AsicMiner's Immersion Cooling Mining Facility

https://bitcointalk.org/index.php?topic=346134.msg3709913#msg3709913

by mdelias • 13 years ago

199 190 13 years ago

mdelias• 13 years ago

The original title of this submission was "A look at the future of Bitcoin mining".

atwebb• 13 years ago

Which isn't really the title of the forum post, on HN post titles tend to be changed to whatever the article/original post's title is to discourage linkbait.

iDemonix• 13 years ago

I'd like to know how much a setup like that could produce and at what profit.

anonymoushn• 13 years ago

Estimates of this sort depend crucially on the rate at which other labs ship ASICs. A good guess is that the hashing power of the network will double monthly until whatever machines people have right now are barely worth running, but there's not a lot of certainty.

omegant• 13 years ago

It seems that this will happen during the first or second quarter of 2014 at this rate of difficulty increase...

redblacktree4• 13 years ago

It seems at that point that it would be down to your power cost per hash.

wmf9• 13 years ago

ASICMiner is a "publicly traded" company, so here are their financials: http://thegenesisblock.com/new-asicminer-financial-statement...

sscalia2• 13 years ago

Has anyone said... buttcoin yet?

GigabyteCoin4• 13 years ago

This isn't 2011. Not sure if you've been here recently but Bitcoin is fairly well respected among HNers today.

sscalia2• 13 years ago

Yes, as a currency for money laundering and buying drugs online - highly respected.

betterunix5• 13 years ago

Bitcoin is respected among some HNers. There are still a lot of people who question its value, its long-term prospects, who call the current price spike a bubble, who call it a toy/gimmick/scam/etc., and so forth.

nakedrobot2• 13 years ago

It makes me sad that all this computation can't simultaneously be used to fold proteins, or some other "externally productive" calculation.

EricDeb• 13 years ago

wouldn't that have been cool. If bitcoin mining could have been done in a way that had practical significance yet also resulted in coins?

termain• 13 years ago

So is there any problem to which ASIC miners can be repurposed?

nwh8• 13 years ago

No. They look for low hashes by incrementing an appended nonce. That's it. Nothing else.

ambiate• 13 years ago

Sadly, not any that I have found other than a few alternate coin systems.

elliottcarlson2• 13 years ago

ASIC - or Application Specific Integrated Circuit - is one purpose only - so unless the exact thing being accomplished is sha256(sha256(x)), then it can't be repurposed - this even limits its use as a password cracking machine unless the password is encrypted using that specific routine.

blhack• 13 years ago

There is an altcoin called "primeCoin" that addresses that. The minining is dedicated to finding ~~new prime numbers~~ mew prime number chains.

(I don't know why this is useful, as I am not a mathematician.)

cryptocoin• 13 years ago

Actually, no, it doesn't involve finding new primes. In fact it states clearly that it is not interested in finding large primes numbers (which is a requirement for finding new primes).

Primecoin's goal is to find chains of prime numbers, like the ones describe at https://en.wikipedia.org/wiki/Cunningham_chain

This is at least more useful than running a double sha on some input to produce some output that is hopefully less than a target t. But a altcoin that does something useful with its proof of work system is still not available.

swalsh• 13 years ago

Think about how much computing power is currently being used to "shoot people" whether it be virtual... or not.

puller• 13 years ago

So what, the existence of something worse doesn't make this better

NathanKP• 13 years ago

The way I see it Bitcoin is inspiring people to create these insanely powerful mining rigs. Once they've developed the technology then its possible to make a more general application to other fields, not with the exact same hardware of course but with similar setups.

nwh8• 13 years ago

There's nothing special about this stuff really. The processors they are using at the moment are literally just standard SHA256 core designs slapped on a wafer. The ASICmimer gear is ridiculously inefficient, drawing something like 11W per gigahash of mining.

Other designs like the anonymous Bitfury's are hand routed and do a lot better, around 0.8w per gigahash, but although there's skills in doing so, it can't really be applied to anything except for Bitcoin mining.

They're pretty basic on the scheme of things, they're just very costly to get to a fabricator in the short timeframes the community wants.

angersock2• 13 years ago

So, this cooling is basically a ghetto monkey-copy of techniques Cray pioneered two or three decades ago--right down to the use of 3M Fluorinert (http://en.wikipedia.org/wiki/Cray-2).

