Slashdot Mirror
Ask Slashdot: Enterprise Bitcoin Mining For Go-Green Initiatives?
Supp0rtLinux writes "Bitcoins are currently trading around $75. I work for a very large organization. We have a fairly large HPC that is usually about 50% idle, as well as about 18K desktops on 4 campuses connected with dark fiber. All stay on 24x7 for after-hours AV scans (weekly) and backups (2-3x a week). All are leases that refresh every 2 years so all have fairly good CPU & RAM specs. As part of a go-green initiative a proposal has come up to use all the PCs for bitcoin in our own mining group; sort of like SETI-at-home style, but with a real dollar value return to us. Additionally, we would setup a queue in our HPC that dedicates 30% to BC mining when in use and up to 99.5% when no other jobs are running. The thought is that all the PCs are on 24x7 anyway and consuming resources so why not allow them to be useful 24x7 as well and generate bitcoins which can then be sold to offset the electrical costs of the running equipment and/or possibly even make a little profit. The guy with the idea says its a no-lose situation as if the price of bitcoins drops to below a certain level and is no longer a financially viable option, we simply stop the mining process. I'm curious what the Slashdot community thinks of this?
"
Read on for a few more details.
Supp0rtLinux continues, Is it viable? Would we generate enough revenue to cover our electrical costs even with CPUs running at 100% utilization all evening? Are there any security risks? Any thoughts on network impact? The consensus is that the proposal sounds good, but no one has enough info to make a knowledgeable decision either way. As a follow-up question and one that came up after the initial proposal, this entire idea has us wondering why the botnet/malware guys aren't doing this already? It would seem like a trivial task to take a botnet of hijacked PCs and have them do BC mining instead of spreading more malware and generate real revenue for the owner's of the botnets wouldn't it?
Why not ask this guy? Seemed to work out well for him.
http://www.smh.com.au/technology/technology-news/secret-money-abc-virtual-currency-racket-probe-20110623-1ggp6.html
On a serious note. If it were viable, the practice would be wide spread. The serious miners have moved onto purpose built hardware.
http://hardware.slashdot.org/story/12/12/05/2242233/race-to-mine-bitcoins-drives-enthusiasts-into-the-chip-making-business
Area51 - We are watching...
From what I have gathered, mining via CPUs is, at this point, not cost effective. It eats more power then it produces in bitcoins. The OP would probably get a better economic result by letting the computers go to sleep.
For a Go-Green initiative, maybe you should consider turning 18,000 computers off over night.
I bet your company ends up with a noticeably higher electricity bill, more so than you'd recover in bitcoins. I ran Seti@Home for a month on a single gaming grade system and my electricity bill jumped a staggering amount. But, I'd love to hear if I was wrong.
This might be financially feasible, but it is NOT green. Bitcoin mining *uses* more power, and does not produce anything "tangible", meaning something that offsets the use of energy elsewhere.
As far as I know, mining bitcoins is extremely CPU intensive. So unless the bitcoins will be spent on planting trees (or something...), the increase in the electric bill due to the increased CPU power dissipation (more electricity to feed the CPU and more electricity to cool it) makes things less green, not more.
But maybe a new definition of "green" is used here? Such as... "profit"?
"The agriculture ministry is not in charge of Gundam" - Japanese ministry official.
One, it's a great idea and even if you don't move forward with this plan I think it's important to give the employee kudos for suggesting something new. If everyone came up with a half crazy/half ingenious idea you'd be able to try out a lot of innovative stuff. That said, I think there's some underlying costs you should evaluate up front: there needs to be an administrator for the Bitcoin mining software, (I assume you know this but) your electrical bill should rise up a bit to reflect the increased power draw when your cluster is crunching numbers and there needs to be a way to record any incidences where you think the BTC software ran into the day or inhibited a normal work related job on the server. That shouldn't be a deal breaker, just have a system in place to assess those monetary incursions on your business. It may also require you to do some interesting tax claims as you might "earn" so many bitcoins that your accounting department has to start including it on an asset sheet so that the books stay legal. I'm not an expert in this.
Something that is important is that the sooner you get this up and running, the better. Read up on the distribution of bitcoins to understand what I mean.
Is it viable? Would we generate enough revenue to cover our electrical costs even with CPUs running at 100% utilization all evening?
Probably? The real problem is the volatility of BTC on trades and, if you become a major reserve of BTC, you would likely have difficulties realizing your USD valuations of BTC in one swift action. One day it could make sense and the next day it could be a bust all based on whether MtGox was hacked or some major holder cashed out. How exactly did you plan on evaluating these holdings?
