4 ms·

"100k transactions per day" There lies your miscalculation. You use the underutilization of the network (~1.1tps currently) to calculate a transaction cost and

by mrb 12y ago

"100k transactions per day" There lies your miscalculation. You use the underutilization of the network (~1.1tps currently) to calculate a transaction cost and you claimed this cost cannot decrease (you said "processing more transaction will or should not lower these costs"). But as it is, the network is in fact capable of processing 7tps, therefore the true minimum electricity cost of a transaction is ~7x lower than what you claimed. "1 GHash/J"* This is a second but smaller error. Miners currently average closer to 1.5 Ghash/J, so that is another 1.5x decrease in your calculated transaction cost. These 2 errors (7 * 1.5 = 10.5) explain why your calculation was 10x off. Regarding micro payments, opinions are split but personally I do not think Bitcoin is well suited to them, or even needs to be suited to them to be a successful transaction processing system. So a $0.60 or $0.06 electricity cost per transaction is good enough right now, keeping in mind cost will go down as the block size limit is increased and as chips become more efficient in terms of Ghash/J.

danbruc 12y ago
Slightly above 1.5 GHash/J seems to be the best available hardware. I have no idea how often people throw away their mining equipment but I find it unconvincing that everybody is already using the newest hardware. The calculation of the transaction costs was for that particular time frame so 100k transactions was a pretty fair assumption. But I was also talking about the future without block rewards so you have a point there that there is room for more transactions. So maybe we will end up at about a Dollar if we throw in hardware costs, rental, cooling, profit and so on. This is still prohibitive for micropayments. And it makes the assumption that the network will not grow if the transaction volume grows. The problem I have with this is that I don't think that it would be a good idea. Processing ten times the volume as of today but protecting it with the same amount of money in hardware and electricity costs as of today kind of makes the network ten times more interesting to attack. I estimated that gaining 50 % of the processing power would cost you 100 million in hardware, add electricity for two, three months and you end up in a range where a determined state, large company or even individual could wreak havoc.
- vanzard 12y ago
  KnC Neptune does 1.8 GHash/J: https://www.kncminer.com/pages/neptune-setup-info https://www.kncminer.com/pages/neptune-setup-info > I find it unconvincing that everybody is already using the newest hardware. Only a third or at most half of the network is mining with the Neptune at 1.8 Ghash/J. The rest is mostly the 1½ year old Bitfury at 1.2 GHash/J. Then there is a bunch of Monarch at 1.4 GHash/J, some Antminer at 1.3 GHash/J, and so on. So yeah a 1.5 Ghash/J average is very likely. You have to remember mining is extremely competitive. If you are not on the latest generation hardware, you are making less money than the competition. > I estimated that gaining 50 % of the processing power would cost you 100 million in hardware These days the estimate is closer to half a billion dollars. (There was a nice article about that - can't remember the link, sorry)
  - danbruc 12y ago
    The list I used has the KnC Neptune at 1.429 GHash/J which seems to be the number they advertised before the update in July 2014. But if you have to replace your hardware every year this makes the whole thing even worse - not only are you spending over 200 millions for electricity each year, you are now also spending a similar amount on hardware.
    - vanzard 12y ago
      Officially they spec 1.4 GHash/J but in practice it measures 1.8 GHash/J: see the "0.57 watts per GH/J" quote in the link I gave. Only a few low-performing units bottom out at 1.4. > But if you have to replace your hardware every year this makes the whole thing even worse You used to have to replace your hardware every year, but not anymore. There was a big race to get the ASICs to the next fab node as Bitcoin startups progressively got better and better funded and could develop for more and more expensive processes. In 24 months we went: ASICMINER 130nm -> Avalon 110nm -> BFL Single SC 65nm -> Bitfury 55nm -> Jupiter 28nm -> Neptune 20nm. But now that we have reached the top-of-the-line 28-20nm, Bitcoin ASIC developers are just waiting for foundries to make the next node available (16nm), so the lifetime of a Bitcoin miner should grow from less than a year to at east ~2 years. If a Bitcoin miner lasts ~2 years, its hardware cost is relatively small compared to electricity costs, so replacing them is not much an issue: a $1500 2kW Neptune consumes $3500 of electricity over 2 years at $0.10/kWh. TCO = $5000. The hardware represents only 30% of that. So not only it is affordable but it makes sense to throw that hardware away and replace it for a more efficient one after 2 years.

by mrb 12y ago

danbruc 12y ago

Slightly above 1.5 GHash/J seems to be the best available hardware. I have no idea how often people throw away their mining equipment but I find it unconvincing that everybody is already using the newest hardware. The calculation of the transaction costs was for that particular time frame so 100k transactions was a pretty fair assumption. But I was also talking about the future without block rewards so you have a point there that there is room for more transactions. So maybe we will end up at about a Dollar if we throw in hardware costs, rental, cooling, profit and so on. This is still prohibitive for micropayments. And it makes the assumption that the network will not grow if the transaction volume grows. The problem I have with this is that I don't think that it would be a good idea. Processing ten times the volume as of today but protecting it with the same amount of money in hardware and electricity costs as of today kind of makes the network ten times more interesting to attack. I estimated that gaining 50 % of the processing power would cost you 100 million in hardware, add electricity for two, three months and you end up in a range where a determined state, large company or even individual could wreak havoc.

vanzard 12y ago
KnC Neptune does 1.8 GHash/J: https://www.kncminer.com/pages/neptune-setup-info https://www.kncminer.com/pages/neptune-setup-info > I find it unconvincing that everybody is already using the newest hardware. Only a third or at most half of the network is mining with the Neptune at 1.8 Ghash/J. The rest is mostly the 1½ year old Bitfury at 1.2 GHash/J. Then there is a bunch of Monarch at 1.4 GHash/J, some Antminer at 1.3 GHash/J, and so on. So yeah a 1.5 Ghash/J average is very likely. You have to remember mining is extremely competitive. If you are not on the latest generation hardware, you are making less money than the competition. > I estimated that gaining 50 % of the processing power would cost you 100 million in hardware These days the estimate is closer to half a billion dollars. (There was a nice article about that - can't remember the link, sorry)
- danbruc 12y ago
  The list I used has the KnC Neptune at 1.429 GHash/J which seems to be the number they advertised before the update in July 2014. But if you have to replace your hardware every year this makes the whole thing even worse - not only are you spending over 200 millions for electricity each year, you are now also spending a similar amount on hardware.
  - vanzard 12y ago
    Officially they spec 1.4 GHash/J but in practice it measures 1.8 GHash/J: see the "0.57 watts per GH/J" quote in the link I gave. Only a few low-performing units bottom out at 1.4. > But if you have to replace your hardware every year this makes the whole thing even worse You used to have to replace your hardware every year, but not anymore. There was a big race to get the ASICs to the next fab node as Bitcoin startups progressively got better and better funded and could develop for more and more expensive processes. In 24 months we went: ASICMINER 130nm -> Avalon 110nm -> BFL Single SC 65nm -> Bitfury 55nm -> Jupiter 28nm -> Neptune 20nm. But now that we have reached the top-of-the-line 28-20nm, Bitcoin ASIC developers are just waiting for foundries to make the next node available (16nm), so the lifetime of a Bitcoin miner should grow from less than a year to at east ~2 years. If a Bitcoin miner lasts ~2 years, its hardware cost is relatively small compared to electricity costs, so replacing them is not much an issue: a $1500 2kW Neptune consumes $3500 of electricity over 2 years at $0.10/kWh. TCO = $5000. The hardware represents only 30% of that. So not only it is affordable but it makes sense to throw that hardware away and replace it for a more efficient one after 2 years.