"Today, all of the machines dedicated to mining Bitcoin have a computing power about 4,500 times the capacity of the United States government’s mightiest supercomputer, the IBM Sequoia, according to calculations done by Michael B. Taylor, a professor at the University of California, San Diego."
And UCSD is home of the San Diego Supercomputing Center (SDSC), a major site for National Science Foundation computing research. So rather than 'obviously wrong' I'd weigh this assessment as 'presumptively right' unless/until other better evidence or testimony is presented.
I don't know if it's wrong, but it is a meaningless comparison since bitcoin mining is done with special purpose hardware that could never, for example, run a linpack benchmark.
I imagine they are comparing raw operations per second (in some measure) but the comparison is totally meaningless.
I'm not sure of this specific calculation but in general they come from the network's total hashrate compared with the known floating point operations per second of the top 500 supercomputers. Not sure how flops has been equated to hashrates though.
Nowadays it makes less sense since bitcoin hardware is ASIC/custom silicon designed & optimized for hashing as opposed to floating point operations making the comparison biased.
The numbers are: 9 petahashes/second (entire Bitcoin network [FTA]), and 20 petaflops/second (IBM Sequoia [1]). If you equate a SHA-2 hash with a floating point op, the ratio is 0.45, not 4,500.
This site (http://www.bitcoinwatch.com/) seems to do a 'flops' type conversion from the hashrate on the bitcoin network.
Currently at 108239.28 Petaflops which is 5400x the performance of the IBM Sequoia. This is approximately the quoted amount of 4500. While I don't agree with the idea that a hashrate can be converted into flops because its more bitwise opts others are drawing similar calculations to UCSD.
A Radeon 7990 gets around 1.2 gigahash/second and 8.2 teraflops.
The ASICS are of course skewed towards hashes from there, and I guess the supercomputer could be more skewed towards floating point than the GPU, but a large factor seems more likely than a fraction.
You are making a terrible apples vs. oranges comparison.
A bitcoin hash is about 1900 32-bit integer operations [1]. This is comparable to 1900 32-bit floating point operations, not one(!). Because most CPU/GPU, when they can execute N integer ops per cycle, can also execute N floating ops per cycle.
So the ratio clearly is in the ~1000 order of magnitude.
"A bitcoin hash is about 1900 32-bit integer operations [1]. This is comparable to 1900 32-bit floating point operations, not one(!)."
That's the amount of work you do in a software implementation of SHA-2, running on a CPU or GPU. This abstracts the hardware complexity inside the chip, which is very relevant when we're going into ASIC's. SHA-2 is bit shifts, masks, and binary adders: it's much simpler than a floating point core.
A hardware comparison would be something like ops/(transistor * time). A recent Bitcoin press release [1] claims 500 billion SHA-2 hashes/second out of >1 billion transistors: ~500 hashes/(transistor-second). A 7990 has a peak throughput of 8.2 teraflops SP [2], out of 8.6 billion transistors: ~1,000 flops/(transistor-second). Both are on 28nm processes.
Back in the GPU mining days people would say that since their GPU could do either 300 MH/s or 2 TFLOPS that every MH/s is equivalent to x GFLOPS. Then you multiply that by the total hashrate and get a ridiculous number and compare it to the Top500. As others have pointed out, this thinking does not apply to ASICs at all but journalists cannot resist it.
I used to hold the opinion that all proof-of-work systems were inherently flawed, because in order to protect against an attacker, you have to waste more resources than the attacker would have gained.
Now I think that in practice, the amount that the attacker could gain might be quite small compared with the total value of the system.
It's not that the attacker does less work than the honest parties. Rather, it is that the honest parties must work at least as hard as the attacker. If the attacker devotes a gigawatt, the honest parties must devote a gigawatt to defense.
What I like to say is this: Would you use TLS if that was how it worked?
TLS depends on a central certificate authority to vouch for the identity of the site you're visiting. If you don't have a centralized authority, a web-of-trust model is the best you can do.
I'm not saying proof-of-work is appropriate in all cases.
To use a more extreme value, a vault containing gold that was protected by proof of work would be useless, since the work required to open it would have to equal the value of the contents in order to discourage thieves.
