• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Processor power consumption when SETI@home is running

E

ElDonAntonio

Senior member
Hi guys!

just a question:
What's the power consumption difference between a processor running idle or one at 100% load? Say a 1Ghz Pentium III...

I'm asking because our university is really short of money (to the point where they have to close at night because heating is too expensive...it's VERY annoying when you have a hell of a digital circuit design to complete before 8:30 tomorrow!!!), and the sysadmin is running SETI@home on every machine during the holidays....I'd like to do some quick math and figure out how much money does this cost the university, so please be honest guys!!! (don't give me lower numbers to support SETI@home, I'm already in your camp!!! 🙂 )

Thanks, and we're leading in the race!!!! 😀
DonAntonio
 


<< Hi guys, I was just wondering, do you think a chick would get turned on if she learned you're a SETI@home cruncher with a high WU count? 🙂

Seriously, now that I got your attention, I have a question...
What's the power consumption difference between a processor running idle or one at 100% load? Say a 1Ghz Pentium III...

I'm asking because our university is really short of money (to the point where they have to close at night because heating is too expensive...it's VERY annoying when you have a hell of a digital circuit design to complete before 8:30 tomorrow!!!), and the sysadmin is running SETI@home on every machine during the holidays....I'd like to do some quick math and figure out how much money does this cost the university, so please be honest guys!!! (don't give me lower numbers to support SETI@home, I'm already in your camp!!! 🙂 )

Thanks, and we're leading in the race!!!! 😀
DonAntonio >>



No change in power consumption, since the systems are running already. If there was a difference it would be very small. The component that uses the most power is the monitor. To save money, I would recommend shutting down the monitors when not in actual use.

Edited: to finish 🙂 anykey.wav lol

The only other increase in power use would be the use of the harddrive, when writing the completed WU. The swapfile shouldn't be a factor, provided there is at least 64 megs of ram. (possibly 32megs would be enough?) Would someone help me out on that one?

Crunch ON!
Force
 
I'm already in your camp!!!

I don't see your name on our team roster... :Q But I sure would like to see it there!!!
 
Hi Wiz,

I'm in fact in the UD research against Cancer team. I only have a laptop and a not too powerful desktop, and as much as I want Anandtech to win the race, I prefer investing my computers towards a cure for cancer.

GunDog, I believe the power consumption goes up when the CPU is at 100% load, because on my laptop (which I don't believe has Speed Step but I could be wrong, it's a Celeron 800Mhz), the CPU fan stays on when Seti@home is running and shuts off when it's not. That indicates that the processor dissipates more heat when it's working, and thus consumes more power, right?

Thanks!
Antonio
 
I believe a P3 1GHz consumes around 16.1W normally. I seem to remember that a study done for Microsoft showed that the coppermine core P3's would consume 50% more power under a full load so you are look at about 8W more per PC. That translates into .008 Kilowatts-hours of Electricity used per machine. That means to create a difference of 1 kWh you would need to run 125 machines. You can do a little research in your area about the cost of electricity per kWh, but I believe the government uses a standard of 6.5 cents per kWh.

If you had it running as a service from 6pm to 6am (12 Hours) 125 machines would cost an additional 78 cents.

Now the additional cost is under the assumption that the machines would have been running at night anyway and that your sysadmin would have the monitors turned off. You can probably throw some more in there for the hard drives, but I doubt it would be a huge amount.

Hope this helps and if I am wrong I am sure someone will correct me. 😀
 
No offence Crazee, but that 16.1 seems very low. I was unaware that P3's used that little amount of power. Athlons are usually in the 50-60 W range.

Let me to some research and I will get some exact Numbers with some links to back me up 🙂
 
This also is putting heat into the bldg. and saves you heating $ during the heating season.
Someone had actually hooked up a meter to their computer and the amount of additional electricity was very little.
 
Actually you are correct I was thinking of the mobile P3 1GHz not the coppermine. It consumes 33W from what I recall which would make the additional power consumption 16.5W. So that means that the difference in 60 computers roughly would equal the 1kWh.

So 60 running overnight 12 hours would be about 78cents using the assumption of 6.5 cents per kWh cost.
 
