ASICMINER: Entering the Future of ASIC Mining by Inventing It

[jimmothy]

Moore's law states that transistor density doubles every 18 months. Something which no longer is true, but even it where, for bitcoin transistor density is almost irrelevant at this point; so what if you can cram twice as many cores in a chip, if it consume twice as much power you havent really gained much. You reduced the silicon cost per GH, but thats arguably one of the least important metrics right now.  Moore's law says nothing about compute efficiency, and I think its fair to say the industry is heading in to a brick wall in that regard too.

Thats not to say I dont expect improvements over where we are now, there is probably still a lot to be had from improving power delivery and riding down the schmoo plot from high voltage/highest performance per mm² to low voltage/highest power efficiency, but thats pretty much going to be a one time deal.

Source for Moore's law no longer being true?

It's true Moore's law says nothing about efficiency but Koomey's law says it should double even quicker.

[Mabsark]

Moore's law states that transistor density doubles every 18 months. Something which no longer is true, but even it where, for bitcoin transistor density is almost irrelevant at this point; so what if you can cram twice as many cores in a chip, if it consume twice as much power you havent really gained much.

It's not irrelevant at all. The smaller transistors use less power. Just because the transistor density doubles, that doesn't mean that the power consumption doubles too. For example, you could end up with twice the hash power for the same power consumption with a die shrink.

What it means is that you can fit more hash power in your DC.

[Puppet]

The smaller transistors use less power.

Not as a rule. Generally smaller gates can work with lower voltage, and that is what is increases power efficiency, its not a direct result and its increasingly untrue. 20 and 16nm offer next to no voltage drop, other than what can be achieved using finfets (unrelated to feature size).

Just because the transistor density doubles, that doesn't mean that the power consumption doubles too. For example, you could end up with twice the hash power for the same power consumption with a die shrink.

quite the opposite, just because you can shrink a transistor doesnt mean it will use less, let alone half as much power.  So you cram more on a die, but power density goes up, not exactly a huge advantage for bitcoin asics which already stretch the limits of power delivery, packaging and cooling.

[Bonam]

I take Puppet's spreadsheet to be a simple demonstration tool, I don't think his intention is to imply it's an accurate model. I would say his assumptions are much more sound than your main premise though. If even with free electricity a miner cannot recover their capital cost, NO ONE WILL BUY MINING EQUIPMENT. Sorry for the caps, but this keeps on coming up and it's a fundamental logic flaw. With free electricity your costs of running equipment are negligible. To say that the capital costs will never be recovered assumes that income approaches 0 as the network hashrate approaches infinity.

Answer one simple question, with the assumption that continued increases in difficulty will mean that even people with free electricity can't recovery their capital costs in equipment, who do you think will be buying the ~250PH/s worth of mining hardware necessary to even double the current hashrate?

People who can get the mining equipment a lot cheaper than it is normally priced for the "consumer". That is, people placing gigantic bulk orders, people with exclusive partnerships with manufacturers, and the manufacturers themselves.

[jimmothy]

I take Puppet's spreadsheet to be a simple demonstration tool, I don't think his intention is to imply it's an accurate model. I would say his assumptions are much more sound than your main premise though. If even with free electricity a miner cannot recover their capital cost, NO ONE WILL BUY MINING EQUIPMENT. Sorry for the caps, but this keeps on coming up and it's a fundamental logic flaw. With free electricity your costs of running equipment are negligible. To say that the capital costs will never be recovered assumes that income approaches 0 as the network hashrate approaches infinity.

Answer one simple question, with the assumption that continued increases in difficulty will mean that even people with free electricity can't recovery their capital costs in equipment, who do you think will be buying the ~250PH/s worth of mining hardware necessary to even double the current hashrate?

People who can get the mining equipment a lot cheaper than it is normally priced for the "consumer". That is, people placing gigantic bulk orders, people with exclusive partnerships with manufacturers, and the manufacturers themselves.

Doesn't matter what the J/GH is if you can never recover the capital cost of miners even with free electricity, which is where bitcoin mining stands right now.

So can some people profit from mining or not?