I don't think that this technology is going to be readily adaptable to anything other than brute-forcing hashes. I don't know of any fundamentally useful engineering or research being generated that would have application elsewhere.

EDIT: Which isn't to say it isn't neat...I just don't think it'll bear fruit anywhere else.

MichaelGG• 13 years ago

Has any new technology been developed that's widely applicable? It seems like straightforward financial pressure being applied to existing technology. What problems exist that'd benefit from extremely high density hash calculation? That is, computing that only needs 32-bytes of input?

peterhajas• 13 years ago

You're correct about the existing hardware not being useful. I don't think the bitcoin ASIC research is generally applicable to other fields.

This came up on the bitcoin Stack Exchange:

http://bitcoin.stackexchange.com/questions/10421/are-bitcoin...

TylerE• 13 years ago

There's really nothing new at all here technologically. ASICs are old hat.

Xdes2• 13 years ago

I don't think this can be practically done. Is there an algorithm for folding proteins? Where is the protein data going to come from?

stdbrouw• 13 years ago

I believe nakedrobot2 is referring to http://folding.stanford.edu/

GigabyteCoin4• 13 years ago

It makes me sad that so many food manufacturers out there create crap for our bodies. They could be synergizing the world's healthiest superfoods.

We all live in the same capitalistic world, for better or worse.

ams61103• 13 years ago

Your blame is misplaced. If people didn't like to eat those foods, they would not buy them and the food would not be produced.

nilkn2• 13 years ago

I'm a cynic, but I like to think of this as an unusually direct measurement of the cost we all have to pay for human selfishness. We simply can't trust each other, so we must expend all this effort so that we don't need trust in the first place.

mseebach2• 13 years ago

I don't have hard data, but I would suspect the effort going into Bitcoin mining is orders of magnitude less than the effort going into securing traditional monetary systems, even systems comparable in financial size to Bitcoin.

brfox• 13 years ago

Its better (for the environment) than mining for diamonds or gold.

seiji2• 13 years ago

What if they're powered by leaded gasoline generators?

jnbiche3• 13 years ago

The answer to this question can be found in the response of David Schwartz (professional cryptographer and long-time Bitcoiner) to this same question on StackOverflow: http://bitcoin.stackexchange.com/a/11651

The long answer is: you cannot secure the network using another any other work. The work done has to be specifically tied to the previous block on the blockchain (made of up transaction data, which is of course unrelated to proteins or anything else "useful" in the view of the questioner). Otherwise, you cannot demonstrably prove the integrity of the blockchain.

The short answer is: they are doing useful work -- they're securing an increasingly large and distributed financial network.

darkmighty• 13 years ago

There are alternatives to hashing proof-of-work. Two interesting ideas are:

- Peercoin (http://en.wikipedia.org/wiki/Peercoin), which introduced an alternative concept called proof-of-stake, which essentially dismisses mining in the long run

- Primecoin (http://en.wikipedia.org/wiki/Primecoin), for which the "work" is engineed to be scientifically useful (I'm not sure how useful it actually is)

I really wish the Bitcoin community had seeked more ingenious/efficient alternatives early on... a lot of energy is going to waste now.

yetanotherphd2• 13 years ago

This does, however, bring into question whether bitcoin can ever compete with the conventional financial system.

It seems that the cost of preventing counterfeiting, and other comparable attacks against the conventional financial system, is much lower than than the cost of the proof-of-work calculations done in bitcoin. In fact, by design the cost of these calculations must be significant, otherwise it would be cheap to hijack the bitcoin system.

I'm aware of some alternatives to proof-of-work, but until we see more of this in practice, I will continue to bet on fiat money as the future of money.

patrickmclaren• 13 years ago

Also related is David Schwartz' answer to this similar question "Is there a way to set up proof-of-work systems so it would be even more useful?"

David responds, "The work is not useless, it secures the transactions. The public hash chain ensures that Bitcoins can only be spent once... There is currently no known way to make the work more useful. " See http://bitcoin.stackexchange.com/a/334

aortega• 13 years ago

I'm sure way more computation power is used to determine your candy crush score.

simias4• 13 years ago

I think you're overestimating the power used to determine your candy crush score by orders of magnitude of orders of magnitude.

DanBC2• 13 years ago

The Bitcoin network is currently more powerful[1] than the entire top500 combined.