Personally, I wouldn't do this. I'd take a more entrepreneurial approach and attempt to lease CPU time to new customers. I know this practice has grown less profitable as people have been better equipped to set up their own clusters but what you might do is look into software that rents out your cluster by the hour and then seek out customers. As far as I know you can only expect a couple cents an hour per core on something like that but it slowly adds up. Plus, it's a little more legit and on the level.
Another idea is to lease this downtime as CPU usage to a local university or high school or something and, whether they use it or not, you might be able to write off some of that donated time and save a little money on taxes or at least build good will.
As a follow-up question and one that came up after the initial proposal, this entire idea has us wondering why the botnet/malware guys aren't doing this already? It would seem like a trivial task to take a botnet of hijacked PCs and have them do BC mining instead of spreading more malware and generate real revenue for the owner's of the botnets wouldn't it?
I'm not an expert in this but I'm pretty sure the answer is simple: when a botnet is up, it takes commands instead of issuing information back to the command and control. The idea behind a botnet is more frequently to control millions of random desktops from one central point. If botnets phone home with a new wallet or information, you run the risk of being traced more easily, being detected more easily and also DDOSing yourself if your botnet gets out of hand. That's my suspicion anyway. I'm sure some do do this, they probably just can't get very large.
My work here is dung.
The one thing that jumps out at me immediately is that, for almost any modern computer, a computer which is powered up but otherwise idle is going to draw less power than a computer which is mining bitcoins. If you want to use GPU mining (and, realistically, if you don't, this is probably a waste of time since you'll probably never mine any bitcoins) then I think you'll find the power draw from a mining machine is substantially higher than that of an idle machine.
So, what you need to do is work out your idle power consumption (with, say, a Kill-A-Watt meter), then work our your fully-loaded power consumption (preferably by trying to mine some bitcoins, but you could get a reasonable quick estimate by firing up a modern 3D game), work out the delta, which is your effective power consumption for mining, and then use one of the various calculators out there (e.g. http://www.bitcoinx.com/profit/) to decide if this is a profitable venture for you or not.
Get a "Kill-O-Watt" for about $20. Test a PC at idle. Then load up the bitcoin processes, and test again. You will indeed find that the PC is drawing considerably more power under load than at idle. Multiply the increase in wattage by the number of hours in a month, then by the number of PCs you're talking about, and then divide by 1000 to get the increase in kilowatt hours. Multiply that by your cost of power per KWh.
Then go read how increasingly impossible it's becoming to mine coins, and how you'll have a very difficult time getting one.
Then don't do it.
You leave 18K systems on 24/7 so you can do a once-per-week virus scan and a twice per week backup? Really?
You mention SETI in the post, but why not pick some distibuted projects that are useful to your company to help push the research forwards? Most big companies want to find ways to improve their public image, this is a great way to do it.
There are many options: https://boinc.berkeley.edu/projects.php
If you don't already you should look into getting a special deal for power consumption at night. It should cost much less than powering on in the daytime. In fact, if you could have your system off in the daytime and only run it at night you could likely get really good deals in certain electric markets.
The proposal is still not green. Increasing profit is not the same as reducing energy. So it sounds like your accounting method has self-induced delusion on this issue.
We know where leadership by an anti-intellectual "strongman" who scapegoats minorities and likes boisterous rallies goes
You are better off powering off all the unused machines at night and saving your company quite a bit of money.
Anyone else notice that this question is being asked during a massive bubble period, and what is necessary to prop up a bubble is public awareness? Just like my post just now is shamefully unrelated to its parent, might the true purpose of this Ask Slashdot be to bring in a few more suckers before it crashes down to 15 or 20?
I will admit my biases against Bitcoin and my post history speaks to this, but this is a tactic also used to prop up stocks and precious metals. So yeah, I'm calling out TFS.
Charisma is the measure of someone's ability to lie with a straight face.
If this is your "Go Green" initiative, you either don't understand Bitcoin mining or "Go Green" means something different than most would assume.
Miners are moving to ASICs, which are an order of magnitude more efficient than the FPGAs they replace which were already an order of magnitude more efficient than GPU mining.
GPU mining was how things were done a year ago, but in terms of the speed at which Bitcoin is evolving "last year" means "back in ancient times".
I've done some testing. I tried to create conditions for success.
1. I had handy an M18 R2 - with dual 7970s.
2. I work a lot of hours and my kit is used by my employer. In trade off I use some juice for testing.
3. CPU bitcoining is worthless - just forget about it totally. Anyone telling you that only high power GPUs, FPGAs or ASICs offer a return are correct, and thats before ever discussing power.
