Folding@Home and Energy Use

Freerepublic ^ | 5/7/2009 | Oshkalaboomboom

[Oshkalaboomboom]

Another site I am on is trying to start a small folding team. One "green" activist is trying to make the case that you would be better off sending money directly to cancer researchers rather than wasting energy by folding. His argument is that projects like folding and SETI have been wasting time and energy for years with nothing to show for it and all you are doing is putting money in the pockets of electric company executives. Are there any studies that show how much it costs to run folding 24/7 as opposed to just letting your pc sit idle 24/7? I can't believe that running your cpu at 100% is going to add significantly to the cost.

BTW, I finally broke the FR top 100 for folding, that was my goal when I started.

[chuck_the_tv_out]

the question would be, would your computer be on anyway. If not, then add up the cost of the extra hours, at perhaps 100W to 200W depending on computer, then look at the cost to you.

then find the cost of computing power if the cancer people bought it in bulk, and decide which is cheaper. it could go either way. it’s just that most people are more likely to leave their PC on than to actually give $10 a month to a charity

[John Valentine]

I confess my ignorance, but other than paper, clothes, bedsheets and napkins, what are you folding and why would electric company executives even take notice?

Now, “greens” want to get their stupid noses into everything we do on the principle that since they are enlightened, they can’t possibly be at fault for any problem and therefore it must be us, but “folding”? This seems like a stretch even for them.

[Straight Vermonter]

Congrats on your top 100. I’m languishing in the 170s. I wish I could get a second computer folding.

[1rudeboy]

Ask the greenie about carbon credits.

[Malsua]

The math should be fairly simple.

((Watts * Hours) / 1000) * electrical rate.

Lets suppose that you’ve got a hot gpu that uses 200 watts to fold. I have a GTX280 at home that draws roughly 200 watts while folding.

200 * 24 = 4800
4800 / 1000 = 4.8 kilowatts a day
4.8 * 11(cents per kwh) = 52.8 cents a day
52.8 * 365.25(day/year) = 19285.2 cents per year
Cost = $192 per year

Now, if you sent $192/year to cancer researchers, how much do you think would actually be spent on cancer research? 100%? Reality shows that while they say “x dollars goes towards research” you have understand that the guy delivering lunch is contributing towards cancer research by that metric. Much of each dollar goes towards administrative, overhead and capital costs. I would bet that actual money spent towards research falls somewhere down at about 25% of donated money.

Do you think that $50/year spent towards cancer research will ever provide meaningful results.

100% of what your donating via folding is actual cancer research, not going towards takeaway Chinese and administrative bonuses.

In my case, at one point, I had 50+ machines folding so the benefit was even greater! :)

[Egon]

Ping.

[nina0113]

Home cancer research in your computer’s downtime:

http://folding.stanford.edu/

I finally have a decent computer so will get this going.

[ikka]

How cold is it where you are? The heat generated by the system goes to warm your house, so if you need the heat anyways it is a wash.

[ken in texas]

The last time I measured power usage was back when I had 3 machines running the folding software. I kept them on 24 hours a day; when idle they drew about 286 watts, and used 373 watts when running folding. At my current electric rates, that works out to about $8.61 a month to run folding.

I’m down to one machine now, though — 47 watts idle, 60 watts when folding. Estimated monthly cost for folding: about $1.29.

[texas booster]

Modern greens are really watermelon Luddites, green on the outside and commie red on the inside, with a disdain for any technology that might advance our lives.

While SETI did not discover firm proof of intelligent life in the universe (including way too often on our own planet) it pioneered the entire discipline of distributed computing. Without the efforts of the SETI group, there would not have been a Genome@home project to help create the equipment to decode genes by the thousands as they do now. It has to start somewhere.

Malsua is correct, most of these systems will be left on anyway, or are doing such simple tasks that we can easily slip in processing to fold proteins. The power consumed is a trickle compared to the results.

http://folding.stanford.edu/English/Papers

In the case of Folding@home, Stanford University has released 63 peer reviewed papers to date, more than any other DC project in the world.

http://folding.stanford.edu/English/FAQ-Diseases

With all due respect, if it wasn't for projects like Folding@home I could never meaningfully participate in this scientific research. I don't have the math background, the medical background or the time to do more than an occational fundraiser. F@H allows me to contribute 24/7 to finding cures that will one day benefit ME personally, and my family.

http://folding.stanford.edu/English/Science

The sheer scale of protein folding is difficult to comprehend. Some proteins can fold a million times per second, but it takes a modern supercomputer a day to simulate one-billionth of a second of folding. There simply isn't enough computing power outside of distributed computing to achieve any detailed results in the field.

