Bitcoin Forum

So the general agreement by those knowledgeable in the field is that by 2020, semiconductor fabrication technology will hit a wall because it will no longer be technologically feasible to produce smaller and smaller transistors. We are already at 14 nm and the wall is expected to be reached once we hit about 5-7 nm, which should be attainable by 2018-2022. Beyond that, processors would no longer function because electrons would simply tunnel through the silicon gates. In contrast, the Pentium 4 was a 130 nm chip and the original Pentium was an 800 nm chip so you can see that Moore's law, i.e. the "engine" behind the exponential increases in computing power during the last couple of decades, is coming to an end.

How will this affect Bitcoin? How will miners be affected by this?

I think effectively 14nm will be our stopping point, to be honest. It's not the electron that stops us, it's the inability to cool a die efficiently, thus shortening the life of the chip.


You're forgetting though, Miners won't be affected by this at all, neither will Bitcoin.
You're talking about an engineering issue.
Chips will either get larger, or more numerous.
Water cooling will probably have to replace air and aluminum.
And Bitcoin miners will continue to hash out blocks.
And BFL will still never have shipped a working Monarch.

LOL  :D :D :D :D :D :D :D :D

stop using "electrical processor" ... use light (laser) processor instead !
http://www.extremetech.com/extreme/143597-pushing-computers-towards-petahertz-with-femtosecond-lasers-and-weird-dielectrics

The massive progress we've achieved in research of graphite lately in combination to quantum computing technologies could bring computing to the next level in the upcoming decades. This could help processing information to a new level. Maybe by that time processors of today won't be needed anymore.

yes. new materials. progress is not only pimp up what is already there, but also inventing new ways. It will never stop, but yes can slow down. But what mostly slow down some progress in past was money. It it is in their interest and they are strong enough they can do it.

Traditional chips will be replaced by something else, just like they themselves replaced vacuum tubes. Computational power / dollar will continue to rise exponentially.

Fixed that for you. The dollar will no longer exist by the time we transcend modern processor technology.

The hashrate and difficulty will not increase as fast as technology improvements slow.

A fairly simple conclusion.

if chip developments stall in 2020 due to cooling limits
there will be improvments in cooling to allow higher density chips

Devs would no longer be able to make more efficient (in terms of electric usage) machines.

Devs could make machines that produce less heat, that have more efficient cooling systems, or have some other advantage.

(I'm taking the assumptions that Moore's law exists at all, really does end in 2020, and implies that processing power will stagnate...after all, there are theoretical limitations on processing power. Even if we're not nearing those yet, we just might do that one day.)
It will continue to produce blocks at 10 minute intervals. It's designed to do so with practically any amount of computing power. No harm done at all.
Without substantially better hardware coming to market every few months, difficulty will level off and reflect the investment people have in the network (instead of that plus the progress of fast-developing ASICs and Moore's law, as we have now). The difficulty may even drop now and then, especially whenever the block reward halves, but as long as it doesn't do so too quickly, we'll continue to have block confirmations averaging under 15 minutes. The lifespan of mining hardware will become more relevant, so people will want more reliability in their ASICs. Mining will become a bit less speculative (because the future difficulty will not be quite so unknown) and market forces will drive the profit margins to be narrower than they are now, putting it further out of reach of people without super-low-cost energy. This might result in some centralization of mining power, but probably no worse than the current GHash.IO situation.
As mining becomes less important to the economy of Bitcoin, it will need to stand on its own legs as a currency, or fall flat. I think it will stand.  :)

Moore's law can never catch up to the crypto in Bitcoin. The only thing that could (MAYBE we don't know yet) would be quantum computers.

There is a limit as to how efficient in terms of electric usage computers can be.

The best posts are the ones that are so serious and truthful and stick the jab in at the end.  Now I have to clean fruit punch off the monitor... well done sir, well done.

Agreed that by 2020 the 1/2s will have made the BTC minimal.  Farms will still exist but I doubt we'll have a race for expansion like we currently have.

I hope that the mining hardware will become cheaper as it becomes more standard, more commoditised and longer-lived. It will fall within the reach of people who don't necessarily have super-cheap electricity, but who are able to make use of the waste heat. I'm hoping it will become a way for hobbyists to heat their homes.

The miners generally have a longer "life" then the amount of time they are actually used as they eventually use more electricity then they generate in bitcoin.

I think you are correct to say that a lot of people may use miners to generate heat, likely mostly up north in the winter.

I am interested to know how efficient they are at heating then other space heaters are.

Hehe right. Or maybe exist but not so important anymore. If some people are to be believed Bitcoin/USD value will continue to rise exponentially longterm. Which give computer power/usd a exponential growth that grows exponentially. That's pretty fast ;) 

mining is maybe what allows the 2020 ~ or whatever moores law to be overcome, eg mining tech drives processor dev. Not intel etc

So wait, if Moore's law ends in 2020 then will the Bitcoin mining difficulty stop increasing? So right now you could buy a 1TH/s miner for something like 7 BTC, make 1 BTC in the first month, and never break even because the difficulty would continue increasing. But a 5-7 nm miner bought in the beginning of 2020 will last you indefinitely?

I'm confused. ???

No, they only cost a little over 3BTC

https://bitmaintech.com/productDetail.htm?pid=00020140429083756017VwGm90Xx06B5

dont worry, 2020 will not be the end.

