6 September 2013 by Chris Edwards
The "nonce" bit refers to a random value used in the process of generating Bitcoins - websites use nonces frequently to control user sessions to stop third parties from attempting to break into transactions. The Golden Nonce in this case is a custom chip designed to perform the calculations needed to generate Bitcoins way faster than the computer CPUs that originally sufficed for the job.
The neat thing about the algorithm is that complexity scales up over time so that as more computer cycles are deployed to obtain or 'mine' Bitcoins, the more cycles it will take. At the heart of the algorithm is hash generation - a technique used to encode passwords among others things. Your ability to mine Bitcoins depends on how many hashes your computer hardware is able to process per second. A regular x86 made today can crunch through about 5 million per second. A custom box made by the likes of Butterfly Labs for a mere $2500 can munch its way through 50 billion per second. Print your money for just $2500? What could go wrong with that?
If you'd managed to send your 50GH/s box back to September 2009 you'd be a milyonaire Rodders, cashing in to the tune of $100m per month. Well, you wouldn't, because Bitcoins weren't worth much at all in 2009 and didn't trade above 10 cents per Bitcoin until 2010. And the nature of the algorithm means that the complexity ratchet would simply have kicked in earlier, although you could probably have cornered the market with your magic money machine at that point.
Curiously, the initial spike in value - midway through 2011 - correlates roughly to the fastest rise in complexity in the history of Bitcoin. Between April and June, complexity rose tenfold so the 50GH/s box would have decelerated from 9000 Bitcoins per month to fewer than 1000. You would have to wait until Spring 2013 to be rolling in the dough though, as the dollar price per Bitcoin surged past $25 to spike at more than $225. And that was just at the point where the algorithm saw complexity move to ten times what it had been in mid 2011. It has taken less than half a year to move up tenfold again to a complexity rating of close to 90 million, five orders of magnitude over mid 2009. The total hash rate now stands at close to 800TH/s.
Butterfly Labs claims that its current-generation chip, based on a 65nm process, can turn in 4GH/s at close to 13W, which does not sound too bad. Blockchain.info works on an estimate of 650W per GH/s, which is closer to the output of a PC rig running GPUs. At that electricity cost, Bitcoin miners are massively underwater. With dedicated hardware, however, the electricity cost starts to make sense. However, Butterfly Labs does not seem to publish power numbers for the rigs, just the bare processors. Results from the web seem to indicate 5W per GH/s is realistic.
Assuming the same electricity rate as Blockchain, on pure electricity cost alone at current complexity, you should be ahead - making about half a dollar per day per GH/s. Scale that up to a 50GH/s rig and you're in the money - to the tune of $25 a day. But, add a zero to the complexity and if you look at the complexity trend right now that could arrive in a matter of months, you are down to less than $4 per day. And you've splurged $2500 to buy the box in the first place. If your box turns up rightaway, you might make the money back in a couple of months. But if not, you could be waiting for more than a year. And, as complexity ratchets up as more people buy this class of hardware, that rig quickly morphs into a doorstop.
The HashFast Technologies Golden Nonce aims to leapfrog the competition and is made on a 28nm process. The chip has only just taped out and so has yet to go into production but the company reckons it will do 400GH/s at a power consumption of 0.6W per GH/s. One of these processors will go into each Baby Jet, costing $5600 a piece. At current complexity levels, it looks really good, making its money back in less than a month with electricity cost a quarter or less than the Butterfly box. But it's not shipping yet. Once complexity increases, the payback looks far less certain even with the higher performance.
The use of pretty advanced chipmaking technology is one of the striking aspects of the whole Bitcoin micro-economy. A 28nm process is not exactly cheap to design with. The market is limited. Just five thousand thousand of these processors would double the total deployed hash rate. The mask cost for 28nm is generally estimated at around $5m and that does not include the cost of design tools and services, which are also far from cheap. There are way to get around each of these items but the company needs to sell hundreds if not thousands simply to break even. But each machine is adding to the hash rate, which pushes up the complexity, which further reduces the payoff.
The phrase "parabolic blow off" springs to mind.
Posted By: Chris Edwards @ 06 September 2013 12:52 PM General
Couldn't agree more. We have seen many enquiries over the last months to design and supply ASIC's for these applications, but when you look at the figures then the whole business model does not make a lot of sense.
|Posted By: Carl Hudson @ 06 September 2013 02:25 PM : Post a reply|
FuseTalk Standard Edition - © 1999-2013 FuseTalk Inc. All rights reserved.
"The Internet of Things used to be a buzz phrase in tech circles, but it's already so last century. Brace yourself for the Internet of Everything"
- Thermal Power Output the Civilian Equivalent to a Submarine Nuclear Reactors [11:40 pm 06/12/13]
- Building 37.5 GWh of Electrical Energy Storage for the Grid [11:07 pm 06/12/13]
- Convert PDF File to HTML Locally in Java Applications Using REST API [09:52 pm 06/12/13]
- Electrical Engineer Career Path [08:18 pm 06/12/13]
- Can we make further productive use of Wylfa Magnox Nuclear Station site at Angelsey? [01:08 pm 06/12/13]
Tune into our latest podcast