And what is wrong with allowing utilities, many of which are owned by the public, from selling electricity? The shareholders certainly care about power sales, and in most parts of the country folding power usage doesn't even effect the grid during the peak usage at 3 - 5 pm.

Now, if the greenie wants to prove how green he is by living totally off grid, then we can have a different discussion. Have fun taunting the little green fool.

[Hyzenthlay]

It also depends on whether or not you let your monitor go to sleep when you’re not using your computer - in my experience, the majority of power (determined by battery life in different situations) my computer uses while idle or even just running something that takes half the night goes to the screen. If the monitor/screen is ‘asleep’, which is possible because it’s not necessary for *you* to see it while your computer is folding, then it probably is more cost-effective.

Plus, you’ve got to consider that the donations would go to publicity, overhead, salaries (even in a nonprofit, in order for people to be full-time they do need a salary), and so on in addition to the research, whereas when you’re folding you’re contributing 100% to the research.

[Oshkalaboomboom]

Congrats on your top 100. I’m languishing in the 170s. I wish I could get a second computer folding.

Thank you. I'm 93 now and at a point where I move up 1 place per month if I'm lucky. But not long ago I was also in the 170's all it take is persistence and patience.

[texas booster]

http://www.freerepublic.com/focus/f-bloggers/1537549/posts#9

This is the first copy of our team scores from 2005. Nearly all of the systems were my personal systems or systems that friends let me maintain (and borg).

This snapshot shows 31 contributors on Dec 9, of which 20 were my systems. Passing a dozen people a week was common. And then when Klutz joined in earnest, the folks at DU howled in protest.

FWIW, SazanEyes produces more points every two days with his folding farm than the entire team did in over a year. How times have changed!

[David Park]

My electric bill has gone up about $60-100 per month since I started folding. It’s hard to get exact numbers because I also added a couple computers. I always left my PCs on 24/7 anyway, so this is just the extra cost of running them at 100% load. My bill will go up this summer (rate increases from 7.77 to 8.92 cents/kWh), whereas my computers helped heat my condo in the winter (seriously).

My primary folding box, with four 9800GX2 graphics cards and a quad-core AMD proc, pulls about 1000W from the wall at 100% load. This accounts for most of my extra electrical cost - using the formula above, about $56 to $64/month depending on the electric rate. The average computer doesn’t use anywhere near that much power.

From an environmental/conservation standpoint, I think it’s more valuable to have my computers doing something useful rather than sitting idle (wasting electricity) while waiting for me to access them. Distributed computing is also a good way to raise the profile of various research projects among computer users. Plus, man-caused global warming and “peak oil” are hoaxes, so I’ll use as much electricity as I want. ;-)

BTW, I’m currently participating in an annual folding contest with another team, but I’ll be back on team 36120 in a few days.

[texas booster]

I added an i7-940 box with a single ATI 4850 which is pumping out about 5,000 ppd. The first sets of console units will hit the points server tonight.

Are you seeing any slowdown in awarding points on GPU work units? With two GPUs I expected to be producing more than I am.

[David Park]

The recent GPU work units have gotten a lot bigger, and not only do they give less PPD than the older units, the PPD is more inconsistent because the units take longer to complete. For example, checking FahMon, the 511-point WUs (known to give low PPD) on my 8800GS cards only give about 2800 PPD, when I used to get over 4000 PPD. On the other hand, the new 1888-point WUs are good for around 9000 PPD on my 9800GX2, but each WU takes about 10 hours to complete. I believe the ATI cards run different WUs than the NVIDIA cards, so you can't really compare them, but they don't generate the same PPD as the NVIDIA cards. ATI keeps making improvements, so make sure you have the latest drivers.

The Core i7 chips are monsters. I've heard if you want to maximize PPD, you should set up several Linux virtual machines to run F@H, since the Linux SMP code is more optimized than the Windows code. I tried to get it working last week but kept running into problems with virtualization. You need to have a 64-bit OS. The instructions are here.

[David Park]

I just saw this post about how to lower CPU usage with ATI cards. You might want to experiment with it. More information is here. The ATI cards usually use a lot of CPU when folding, so you have to make sure if you're running F@H on the CPU as well, that the CPU is not robbing the GPU of cycles (or vice versa).

[texas booster]

Thanks. I'll check it out.

I installed an ATI 4850 and have seen a decent pop in points.