[MrTeal]

I take Puppet's spreadsheet to be a simple demonstration tool, I don't think his intention is to imply it's an accurate model. I would say his assumptions are much more sound than your main premise though. If even with free electricity a miner cannot recover their capital cost, NO ONE WILL BUY MINING EQUIPMENT. Sorry for the caps, but this keeps on coming up and it's a fundamental logic flaw. With free electricity your costs of running equipment are negligible. To say that the capital costs will never be recovered assumes that income approaches 0 as the network hashrate approaches infinity.

Answer one simple question, with the assumption that continued increases in difficulty will mean that even people with free electricity can't recovery their capital costs in equipment, who do you think will be buying the ~250PH/s worth of mining hardware necessary to even double the current hashrate?

People who can get the mining equipment a lot cheaper than it is normally priced for the "consumer". That is, people placing gigantic bulk orders, people with exclusive partnerships with manufacturers, and the manufacturers themselves.

What are you going on about? All you're describing is that some larger groups can get equipment cheaper than others, which is no different than some people getting cheaper power than others. Change $330 per TH/s to something else like $200/TH/s, and you're done. All that does is change one variable. In the end, you still end up with the same result; even the manufacturer won't make more equipment as it is not cost effective to do so, and the hashrate will stabilize.

[Mabsark]

The smaller transistors use less power.

Not as a rule. Generally smaller gates can work with lower voltage, and that is what is increases power efficiency, its not a direct result and its increasingly untrue. 20 and 16nm offer next to no voltage drop, other than what can be achieved using finfets (unrelated to feature size).

Just because the transistor density doubles, that doesn't mean that the power consumption doubles too. For example, you could end up with twice the hash power for the same power consumption with a die shrink.

quite the opposite, just because you can shrink a transistor doesnt mean it will use less, let alone half as much power.  So you cram more on a die, but power density goes up, not exactly a huge advantage for bitcoin asics which already stretch the limits of power delivery, packaging and cooling.

Yes, as a rule. What you are saying is simply wrong.

TSMC's 20nm process technology can provide 30 percent higher speed, 1.9 times the density, or 25 percent less power than its 28nm technology.

Comparing with 20SoC technology, 16FF+ provides extra 40% higher speed and 60% power saving.

Also, I didn't say a die shrink would use half the power, I said it could. For example, going from 130nm to 40nm.

[Puppet]

TSMC's 20nm process technology can provide 30 percent higher speed, 1.9 times the density, or 25 percent less power than its 28nm technology.

You're basically proving my point even with TSMC marketing material. The real gain by their own admission is just transistor density. Note the word "OR". Speed and power improvements are a joke, ask anyone who's actually used their process like nVidia.

Comparing with 20SoC technol
ogy, 16FF+ provides extra 40% higher speed and 60% power saving.

Yes, due to finfet, just like I said. Not because of the shrink.

Also, I didn't say a die shrink would use half the power, I said it could. For example, going from 130nm to 40nm.

Which only took what, 10 years? 

[Bonam]

What are you going on about? All you're describing is that some larger groups can get equipment cheaper than others, which is no different than some people getting cheaper power than others. Change $330 per TH/s to something else like $200/TH/s, and you're done. All that does is change one variable. In the end, you still end up with the same result; even the manufacturer won't make more equipment as it is not cost effective to do so, and the hashrate will stabilize.

My original point was that people buying miners are likely to make very little profit if any at all. Implicit in that statement is that this refers to consumers buying these products at "retail" prices. That is the statement I was defending. Which, I agree, is at this point only tangentially related to the discussion that is now going on.

I do agree that manufacturers will not make equipment when it is not cost effective to do so. That much is obvious. But the graph still does not accurately describe this, as, again, implicit in the graph is a technologically static environment, which does not correspond to reality.

As for the discussions about nearing the limits of Moore's Law, or how that corresponds to power efficiency, etc... the thing that people often forget is that the current paradigm of silicon integrated circuits is just one embodiment of the general trend of exponentially accelerating technological progress. There are many aspects of progress that we are just getting started on now: three-dimensional as opposed to two-dimensional elements and ICs, materials other than silicon with better performance, optical computing, quantum computing, etc. As any particular paradigm like just shrinking integrated circuit process size runs out of steam, other ones will pick up the slack. We're not gonna suddenly hit a brick wall where computational power just stands still... at least, not for a very long time, until we start to reach the fundamental physical limits that come from information theory.