[1] At least, that's what they tell me. I suppose it depends what you define as power.

DanBC2• 13 years ago

simultaneously

Worse, the ASICs can't be used to fold proteins when they've finished being used for Bitcoin mining.

simias4• 13 years ago

That's a very good point. An other comment points out that the computation has to be tied to previous blocks to maintain integrity but it would be great if once the hardware is decommissioned (newer hardware came out/bitcoin crashed) it could be repurposed to do something else.

AlexCoventry• 13 years ago

I put some thought into how to do this. I think it would be hard to get right. One way, which is total vaporware and probably riddled with security flaws:

https://docs.google.com/file/d/0B3qaT-ZL6aeKOHNEQWdpZEtRYWc/... https://bitcointalk.org/index.php?topic=64421.5

csomar• 13 years ago

Sorry but this computation is being done to secure the network and secure a wealth estimated at around $10bn at today market prices.

I don't think it's unproductive at all.

betterunix5• 13 years ago

Are you suggesting that there is nothing this energy might be used for that would be worth more than $10bn? I can think of a few things that might be more valuable. What if this vast amount of electrical power were used to solve traffic flow problems for major cities, thus saving fuel and reducing pollution? Maybe we could find better TSP solutions and improve transportation. Maybe we could be creating more power-efficient chips. Maybe we could be curing deadly diseases.

Maybe instead of spending this energy on computers, we could be using it to build up the infrastructure of the developing world and help them become self-sufficient. I think that would create far more wealth for far more people than Bitcoin ever will.

dllthomas7• 13 years ago

It could lead to strong demand for a "print your own high-performance ASICs" product.

Sheepshow• 13 years ago

Good luck. Making hardware in general, and silicon specifically, is a clusterfuck compared to the software world HN readers likely expect.

Know how a good undergraduate CS degree involves writing hundreds of fully functioning programs? A great graduate EE degree might involve one single functioning piece of silicon.

Know how we bitch about compile time in the minutes or, god forbid, hours? Turnaround time on silicon is weeks at the minimum.

melling3• 13 years ago

Not to worry, the unintended consequence of the increasing need for large computational power will lead to faster and cheaper computers. The fact that someone can directly justify spending large amounts of money means that they will. We all win in the end.

JoachimSchipper• 13 years ago

No. This was (insignificant but) true in the CPU and GPU mining days, but ASICs dedicated to double hash collisions do not have any non-Bitcoin use, nor do these appear to push the state of the art in any way.

joosters• 13 years ago

Surely they can make use of the heat output at least?

runeks6• 13 years ago

There was a merchant once that sold dried fruit for Bitcoin. Fruit that was dried over his mining equipment. Not sure if it still exists though.

kurige• 13 years ago

While I'm not entirely sure how I feel about the "me too" altcoins, Gridcoin[0] was created for exactly this reason. Solving blocks contributes to grid computing for real world problems. Using ASICs would prevent them from switching, though.

[0] http://www.gridcoinnetwork.org/

mbreese7• 13 years ago

And this is why mining makes zero sense for any individual, at least as a profit making endeavor. When confronted with this scale of computational power, it will be impossible for "the little guy" to make any profit mining.

As a side effect, it will effectively centralize control of the currency to a few major computational groups that have the resources to make such big investments. If they ever wanted to cooperate, just a few of these groups could be able to determine policy for bitcoin as a whole.

applecore• 13 years ago

Mining pools were asked to switch to an earlier version of the Bitcoin software in March 2013 to roll back a blockchain fork[1].

[1] http://bitcoin.org/en/alert/2013-03-11-chain-fork

unavoidable• 13 years ago

I am curious - how do you think that these large groups will be able to set bitcoin policy? Technologically, they don't have much influence over the algorithms or the distribution of bitcoins as a whole. What meaningful policy decisions can be made?

wmf9• 13 years ago

In Bitcoin a block can be up to 1 MB, yet mining pools are only generating 150-250 KB blocks and leaving many transactions unconfirmed.

betterunix5• 13 years ago

They could selectively deny transactions. They might, for example, demand that you tie a real-world identity to your Bitcoin address before you can engage in any transactions, then deny transactions to people they dislike.

mrb6• 13 years ago

This is false. "Little guys", think the guy mining with 1 or 2 USB thumbdrive-sized miners, make more relative profit than large scale miners. Whether you spend $50 to mine at 3 gigahash/sec, or $50,000 to mine at 3,000 gigahash/sec, in theory the relative profit is the same, but the extra costs associated to the 1000x scale become non-negligible.