Here is a libnk to one of AnandTech's articles: 1.0 P3

It says that the P3 should be puling less than 50 watts. Assuming that the machines are Turned off at night, that would be 50 watts per machine that could be saved. Assuming that wyou have 100 computers that would be 100 x 50 = 5000 watts that is 5 kilowatts. Assuming the same 6.5 cents per KW/H you would be looking at a Potential cost savings of 32 cents an hour. with 100 computers turned off instead of running SETI

Assuming the computers will not be turned off, you can take 2/3's off of that, as most machines will pull 2/3's of max while at idle, as an idle CPU is still running at 100%, it is just doing the "idle process" you would be looking at about 15 cents per hour if you have 100 computers, and you uninstall SETI. That is for a total of 100 computers, not per computer 🙂 For a university, that is a grand total of nothing. putting water saving showerheads in one dorm shower would save more on a yearly basis 🙂
 
I don't know if running SETI@HOME will up power consumption. However, while my laptop runs on battery with SETI,
the battery is drained in less than 30mins compare to 2 - 3hr with normal use. I would say the harddrive
and CPU fan is the major cause of battery drain.
 
ElDonAntonio,


No matter how you look at it, if he is running it on 500 machines as long as he is shutting the monitors off, they aren't spending more than $20 a night to run Seti even versus turning the machines off at night.

Now you need to talk to the sysadmin about joining the TeAm 😉
 
These chicks sure look turned on!!! 🙂

Crazee, yeah, I guess that would make it worth it 😉 I just discovered our university already has a team, but it's far in the rankings...and I don't think the administration is even aware that the sysadmins are running SETI@home. I had to complain last session because they were running all the time (even during school term) and compilation of sotware on these 1Ghz machines was SLOOOWWWWW....
 
Seti will nto slow down the machines at all if it is set up correctly. SETI only uses the un-used CPU cycles. If the computer is getting heavy use, then it will basicly shut off seti. I use SETI on my web servers, e-mail servers and game servers with no adverse effects. IBM has it running on their web servers as well ( well, according to my father anyway, I never actually looked 🙂 )

The speed slowdown is probably all in your head 🙂 DO you have any way of timing the complication on the same machine of the same program with SETI on and off? You may be suprised 🙂
 
EvadMan, I did some qualitative testing on lab computers (ran the same compilation on two identical machines, one running seti@home and the other not), and the one without seti@home finished in about half the time (2 pentiumIII 1Ghz running Linux/KDE)

Maybe it is not properly configured, I don't know. I know for sure though that it never shut off. I also doubt that Seti@home would not influence at all performance, if only for cache hits that would go way down for your running application.
 
The biggest performance difference I have observed in really strenuous tasks is about 5-10% with S@H running, mainly due to cache hits as you stated.

I find it extremely hard to believe that S@H could cause ANYTHING to take twice as long. That's a HUGE performance gap. I question your emperical techniques that caused a 1Ghz P3 to seem as slow as a 500Mhz P3.
 
That is a huge difference. Either it is installed incorrectly, or something else is running in the backround of the slower computer. I could see a 5-10% diff, but not a 50% that is huge! if it was 50% I would not be running it on my buisness machines. Something fishy is going on there.
 
If some Admin did a mass install of SETI as a service everywhere, where the batch file had the service installed and set for a "normal" priority (inadvertantly or not) rather than "low", then it is possible (depending on the OS) that some impact would be felt when doing stuff like compiling.
 
I can see three explanations for what I experienced:

1) The explanation above (although I don't remember the nice values set by the sysadmin)

2) someone else was logged on my machine and running tasks. I am pretty sure I checked though and nobody was logged on.

3) The performance hit from cache misses becomes pretty important when compiling a program. This is possible as I guess during normal compilation, the librairies and object files are already in the cache and speed up considerably the subsequent compiling and linking. If my memory is correct, the program I was compiling took with seti@home about 5 minutes and without around 3 minutes.

I'll do some more research on the topic.
 



<< GunDog, I believe the power consumption goes up when the CPU is at 100% load, because on my laptop (which I don't believe has Speed Step but I could be wrong, it's a Celeron 800Mhz), the CPU fan stays on when Seti@home is running and shuts off when it's not. That indicates that the processor dissipates more heat when it's working, and thus consumes more power, right?

Thanks!
Antonio >>



A laptop would be different, but you didn't mention laptops 🙂

A desktop the cpu fan runs all the time.

Force
 
Gundog, I know the CPU fan of a desktop runs all the time, I am not arguing that the fan is drawing more power, but that the fact the fan of my laptop turns on is a sign that the CPU is drawing more power when SETI@home is running.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top