[Mabsark]

You're basically proving my point even with TSMC marketing material. The real gain by their own admission is just transistor density. Note the word "OR". Speed and power improvements are a joke, ask anyone who's actually used their process like nVidia.

I'm not proving your point at all. The only way you could think that is if you completely misunderstood what you just read. The greater transistor density means that you can fit more transistors into the same space. In other words, you can keep the chip the same size and increase processing power and possibly power consumption or you can keep the processing power the same and decrease the power consumption and chip size.

Also, nVidia's argument against 20nm had nothing whatsoever to do with power consumption, it was based on transistor and wafer cost.

Yes, due to finfet, just like I said. Not because of the shrink.

Because of both actually.

Which only took what, 10 years?  

1 year for AM.

[explorer]

 

author=zinodaur link=topic=99497.msg9147280#msg9147280 date=1412897748]
Still nothing about dividends ? ...

when share price explodes up, you will know that divs have been announced.

[101111]

Bitfury just raised another $20M, so there's some confidence out there that asic development can still be a profitable.

[RoadStress]

Which only took what, 10 years?  

1 year for AM.

You are lacking reading comprehension.

Bitfury just raised another $20M, so there's some confidence out there that asic development can still be a profitable.

Source?

[Rival]

There is still a tremendous amount of "stupid" money out there just waiting to be plucked. The ability to raise capital has almost nothing to do with the ability to deliver value to consumers.

[freedomno1]

Bitfury just raised another $20M, so there's some confidence out there that asic development can still be a profitable.

Source?

From the Press Feeds
http://www.coindesk.com/bitfury-raises-20-million-asic-development-mining-output/

Or just here I guess

BitFury Raises $20 Million to Power New ASIC Chips, Increase Bitcoin Mining Output
BitFury has raised $20m in additional funding to complete development of its 28nm ASIC chip capable of achieving energy efficiency of 0.2 joules-per-gigahash (J/GH) and first announced in September.
[...]
The funding will seek to power the company’s development of what it hopes will be industry-leading ASIC chips, as well. BitFury announced in September that it is seeking to achieve energy efficiency of 0.2 J/GH by the fourth quarter of 2014, and sub-0.1 J/GH efficiency by mid-2015.

[jjdub7]

Still nothing about dividends ? ...

Wait...ASICMINER pays dividends?



 

[Puppet]

From the Press Feeds
http://www.coindesk.com/bitfury-raises-20-million-asic-development-mining-output/

from your link:
BitFury further indicated it would use the funds to increase the capacity of its data centers to 100 megawatts

100MW. Thats ~125PH with todays tech and if he achieves his 0.2J/GH goal later this year, thats 500 PH. Or 1EH next year if he does get it down to 0.1JGH

[Lincoln6Echo]

From the Press Feeds
http://www.coindesk.com/bitfury-raises-20-million-asic-development-mining-output/

from your link:
BitFury further indicated it would use the funds to increase the capacity of its data centers to 100 megawatts

100MW. Thats ~125PH with todays tech and if he achieves his 0.2J/GH goal later this year, thats 500 PH. Or 1EH next year if he does get it down to 0.1JGH

[RoadStress]

100MW. Thats ~125PH with todays tech and if he achieves his 0.2J/GH goal later this year, thats 500 PH. Or 1EH next year if he does get it down to 0.1JGH

I wonder if FC will address this issue.

[Chris_Sabian]

From the Press Feeds
http://www.coindesk.com/bitfury-raises-20-million-asic-development-mining-output/

from your link:
BitFury further indicated it would use the funds to increase the capacity of its data centers to 100 megawatts

100MW. Thats ~125PH with todays tech and if he achieves his 0.2J/GH goal later this year, thats 500 PH. Or 1EH next year if he does get it down to 0.1JGH

100MW is a huge amount of power.  About 1/100 the size of a coal powerplant.