But the Mayans said 2012 so we are overdue....
It has to be 2020.....

 ::) ::) ::) ::) ::) ::) ::) ::) ::) ::) ::) ::)

The increase of mining difficulty is largely affected by two things

1) Increasing number of miners
2) Advancing technology (most importantly the amount of J/GH)

The rate of which technology advances in mining is pretty predictable, as we quickly follow the computer technology (recent miners are at about 28/24nm right now and we're quickly advancing to 14nm to keep up with computer chips. By next year ASIC chips will probably be as small as computer chips, thus the technology in ASICs will slow down compared to how fast they currently advance in technology.

This will make it more profitable to mine with the 14nm chips, as it will take a while before there will be more efficient chips. Theoretically that would mean 14nm miners would be guaranteed to be profitable, however there are a few catches to this.

First of all, because many people will expect the 14nm miners to be profitable, many more miners will buy that hardware. As well as we can expect bitcoin to be more popular in the first place, and by extension more people will be interested in mining them. More miners mean more difficulty and thus the 14nm miners may not be as profitable anymore just because there are too many miners using them.

Second there's the fact that electricity is much more expensive in Europe ($0.30~$0.45 per kWh) than it is in Asia and America ($0.08~$0.15 per kWh). Therefore we can assume large 14nm mining farms will be build in Canada, Russia or China which will be able to stay profitable while miners in other countries will not be able to be profitable because of overly expensive electricity.

Maybe I should start a kickstarter campaign to fund a private mining farm with 14nm chips in canada with chips to be designed privately. Investors being stakeholders of the mining farm ofc. This would be pretty interesting.

It will be the invention of a true general artificial intelligence that will one day render Moore's theory irrelevant. That or nanotechnology chips.

The end of Moore's law has been predicted since the early 1980s, just like 300 baud was the max modem speed.

Perhaps the "experts" are right this time, but I wouldn't count on it.

What about storage and bandwidth?  At the moment, using the core Bitcoin client is not viable for many people due to space requirements as well as the amount of time required to sync.  Will technologies in both storage and bandwidth outpace the current growth rate of the blockchain?

I'm running a core node, 21gb isn't a lot of data to hold right now. With current SSD and hybrid SSD drives, storage isn't even a concern, and probably will be likely to improve in both reliability and longevity.

Bandwidth? Yeah, no, a full node doesn't even begin to bump up to my limit even now. It would take 16x larger of a transaction size maximum to brush against the throughput limitations of a DOCSIS 3.0 cable modem.

Since I'm running my node on a dedicated 512gb SSD, maybe my kids will need to upgrade it after I'm gone.

As an aside, how do you find the Bitcoin Core client 'not viable for many people'?

The problem is quantum mechanical one, and not an estimate (Though the specific year 2020 is an estimate). When transistors get too small (specifically, when the channel gets too small), electron tunneling (https://en.wikipedia.org/wiki/Quantum_tunnelling) will take place and electrons can cross the channel (turning the transistor on) without needing the gate, which usually facilitates this.


I actually posted a thread (https://bitcointalk.org/index.php?topic=547893) on this back in March.

Fixed that for you.

Computer chips were once vacuum tubes. Do you think these people could have imagined the modern microchip?

http://www.computersciencelab.com/ComputerHistory/HtmlHelp/Images2/eniac3.gif

I doubt it. The past always seems incomprehensible to the people of the future, I see no reason why THAT rule would not apply in 2020.

I think if you're a user that only needs to make a few payments per month, you're not going to be running the core client every day.  Syncing only whenever you need to make a payment can take a long time.  That can be a problem for some users I'd think.

The current average block size is roughly 0.22MB.  A year ago the average was around 0.12MB.  The average block size is growing at an exponential rate.  However, even extrapolating the growth rate linearly, we get to 500GB by the year 2025 and would hit the 1MB block size limit by the year 2022.  

So if you do actually live past 2025, you might need to dish out for some extra storage.  However, I think by the year 2025 500GB of storage will be pennies.  ;)

I'm not very well educated on this subject, but I know it would be possible to crack private keys if you just had enough computing power (like, a lot). Is it possible to estimate a time frame for this using Moore's law?

Millions of years, that including moores law.

Banks will be hacked before a single bitcoin wallet will be compromised. (Not counting careless persons who get their private key stolen).

Regarding the full nodes, bitcoin was designed not to trust anyone, running a full node yourself is the best way to do this as otherwise you need to trust a 3rd party node. I'm sure the bitcoin developers understand this and will do everything they can to keep the bitcoin core viable for everyone. Everyone who wants to run a node can (and should) do it.

I'm currently not running a dedicated node yet, but I will at some point. However my gaming computer is a node when it is on (it's off during the night).

well done zimmah

You don't need to run a full node to fully benefit from the trustless nature of bitcoin.  It's just that doing so supports the network.

QT connects to multiple nodes and will only connect to nodes that are "trusted" by the network, eg not ones that relay invalid TX or other TX that should not be relayed.

yes. new materials. progress is not only pimp up what is already there, but also inventing new ways. It will never stop, but yes can slow down. But what mostly slow down some progress in past was money. It it is in their interest and they are strong enough they can do it.