ASICMINER: Entering the Future of ASIC Mining by Inventing It

[runeks]

Enjoy your 6% of blocks over the last 4 days - BitFury has eaten your lunch

Let's get him guys!!!

[Swordsoffreedom]

Enjoy your 6% of blocks over the last 4 days - BitFury has eaten your lunch

Let's get him guys!!!

Moon units invade the fury base!!!

[Vycid]

Anyone got a poisson distribution calculator that can tell me what the odds of 0 blocks in 6 hours at 50 TH is? Probably less than 5%?

http://www.asicminercharts.com/live/

[David Chen]

Anyone got a poisson distribution calculator that can tell me what the odds of 0 blocks in 6 hours at 50 TH is? Probably less than 5%?

http://www.asicminercharts.com/live/

what happened?  
maybe, Friedcat is testing and deploying?

[Vycid]

Anyone got a poisson distribution calculator that can tell me what the odds of 0 blocks in 6 hours at 50 TH is? Probably less than 5%?

http://www.asicminercharts.com/live/

what happened?  
maybe, Friedcat is testing and deploying?

I have no idea.

http://blockchain.info/blocks/ASICMiner

As of this moment, it has been 7.5 hours since the last discovered block.

A back-of-the-envelope calculation (could someone do this properly, please?) tells me that AM has no less than 7% of the network hash at 50TH. If we assume 6 blocks per hour (I believe there are slightly more), the odds of this no-block streak are less than

(1-0.07) ^ (6*7.5) = 0.0381 ~ 3.81%, so I think it's worthwhile to start paying some attention here.

(If 50TH constitutes 8% of the network hash, that'd be (1-0.08)^(6*7.5) = 0.0235 ~ 2.35%)

EDIT: AM has just discovered a block, after approximately 8 hours since the last.

[kano]

Anyone got a poisson distribution calculator that can tell me what the odds of 0 blocks in 6 hours at 50 TH is? Probably less than 5%?

http://www.asicminercharts.com/live/

what happened?  
maybe, Friedcat is testing and deploying?

A block takes ... on average over a large sample ... X amount of time to find.

For Y TH/s ... X = Difficulty / (Y * 10^12 / 2^32)

However, that is a statistical average ... variance is a LARGE factor in it.

It's not hard to find a block on a pool with a factor of 8 times the expected time, and even on rare occasions 10 times ...

Edit: oops got Difficulty the wrong way around Fixed

[runeks]

Anyone got a poisson distribution calculator that can tell me what the odds of 0 blocks in 6 hours at 50 TH is? Probably less than 5%?

http://www.asicminercharts.com/live/

I don't have the numbers but I'm pretty sure it's a lot higher than 5%. More like 25%.

[coldguy]

I don't know where you got those numbers, but if that is an accurate schedule of Asicminer, we are pretty much fucked.
CoinTerra 2 PetaHashes by December, BitMine will bring 4 Petahashes by March, FastHash and KnCMiner who knows how much more.
CoinTerra and KnC are both already going full throttle using 28nm.

IF AsicMiner doesn't put their shit together bringing a couple of petahashes to the network, even 2 BTC per share will be overpriced.
There is no more reason to be holding back, being the most capitalized company in the bitcoin world, they should be leading this fierce competition.

Cointerra - doesn't have anything yet
BitMine - doesn't have anything yet
FastHash - doesn't have anything yet
KnCMiner - doesn't have anything yet

I don't know where is the informatino that CoinTerra and KnC are going full throttle using 28 nm, but I haven't seen proof of such a thing.

Asicminer has already experience with 130 nm chips and the new generation will be produced pretty soon afaik. You should walk before you start running.

Correct me if I'm wrong in anything

There is also Avalon, BitFury and BFL.

Cointerra founders claim they have previous experience in 28nm fab.

