another "room" which used to be a cistern way back when and even in the middle of summer when it's three digits outside that room stays around 60-70F. It would be the perfect place for boxen after doing a bit of sealing up and a better door leading to the outside to keep the bugs and dust out.

 
It all depends on your hardware and which protien you get to fold.

Quads are better than Duals, as they give 100% more folding power for only 50% more power draw.
Low energy CPU's rule.
Get the most energy efficient PSU you can and aim for it to be only 2x bigger than the system power draw.
A simple Quad core system will only pull 150-180 watts max depending on overclock, so a good 300-400 watt PSU if ideal.


****** Disclaimer: If this is too obvious to answer , please disregard :D ******

Sir:..., (Tigerbiten) In the above abbreviated post you said a quad pulls about 150w to 180w. My question is, perhaps dumb, was that a reading at the wall for the whole boxen, a multi meter read on the board or an estimate of core draw?

The reason I ask is, at the wall, using a Killowatt meter, I'm getting about 160w a piece for each of 2x E6600's. Perhaps my WC'ing pumps and rad fans use extra watts ? Plus I have side fans blowing on the mobo and a "blow hole" in the top of both cases, they're all set at 5 volts. (yeah, I know I don't need all those fans, but I'm kinda' temp anal :p)

Thank you for any answer, I have respect for a great folder like you. :)


 
I can tell you my Wattage #'s come from my UPS which is a 900W/1500VA setup and I monitor each one under load. Those #'s tigerbitten has quoted are very accurate. My Quad's have peaked at 198W on the high side and 145 on the low side. So Average is right around 175ish...

 
I can tell you my Wattage #'s come from my UPS which is a 900W/1500VA setup and I monitor each one under load. Those #'s tigerbitten has quoted are very accurate. My Quad's have peaked at 198W on the high side and 145 on the low side. So Average is right around 175ish...


My appologies, maybe you misunderstood me? :confused: Having read several of Tigerbitten's posts there was no question in my mind that the numbers he gave were anything but accurate.

As I stated, I thought clearly, the reason I was asking I was getting on my Killowatt meter at the wall, after my Tripplite 850 PRO about 160 watts per E6600 (that's about 320w). So... perhaps my fans and pumps were drawing extra wattage :rolleyes:

Like I said before and I'll say it again: I was not under the impression that Tigerbitten's figures were anything less than accurate. FWIW I've read pretty much the same conclusion on other forums and associated articles on the internet.

 
I would say that between my E6600 and my Q6600 I'm seeing about 15-25W difference and your 160 is probably due to the water pump. That does take a bit of juice to run.

So I would say your guess is 100% accurate. Although My E6600 without a pump is taking 160W but i have it pretty well Oc'd.

 
I would say that between my E6600 and my Q6600 I'm seeing about 15-25W difference and your 160 is probably due to the water pump. That does take a bit of juice to run.

So I would say your guess is 100% accurate. Although My E6600 without a pump is taking 160W but i have it pretty well Oc'd.


Wow, thanks for the speedy reply Sunin. :D

By my sig you can see my E6600 boxen are only OC'ed to 3.0 GHz and 2.8 GHz. I'm uncomfortable running these boxen any harder than 3 GHz 24/7. From what I understood an OC'ed quad pulled a lot of watts, that's why I almost fell out of my chair when I read Tigerbitten's and your wattage figures. I understand how vcore and the other voltage adjustments can effect the total wattage draw. The info was much appreciated :)

Being "temp anal" I could probably loose some of my fans, obviously, I didn't go WC'ing for quietness and for the real "kicker" one of my cases is a Antec 900 with a huge "blow hole" fan :eek: . After HD tray mods, rad bracket mods, etc, I've been kind of leary about doing anything with the Antec 900 case. :( (IMO a great air cooled gaming case, but a pain in the rear end for WC'ing internally :rolleyes: )

 
Wow, thanks for the speedy reply Sunin. :D

By my sig you can see my E6600 boxen are only OC'ed to 3.0 GHz and 2.8 GHz. I'm uncomfortable running these boxen any harder than 3 GHz 24/7. From what I understood an OC'ed quad pulled a lot of watts, that's why I almost fell out of my chair when I read Tigerbitten's and your wattage figures. I understand how vcore and the other voltage adjustments can effect the total wattage draw. The info was much appreciated :)