My point was that in some cases, like bitcoin, it might be possible that the value to the attacker of a successful attack on bitcoin was significant, but still much less than the total worth of bitcoin.
Here's the difference: banks have every incentive to try to reduce their power consumption, Bitcoin miners have every incentive to increase their power consumption.
You are pretending that miners somehow do not double their income when they double the number of mining rigs they operate. A bitcoin miner makes more money by devoting more electrical power to Bitcoin, regardless of the hashes-per-second-per-watt ratio.
Increasing hashing only increases the proportion you get from subsidy + fees. It's a zero-sum game. If the other guy increases his hashing rate, you will lose out. Eventually, at some point, energy cost dominates.
OK, let's suppose that everyone is already at the point where consuming more power is not economical. Then a new technology that would allow everyone to use only 1/10th as much power for the same hash rate comes along. Power consumption will not decrease in that scenario; instead, it will remain the same, and the hash rate will increase.
My original point was that when a technology comes along that allows banks to be more energy efficient, the incentives are all in place for banks to adopt that technology and reduce their energy consumption. There is never an incentive for a Bitcoin miner to reduce their power consumption.
Not true. Since using more electricity costs more money, bitcoin miners have an incentive to use less electricity. Their incentives are to maximize hashes/s/watt.
Given that energy efficiency is mostly a function of the underlying silicon, miners don't have much control over it (especially once everyone has converged on 28 nm).
This is true, but the reward is currently far larger than the electrical consumption, so mining is currently mostly a contest to bring more hardware online.
(kicking CPUs and GPUs out of the contest probably did a lot to lower power consumption, but I wonder if the growth of the network has overshadowed that)
The reward for mining is currently 25 BTC/block, plus transaction fees. That's pretty high at today's conversion rates, but it will be cut in half continually over time. If miners spend more money in power and cooling than the reward, they're losing money. Even with deflation, the amount of money they can spend on power will go down.
Whenever I see advances in mining rigs I just hope that someone invents a cryptocurrency where the mining actually does some useful work, like protein folding.
Spot on, this is exactly what I've been thinking for a while. Like the SETI@Home screensaver -- processing astronomy data, or number crunching for public institutions.
Not possible- Having the calculation perform some alternate activity would destroy the utility of the calculation for protecting bitcoins.
The best you could do is create a cryptocurrency that requires similar hardware to the hardware used for other tasks, so that this hardware is commoditized. This would help other fields indirectly.
The core of the proof-of-work concept is that the work is easier to verify than to do the work again. If a way can be found to do that with arbitrary computation, it would be a very big deal.
The fascinating thing about this idea is that you'd get an actual commodity money. Solving useful problems like protein folding is something people pay for, it has intrinsic value. This would be utterly unlikely Bitcoin.
Mining does some useful work: it keeps the network secure. So it's not entirely useless. I agree that the proof-of-work could be built on a more useful algorithm rather then simply finding hashes smaller than a certain value.
Remember though that the proof-of-work calculation needs to be hard to generate and also easy to verify. If protein folding isn't easy to verify, it won't work as the basis of a decentralized consensus network such as Bitcoin.
Exactly. The cost of all the energy used to mine bitcoins and process transactions needs to be compared to all the financial infrastructure we have in place today. Think of how many much of the financial infrastructure today would disappear if the world operated mostly on Bitcoins. If an when that happens we'll reap enormous savings and it should end up being more environmentally friendly because mining operations do not require locality. It's very easy to put a mining operation in a region/country with access to cheap, renewable energy, whereas lots of our current financial infrastructure is located in places where the energy consumed is dirty and expensive.
I would love to see someone compare the energy costs of our current infrastructure and externalities vs how thing would be if Bitcoin or other compute intensive crypto-currency replaced all that.
There are fundamental reasons why this can't work, such as the need for proof of work to be progress free, to build off previous work, and security vulnerabilities which arise if proofs if work are not intrinsically worthless.
Instinctively I reel the same way, but if you think it through, it's equivalent to complaining about all the ink and paper and cotton and metal being "wasted" on bills and coins.
Question: If BTC were to disappear overnight, could all these machines be retooled to do something else? Or are they so specifically made that their only possible purpose is coin mining?