For the little guy, space, power, and cooling are effectively free. He makes use of space that was already available and unused (eg. his desk), cooling is easy (a few watts can be passively cooled), and power is so small (2.5W per port) that he effectively writes it off as insignificant.

But a large scale miner like friedcat has to build or rent a data center facility, has to pay for air conditioning units, has a electricity bill he cannot ignore as insignificant, has to pay people to deploy and maintain the dozen of racks, etc.

Source: I have been mining since 2010, at some point I was mining with 20 killowatt of GPUs (ten racks or so).

zhaodaxiong• 13 years ago

This seems incredibly similar to the claim that: all can make money, as long as you work hard! Is that really true? Think again...

jorgem2• 13 years ago

Hmmm. Regarding the "relative profit" of small scale mining: There must be some gain in efficiency to moving to these largescale mining systems. Or everyone would just make lots of the small scale systems, right?

jason_slack2• 13 years ago

So are you saying that someone could mine bitcoins from their day-to-day MacBook Pro and actually see a return on that?

Edit: I get that I would need a USB miner. I could pick up a 12 port USB HUB from Amazon and 12 USB Miners for a few hundred dollars of an investment.

fpgaminer2• 13 years ago

Whether you spend $50 to mine at 3 gigahash/sec, or $50,000 to mine at 3,000 gigahash/sec

You forget to take into account that cost per GH plotted against volume is non-linear. At $50,000 investment one won't see much difference and the graph will look linear. At $500,000, though, one can design and manufacturer their own ASIC and purchase everything in high volume. Then the cost plummets like a rock. And I would argue $500k is where large scale mining starts these days, not $50k.

mbreese7• 13 years ago

The only way to make money is to be able to mine more than your costs. A small USB ASIC miner is now around $30-40 for 330 Mhash/sec and the 2-3Ghash/sec USB models are ~$200 on ebay. You'd have to make more than that much money with your mining. At current exchange rates, you can make about 50-75 cents a week with a 330 Mhash/sec miner. That would require 40-60 weeks to pay back the costs. And that's not counting the electricity that you're using, which is small, but is probably more than 50 cents a week in some areas of the country. And that also doesn't then into account the expected drops in mining rewards.

At some point when you are scaling, your costs are lower than your expected income. You have to be doing this at scale in order to break even, let alone make money. And, if you have a sufficiently large setup, you aren't talking about mining in pools, you have your own pool, where you keep the rewards for each mined block.

Little guys mining get minuscule rewards for negligible costs. You aren't going to make any significant money with it.

Source: I've been running a 330Mhash/sec USB ASIC miner for the past two weeks :) I'm a total newbie, but the economics of it aren't good.

ihsw• 13 years ago

Absolute rubbish -- the only thing centrally controlled will be the generation of new BTC. That eventuality is expected and after all coins are mined then the reward will be transaction fees.

Up until then, you are correct -- the rate of new BTC generated will be dominated by the few.

mbreese7• 13 years ago

But in theory a majority cartel could control which transactions get added to the block chain. You'd be forced to follow their chain. And if you wanted your transaction processed, you'd have to follow their rules.

I'm not saying this will happen, but it's possible.

redblacktree4• 13 years ago

I mined 14 coins last year with a GPU I had anyway. My only regret is selling them immediately. At this moment, they'd be worth $11,704

seiji2• 13 years ago

Another title: Why the future of America is not bright.

The Chinese computer engineers can go out and assemble this thing from spare parts in their city.

Americans would have to order through five different middlemen and face language interop issues along the way.

scott_karana• 13 years ago

That was my takeaway as well.

China looks like a fantastic country to live if you're a talented engineer.

(Obviously, you might not make as much money, but what could you pioneer with the agility you'd have?)

beachstartup• 13 years ago

eh, no. you might get to work on some interesting stuff that's unique to china/asia, but quality of life is much higher in the US if you are a worker bee, aka a talented engineer.

ck22• 13 years ago

This is a very strange hardware race.