Being "temp anal" I could probably loose some of my fans, obviously, I didn't go WC'ing for quietness and for the real "kicker" one of my cases is a Antec 900 with a huge "blow hole" fan :eek: . After HD tray mods, rad bracket mods, etc, I've been kind of leary about doing anything with the Antec 900 case. :( (IMO a great air cooled gaming case, but a pain in the rear end for WC'ing internally :rolleyes: )


My highest Quad at 3.6 which took a bit of voltage, I think 1.4v to the cpu is taking 198W, which for the 40watt increase over my E6600 is producing 2x the work... your never going to be able to get that sort of watt to performance difference anywhere else thus why it is strongly recommended to use Q6600's if you can afford them.

Also for temp's get a TRUE HSF that puppy right there will eliminate 99% of peoples concerns for temps on the Q6600 with Two 78CFM fans = hard to break 55c at 3.6Ghz When it gets 80F in my condo I hit up to 59, but that is extremely warm in my condo obviously! :)

 
My dual CPU boards tend to pull around 1.5x the CPU draw.
My X5355's pull 220 watts and the whole system pulls 330 watts.
My dual K7's also had a CPU draw of ~220 watts and a total system draw of ~330 watts.

My single CPU boards tend to pull around 2x the CPU draw.
My E6600's pull 65 watts and the whole systems pull 130 watts.

All figures are from a Kill-A-Watt meter.
I've yet to put a Kill-a-Watt meter on my new Q6600 systems but I expect them to be somewhere around the 1.5-2x the CPU draw of 95 watts.
So the 150-180 watts is a guessamate.

But things like a high end vid card will add another ~40 watts idle, ~100 watts load to your figures.

I'm another one who thinks the GPU client is very under valued in terms of PpDpW vs the SMP client.

Luck ............ :D
 
Wow! I didn't expect this kind of response. Thanks guys.

Well my goal really is to keep folding just as long as I can afford to pay the bills this summer and stay cool. The heat is probably the main concern for me. I typically keep my thermostat at a high setting (80F) during the summer and have my ceiling fans going. Forgot how much power it draws but it's one of those energy efficient types. Anyway, I guess I could always get another fan to circulate air in/out of my home office. I might just get one of those Kill-a-Watts just to see how much my systems are eating up.


 
Look around you can get a kill-a-watt for under $15 bucks.

 
Just tested the 3 new Q6600's systems I've got set up so far.
Running at 400x8 for 3.2 Ghz, core voltage 1x 1.350v, 2x 1.375v.
Gigabyte G33M-DS2R mobo with 2x2 Gb memory.
Running 2 copies of Orthos Gromacs stress test to load core to 100%.
Power useage recorded by a Kill-a-Watt meter is 149 watts, 157 watts, 159 watts.
The CPU uses 10 watts more going from 1.350v to 1.375v.

8 systems at 160watts + 1@330watts + 1@370watts = 1980watts.
Looks like I'll just get under my 2 kW/h limit.

Luck ............ :D
 
My highest Quad at 3.6 which took a bit of voltage, I think 1.4v to the cpu is taking 198W, which for the 40watt increase over my E6600 is producing 2x the work... your never going to be able to get that sort of watt to performance difference anywhere else thus why it is strongly recommended to use Q6600's if you can afford them.

Also for temp's get a TRUE HSF that puppy right there will eliminate 99% of peoples concerns for temps on the Q6600 with Two 78CFM fans = hard to break 55c at 3.6Ghz When it gets 80F in my condo I hit up to 59, but that is extremely warm in my condo obviously! :)


Damn, you are only using 1.4 volt, while I had to bump it to 1.46xx in the BIOD and the vdroop takes it to about 1.40v and that's for the 3.4Ghz, I'll bump it back to 3.6Ghz soon.

I agree that True kicks ass, at 3.4ghz mine nearly even breaks 50c, I would suppose the true is the same for 3.6Ghz (I've done it before, it hovers around 52c each cores)

Go C2Q whenever possible.
 
