Page 1 of 1

Computing power vs cost

Posted: Tue Feb 09, 2016 2:28 pm
by Naze
So I ran my computer overnight for the first time folding proteins. I started thinking about the cost of running my PC for extended periods of time(electric bill) vs just donating money. So I figured I'd ask here:

Let just say speaking hypothetically it costs me an extra 40 dollars a month to run my PC every night folding proteins. I will use last night as a guideline to how many proteins I can fold, I earned ~12,300 points and ~9 WU's(#8 is about 80% done).

Would you guys rather just have 40 dollars a month from me or would you rather me earn the 369,000 points / 270 WU's? Since I am unsure of how much my electric bill will increase because of this... At what point, money wise, would you rather have the donation over the processing power? $30? $40? $50? $100? How much value do you place on 270 WU's, or even a single WU for that matter?

I ask for a couple reasons. If the money would further the research better then it makes no sense for me to continuing to donate this way. Also if it is kind of close either way it might be worth saving the wear and tear on my pretty graphics card / CPU getting maxed out for 9 hours at a time.

Any admin response is much appreciated! Thanks!

Re: Computing power vs cost

Posted: Tue Feb 09, 2016 4:20 pm
by foldy
I think if your PC already has much GPU power it is more value to FAH to let this GPU fold.
e.g. fast GPUs fold like 200k to 500k PPD (thousand points per day).
Your PC seems little weak with only maybe 30k PPD.
If we cannot configure FAH to get out more PPD of your PC then donating just the money maybe a good option?

Re: Computing power vs cost

Posted: Tue Feb 09, 2016 5:09 pm
by wilding2004
foldy wrote: Your PC seems little weak with only maybe 30k PPD.
Or possibly he's only getting base points? I don't know. I think the science done is more valuable than the money. If you donate small amounts of money, somebody has to then spend time collecting the hardware to build the systems to do the science.

Since he's already got the hardware he can get straight to the "do science" bit. Now if somebody was to donate some serious money, that might change the equation somewhat.

Re: Computing power vs cost

Posted: Tue Feb 09, 2016 5:44 pm
by Nathan_P
It would also be useful to know what machine he has folding - we may then be able to give some idea of monthly running costs.

Re: Computing power vs cost

Posted: Tue Feb 09, 2016 6:06 pm
by 7im
Also consider that donating money is tax deductible in the US (and maybe elsewhere). Some employers will even match your donation to a non-profit which doubles your donation.

Stanford welcomes all donations. Some people prefer to donate in small portions using hardware they already have. Others purchase more hardware just to run FAH. Some people answer questions in the forum. Or do beta testing on new fah software. Some fold and donate money.

I don't think they have a preference of cash vs. work units or any of that. It would cost Stanford about the same amount of money either way to run those work units. But considering the current penchant of the US Congress for not funding a lot of research, the cash may actually be more helpful at times; upgrading servers, more storage drives, etc.

Let each person give according to their ability.

Re: Computing power vs cost

Posted: Tue Feb 09, 2016 11:22 pm
by Naze
Just an FYI for people asking. My GPU is a GTX 980ti, and my CPU is an Intel i7 4790k.

Re: Computing power vs cost

Posted: Wed Feb 10, 2016 1:14 am
by Ricky
Naze,

If you have your key entered, you will get early return bonus. After 10 work units, you should see more than 500k PPD. The card you have is one of the most efficient for folding.

Re: Computing power vs cost

Posted: Wed Feb 10, 2016 3:22 am
by bruce
There are two important considerations.

First, there's equipment costs.
If you already own the equipment and you give money to FAH so they can buy some percentage of a more powerful machine, it's obviously going to be more efficient to use equipment that you bought so you could get your email/browse the web/whatever and also use id for folding. If you're spending money specifically to buy hardware for FAH, then it doesn't work out that way.

Second, there's the power costs..
If you're considering donating money to Stanford to do the processing that you might be doing, it's a more complicated calculation. First, the cost of business power is rarely lower than the cost of the same amount of power for your home -- but that tends to be a very complicated calculation. Second, buying power is taxable, whereas donating money is not. I don't know enough about those two power cost alternatives to make an intelligent recommendation.

Re: Computing power vs cost

Posted: Wed Feb 10, 2016 5:07 pm
by foldy
@Naze: If you are unsure if your FAH client is configured for max performance on your GTX 980ti then you can post the fah logfile and we can check it.