They can be used to power another decentralized consensus system. Currency is only one application being used on the bitcoin network. If bitcoin dies for whatever reason (political or technical) the hashing power could still be used to secure and build another decentralized consensus network that could have many other functions other than being a currency: such as ways to send stocks, bonds, smart property and so on... in an entirely decentralized manner.
55 comments
[ 2.3 ms ] story [ 36.5 ms ] threadThis is obviously wrong. Where did it come from?
I imagine they are comparing raw operations per second (in some measure) but the comparison is totally meaningless.
Nowadays it makes less sense since bitcoin hardware is ASIC/custom silicon designed & optimized for hashing as opposed to floating point operations making the comparison biased.
[1] http://www.top500.org/system/177556
Currently at 108239.28 Petaflops which is 5400x the performance of the IBM Sequoia. This is approximately the quoted amount of 4500. While I don't agree with the idea that a hashrate can be converted into flops because its more bitwise opts others are drawing similar calculations to UCSD.
Also a bit more of a discussion on this: https://bitcointalk.org/index.php?topic=7675.20
A Radeon 7990 gets around 1.2 gigahash/second and 8.2 teraflops.
The ASICS are of course skewed towards hashes from there, and I guess the supercomputer could be more skewed towards floating point than the GPU, but a large factor seems more likely than a fraction.
A bitcoin hash is about 1900 32-bit integer operations [1]. This is comparable to 1900 32-bit floating point operations, not one(!). Because most CPU/GPU, when they can execute N integer ops per cycle, can also execute N floating ops per cycle.
So the ratio clearly is in the ~1000 order of magnitude.
[1] https://en.bitcoin.it/wiki/Why_a_GPU_mines_faster_than_a_CPU...
That's the amount of work you do in a software implementation of SHA-2, running on a CPU or GPU. This abstracts the hardware complexity inside the chip, which is very relevant when we're going into ASIC's. SHA-2 is bit shifts, masks, and binary adders: it's much simpler than a floating point core.
A hardware comparison would be something like ops/(transistor * time). A recent Bitcoin press release [1] claims 500 billion SHA-2 hashes/second out of >1 billion transistors: ~500 hashes/(transistor-second). A 7990 has a peak throughput of 8.2 teraflops SP [2], out of 8.6 billion transistors: ~1,000 flops/(transistor-second). Both are on 28nm processes.
[1] http://finance.yahoo.com/news/uniquify-hashfast-announce-wor...
[2] https://en.wikipedia.org/wiki/Radeon_HD_7000_Series#Southern...
Now I think that in practice, the amount that the attacker could gain might be quite small compared with the total value of the system.
What I like to say is this: Would you use TLS if that was how it worked?
To be clear, each attacker must commit more work than every honest miner combined together.
To use a more extreme value, a vault containing gold that was protected by proof of work would be useless, since the work required to open it would have to equal the value of the contents in order to discourage thieves.
My point was that in some cases, like bitcoin, it might be possible that the value to the attacker of a successful attack on bitcoin was significant, but still much less than the total worth of bitcoin.
My original point was that when a technology comes along that allows banks to be more energy efficient, the incentives are all in place for banks to adopt that technology and reduce their energy consumption. There is never an incentive for a Bitcoin miner to reduce their power consumption.
Well that's one way of writing /joule, I suppose.
(kicking CPUs and GPUs out of the contest probably did a lot to lower power consumption, but I wonder if the growth of the network has overshadowed that)
The best you could do is create a cryptocurrency that requires similar hardware to the hardware used for other tasks, so that this hardware is commoditized. This would help other fields indirectly.
https://en.wikipedia.org/wiki/Commodity_money
Remember though that the proof-of-work calculation needs to be hard to generate and also easy to verify. If protein folding isn't easy to verify, it won't work as the basis of a decentralized consensus network such as Bitcoin.
I would love to see someone compare the energy costs of our current infrastructure and externalities vs how thing would be if Bitcoin or other compute intensive crypto-currency replaced all that.
Ripple Labs is giving away Ripple (XRP) for donating your computing power to solving scientific problems at https://www.computingforgood.org/.
https://bitcointalk.org/index.php?topic=346134.msg3709913#ms...