My 2 E2160's @ 2.7 (300x9) are using between 100-110 watts each. They are diskless, using Gigabyte P945 board, 500w generic PS's 1gb ram (1gbx1)
 
My 2 E2160's @ 2.7 (300x9) are using between 100-110 watts each. They are diskless, using Gigabyte P945 board, 500w generic PS's 1gb ram (1gbx1)

The only numbers that have any meaning at all to your wallet are total KWH used by all your machines.

The first step is plugging a kill a watt into the wall and everything else into that. The actual relationship between a reference motherboard and several different CPUs is actually pretty minor. A big bad video card will certainly make a difference. Hard drives draw so little these days they are all but un-noticeable.

Throw a few big 120mm fans in the system at .5 amps each, now you have a difference.

Next look at the specs on your Power Supply. A decent 550 to 600 watt PSU tells you right up front it’s going to draw 4 to 10 amps from the wall. In this regard the Kill a Watt is usually a bit off.

On the other hand the Kill a Watt does very well with KWH or kilowatt hours and those kilowatt hours are how you are billed by the electric company. KWH-X-Your Rate=The dollars you pay per month.

The so-called 80 Plus PSU’s because like many specs each company takes those measurements in their own factory under their control. Do some checking and you will find there are now at least 4 ratings on these PSUs, these new ratings coming from a third party of course which backs what I just said, right now 80 Plus means little except you probably bought a slightly more efficient PSU. The 80 Plus doesn’t mean you are consuming 80% or the difference 20% less power, right now it means slightly less power, maybe, on Thursday, if the moon is right.

The closest I can come to a parallel on power consumption is a basic quad with basic video, 2 gigs of JDEC spec RAM and the usual array of cooling fans draws about the equivalent of 4 100 watt light bulbs running 24/7/365.

My main machine with an 8800 GTS 4 gigs of RAM a Q6600 at 3.4 with a vcore of 1.41 including my 24” LCD monitor, and router works out to about 5.5 100 watt bulbs, 5 with monitor off.

I only used the 100 watt bulb comparison because if you go to your power company’s power consumption page their typical expected use in dollars for a home computer seems to date back to the 286/8 days.

Now I’ll throw another item into the ring. Your own electric meter. A design done in the 1940’s or earlier it reads by inductance and as such is non linear. Long and short of that is, IT’S USUALLY WRONG. Worse, there isn’t a damn thing you can do about it. If you pay the $25.00 fee (Here in PA) to have it replaced you are stuck with one of two options. IF it reads higher, you are stuck with it. IF it reads lower the Power Company reserves the right to replace it.

I have pages of rants on power companies all backed by fact but in the end we are all being screwed there just as hard as were are at the gas pumps.

Be thankful your computers don’t run on diesel fuel;):rolleyes:


 
Right now supermediastore
has a 2-fer on kill-a-watts, one fancy, one plain. Well worth it.


Holy crap I picked one up for 20 bucks... wtf?:confused:

http://www.amazon.com/P3-Internatio...4?ie=UTF8&s=electronics&qid=1207004393&sr=8-4

It actually appears to be teh P4460 version.. /shrug I don't get it but hey this one did everythign I wanted and will run you about 10 bucks cheaper. Whats the diff? Ahh I see it calcs the actual cost for you.. ok so well yoru choice... most the info you need will be on the one I linked...

 
As an Amazon Associate, HardForum may earn from qualifying purchases.
I've two of the older type, each was about $16. The new one is a little fancier, one of the "neat" things is that it retains prior measurement, which the old one did not.
For someone who thinks a case just gets in the way of cooling, I don't need the fancy features.
 
I think mine was half price when I got it.

The way I checked my figures.
I worked out what my farm should be useing by measureing the power draw of each box with the Kill-a-Watt meter and adding them together.
Then before I went to bed I took a meter reading, when I got up I took another one.
The difference devided by the time was almost the same as the figure I worked out with the Kill-a-Watt meter.
The only other thing running all night was the fridge.

Luck ......... :D
 
Back
Top