What are they going to do with all this hardware worldwide when the entire chain is finished in a couple years?

mrb6• 13 years ago

The chain will never be finished. Mining will (should) go on forever. This hardware will only become obsolete when Bitcoin's proof-of-work is changed from SHA-256(SHA-256(x)) to something else (which might happen in a few years/decades, but until then the hardware will have largely paid itself off).

evolve2k• 13 years ago

Oh it was my understanding that at a certain date (2015?) the system would stop issuing new bitcoins. How are miners paid then?

grecy• 13 years ago

This hardware will only become obsolete when Bitcoin's proof-of-work is changed from SHA-256(SHA-256(x)) to something else (which might happen in a few years/decades)

Once they announce whatever comes after SHA-256, would it not be possible to immediately start building ASIC's for whatever that is, and put them to work as soon as the new proof-of-work is used?

pilom• 13 years ago

Miners earn bitcoin from 2 sources when they find a block. One source is the award. This award is halved periodically so each block is worth fewer and fewer bitcoin. Theoretically, these awards will continue to decrease until the awards are extremely tiny and eventually (A REALLY LONG TIME FROM NOW) the last award will be given and there will be about 21 million bitcoins.

The second source is transaction fees. When you make a transaction (for instance when you buy a flight on Virgin Galactic) you have no guarantee that the rest of the world agrees that you made that transaction until someone finds a block which includes your transaction. You increase the likelihood of your transaction being approved by offering a transaction fee. This incentivizes miners to look for blocks which include your transaction.

Eventually, the awards will be very very small but there will be more transactions and thus more transaction fees available to miners. Thus miners must continue for as long as bitcoin transaction are being made.

jrockway• 13 years ago

What a waste. I can't wait until the bubble bursts.

thisiswrong• 13 years ago

The USD bubble ? The whole world is waiting eagerly for it to bust. At least it will put a stop to the US war machine.

fierycatnet• 13 years ago

The whole world is not waiting for the #1 economy in the world to tank with its currency.

ChuckMcM2• 13 years ago

While the bitcoin craze is on-going I've heard anecdotally of one retired GPU system being used to do CFD work, the reasoning was that people would pay $50 - $150 per hour for CFD analysis run time and that was more than the machine had been "earning" when it had been build for Bitcoin mining. That said, with an ASIC you're rather constrained in your choices.

But something in the same form factor as the mining ASIC might make this a pretty powerful engine for computing 'X' for some useful definition of X, post BitCoin exhaustion.

dmix• 13 years ago

a pretty powerful engine for computing 'X' for some useful definition of X

Password cracking-as-a-service for China's government/companies?

wlian• 13 years ago

Unfortunately Bitcoin ASICs are only good at double SHA-256 hashing; meaning it only does sha256(sha256(INPUT)). Probably not remotely useful of any modern password systems.

seabrookmx• 13 years ago

Maybe I'm blind, but did he mention what liquid they're using for immersion cooling?

Something like 3M Novec maybe?

alsocasey• 13 years ago

Something called 3M fluoinert.

VMG• 13 years ago

I love living in a science fiction cyberpunk novel

runako• 13 years ago

Are there analogous racks of equipment supporting the use (vs mining) of BitCoin for e.g. purchases? If so, where?

fiatmoney• 13 years ago

Why is the coolant boiling? Surely it must be more efficient to cool & retain the coolant than let it evaporate. Or are they using air bubbles to agitate or cool it?

ams61103• 13 years ago

Phase change (liquid to gas) removes a lot more heat.

It's not being evaporated into the atmosphere, it is undoubtedly being recondensed to liquid state and recirculated.

runeks6• 13 years ago

I believe you can see that in this picture: https://images.weserv.nl/?url=i41.tinypic.com/29nxc9x.jpg&fn...

The coolant is pouring off the copper pipes, which it presumably has condensed on.

sbierwagen2• 13 years ago

Huh, why immersion cooling? I thought the limiting factor in bitcoin mining was $/KWh, not floor space.

marshray2• 13 years ago

Presumably things are different in Hong Kong.

sbierwagen2• 13 years ago

Yeah, but how could electricity possibly be any cheaper there? What are the real estate prices like, then, $1000/m^2?

jimworm• 13 years ago

Someone else already noticed that the external pictures were easily identified. The location is no longer secret.

AsicMiner's Immersion Cooling Mining Facility

Related Stories