That is correct. Ravi Iyengar (CoinTerra lead) was a big shot at Samsung (CPU Lead Architect at Samsung Austin Research Center). The star power is even greater on CoinTerra's advisory board - their chairman, Naveed Sherwani, is also the chair of the Global Semiconductor Association Technical Steering Committee.

http://www.anandtech.com/show/7246/the-rush-to-bitcoin-asics-ravi-iyengar-launches-cointerra

http://arstechnica.com/business/2013/08/chip-veterans-form-new-startup-will-sell-high-end-bitcoin-miner-for-16000/

Even more important than the experience is the industry connections. Nothing gets done without knowing the right people.

There is no need to have so many big shots in bitcoin mining industry, which only increases the cost. Several smart heads can do the job pretty well.

Firedcat has proved himself and his teammates to be smart enough to do the job well. No other bureaucrats are needed.

[VeeMiner]

Anyone got a poisson distribution calculator that can tell me what the odds of 0 blocks in 6 hours at 50 TH is? Probably less than 5%?

http://www.asicminercharts.com/live/

I don't think you should sit behind your table and hit F5 and watch if there is a new block. The luck/variance is a very big factor. You should realize that the luck is only eliminated in a very long timeframe, that being 14 days+. I for example mine at Slush's pool (Which is usually around 10 percent of the whole network) and I have seen such streaks of bad luck - even 17 hours with no block found. Then there were five block found in 15 minutes. That's just the way it works. If the market is so irrational and bases the buy/sale of AM shares on the luck in last 6 hours, then that's another thing. It seems to me however that the weak hands were already shaken off and now we're stable around 2.5BTC. The share price will probably start rising again after the new chips are developed. I think there will be weeks now with pretty weak dividends and pretty low share prices... All of that said I don't plan on selling any of my shares

[Vycid]

I don't know where you got those numbers, but if that is an accurate schedule of Asicminer, we are pretty much fucked.
CoinTerra 2 PetaHashes by December, BitMine will bring 4 Petahashes by March, FastHash and KnCMiner who knows how much more.
CoinTerra and KnC are both already going full throttle using 28nm.

IF AsicMiner doesn't put their shit together bringing a couple of petahashes to the network, even 2 BTC per share will be overpriced.
There is no more reason to be holding back, being the most capitalized company in the bitcoin world, they should be leading this fierce competition.

Cointerra - doesn't have anything yet
BitMine - doesn't have anything yet
FastHash - doesn't have anything yet
KnCMiner - doesn't have anything yet

I don't know where is the informatino that CoinTerra and KnC are going full throttle using 28 nm, but I haven't seen proof of such a thing.

Asicminer has already experience with 130 nm chips and the new generation will be produced pretty soon afaik. You should walk before you start running.

Correct me if I'm wrong in anything

There is also Avalon, BitFury and BFL.

Cointerra founders claim they have previous experience in 28nm fab.

That is correct. Ravi Iyengar (CoinTerra lead) was a big shot at Samsung (CPU Lead Architect at Samsung Austin Research Center). The star power is even greater on CoinTerra's advisory board - their chairman, Naveed Sherwani, is also the chair of the Global Semiconductor Association Technical Steering Committee.

http://www.anandtech.com/show/7246/the-rush-to-bitcoin-asics-ravi-iyengar-launches-cointerra

http://arstechnica.com/business/2013/08/chip-veterans-form-new-startup-will-sell-high-end-bitcoin-miner-for-16000/

Even more important than the experience is the industry connections. Nothing gets done without knowing the right people.

There is no need to have so many big shots in bitcoin mining industry, which only increases the cost. Several smart heads can do the job pretty well.

Firedcat has proved himself and his teammates to be smart enough to do the job well. No other bureaucrats are needed.

Depends what "do the job well" means.

These guys are going straight for 28nm.