4. I am vaguly able to generate 0.05 BTCs a day in test conditions. This currenty equates to 1/20th of a bitcoin per day, so 20 days would get me around $75.
Notes: At current rates, it will take 400 days to recover the cost of the laptop - assuming power was free, which it is not. Power draw is not far off a nasty P4 desktop box - but daya by day 24/7 I don't like the look of the power costs.
ASICs and FGPAs seem to be full of start ups, and small companies offering various now and future hardware and G/hs rates. These offer performance that will - as far as I can see eliminate GPUs in the near future, and make BTC mining the home of specialist miners only. Which is perhaps where it needs to go given the silly power required now on commodity hardware for minimalist return.
I can't really equate this being green, in any way, no matter how I cut it. I think the idea is to cut your power use, not get creative in trying to generate money of silly power usage.
I think previously, some crazy folks might have been able to make large btc mining operations on commodity hardware - I think its moved away from that now. At least where I live, and with energy costs what they are.
And the figures now are not what they were, when BTCs fell to 2$ - its fairly scary to see the loss basis if you make assumtions of hardware costs/ profit. For me, it has to be hobbyist mining, and with no direct aim 'to make' money.
If someone else buys your hardware for you; and if someone pays your power costs and colling / air con costs - then on the surface money could be made... but its a mirage really. Someone has to pay
We`re all equal
I'd recommend looking at the World Community Grid and BOINC. You can pick any number of projects to contribute resources to from solving clean water problems to finding a cure for AIDS to processing massive antennae data sets to detect asteroids that may be on a collision path with earth.
And just how many CPUs would you need to calculate 30 trillion SHA256 hashes per second?
Really ? I doubt they can lose much value at all, being a deflationary currency, save for total protocol/algorithm compromise.
Value of currency is in the eye of the users, and people are using it more and more (really ! the drug scare is just a tiny tip of the iceberg of all BTC transactions).
FWIW
I have nothing to lose but my bindings.
No, you see, the COMPANY saves money because they're not paying for some of that electricity. So they can claim that they're "green" and use "less electricity." When all that's really happening is that their employees are plugging the devices in and charging them at home (and paying for the electricity to do so). Then when they come to the office, the company gets a few hours of free electricity out of them!
It's really just a diabolical way to shift the costs of your energy consumption onto your employees, and claim you're "doing good" while doing it.
I was wondering about whether computing resources could be donated and you get a tax break. Found this related thread.
http://boinc.berkeley.edu/dev/forum_thread.php?id=7201
Key is as others have mentioned that your resources are most likely using minimum power when idle and increasing use will increase electrical fees, air conditioning fees, and might I suppose also wear them out sooner if it uses disk space.
Also imagined you might have enough scale to sell elastic computing services but there is a competitive market for that too.
Go Litecoin, or Namecoin, or something else that will become a competitor to Bitcoin.
In the end, it's all virtual currency anyways, even the US Dollar (which is a fiat currency, and thus essentially virtual.)
Add Timekoin in there as well. It takes a total opposite approach to digital currency that minimizes CPU usage while providing a lot better protection than Bitcoin in terms of double spending and wild spikes in prices.
it's not worth it. And as more ASIC"s hit the market, GPU's will become unviable to make back what you spend in electricity. As it is, it's getting close to that point with FPGA's already out last year using much less watt per mh/s.
You PC idle is prolly 200-300. And again only the AMD gpu's are good at BTC mining. Nvidia stink and most intels don't have openCL and stink even worse then nvidia.
You also have to think about heat. Running full load over night if the machines are near each other is going to create lots and lots of heat. Your AC system may kick on when it never did before over night. Causing even more electricity to be used.
You'll spend more in electricity then you make back.
Just put them to sleep for like 6 hrs each night and you'll save a ton.
Over half of all bitcoins that will exist, already exist. The remaining half will take about 140 years to be produced. The Chinese do not concern me. :)
Is "a billion" supposed to impress anyone with any amount of experience in cryptography? Would you use a cipher that could be defeated by a billion modern CPUs?
There is a reason researchers working on secure multiparty computation require that the attacker's work scale exponentially with the work done by honest parties. In Bitcoin, the attacker's work scales at most linearly with the work done by honest parties -- which is cryptographically worthless.
Again, your argument is what the German cryptographers thought about Enigma. It was a mistake then, and it is still a mistake today.
Palm trees and 8
When there's no way to exchange real money for BitCoins (inherent in that statement is my feeling about BitCoin
Just as a mental game, what makes you so sure other 'real' currencies are any more real outside of people's perception of their value?