Anyone got a poisson distribution calculator that can tell me what the odds of 0 blocks in 6 hours at 50 TH is? Probably less than 5%?

http://www.asicminercharts.com/live/

I don't think you should sit behind your table and hit F5 and watch if there is a new block. The luck/variance is a very big factor. You should realize that the luck is only eliminated in a very long timeframe, that being 14 days+. I for example mine at Slush's pool (Which is usually around 10 percent of the whole network) and I have seen such streaks of bad luck - even 17 hours with no block found. Then there were five block found in 15 minutes. That's just the way it works. If the market is so irrational and bases the buy/sale of AM shares on the luck in last 6 hours, then that's another thing. It seems to me however that the weak hands were already shaken off and now we're stable around 2.5BTC. The share price will probably start rising again after the new chips are developed. I think there will be weeks now with pretty weak dividends and pretty low share prices... All of that said I don't plan on selling any of my shares

Yeah - I wasn't refreshing, just opened up that chart webpage and noticed there hadn't been anything in the past 7 hours (statistically significant at 95% confidence). Something showed up an hour afterward so it turned out to be a false alarm.

[VeeMiner]

Depends what "do the job well" means.

These guys are going straight for 28nm.

good for them. However friedcat obviously has much more experience with creating chips and devices for bitcoin mining. I would sooner trust Avalon or Asicminer to create new chip before KnCMiner and Cointerra and Hashfast since they have never created bitcoin mining ASIC chip before.

[Vycid]

Depends what "do the job well" means.

These guys are going straight for 28nm.

good for them. However friedcat obviously has much more experience with creating chips and devices for bitcoin mining. I would sooner trust Avalon or Asicminer to create new chip before KnCMiner and Cointerra and Hashfast since they have never created bitcoin mining ASIC chip before.

Making a functional ASIC is not excessively complicated. The complex part is the optimization.  I think you'll find it is really FC's team that lacks experience here, having designed only one chip - I quote, from AnandTech,

Ravi’s focus at SARC was on the CPU Mid-Core, including integer execution and special purpose registers

Being the chief architect on CPU mid-cores at one of the biggest consumer electronics companies in the world makes this guy far more qualified than a team of guys in China that made one simple chip at an archaic process node.

The hardware and boards should be the easy part, given the preponderance of existing packaging companies to partner with (at least for anyone with connections in the industry).

Point is - you should recognize FC has one year of direct experience and this guy has decades of related experience.

[stripykitteh]

Point is - you should recognize FC has one year of direct experience and this guy has decades of related experience.

That also ties in to a wider point; i.e., there are *hundreds* of vastly experienced chip designers with experience working for Intel, AMD, Samsung, et cetera who are capable of designing a 28nm ASIC. Acquiring the talent is not a big hurdle and there are probably a few more Cointerras lurking in the wings right now.

[FNG]

Point is - you should recognize FC has one year of direct experience and this guy has decades of related experience.

That also ties in to a wider point; i.e., there are *hundreds* of vastly experienced chip designers with experience working for Intel, AMD, Samsung, et cetera who are capable of designing a 28nm ASIC. Acquiring the talent is not a big hurdle and there are probably a few more Cointerras lurking in the wings right now.

Exactly. Money buys talent. AM has money. I don't see how one name attached to cointerra threatens AM so significantly.

AM has a massive head start and funds to buy whatever skillset or expertise they may be lacking

[coldguy]

I don't know where you got those numbers, but if that is an accurate schedule of Asicminer, we are pretty much fucked.
CoinTerra 2 PetaHashes by December, BitMine will bring 4 Petahashes by March, FastHash and KnCMiner who knows how much more.
CoinTerra and KnC are both already going full throttle using 28nm.

IF AsicMiner doesn't put their shit together bringing a couple of petahashes to the network, even 2 BTC per share will be overpriced.
There is no more reason to be holding back, being the most capitalized company in the bitcoin world, they should be leading this fierce competition.

Cointerra - doesn't have anything yet
BitMine - doesn't have anything yet
FastHash - doesn't have anything yet
KnCMiner - doesn't have anything yet

I don't know where is the informatino that CoinTerra and KnC are going full throttle using 28 nm, but I haven't seen proof of such a thing.

Asicminer has already experience with 130 nm chips and the new generation will be produced pretty soon afaik. You should walk before you start running.

Correct me if I'm wrong in anything

There is also Avalon, BitFury and BFL.

Cointerra founders claim they have previous experience in 28nm fab.

That is correct. Ravi Iyengar (CoinTerra lead) was a big shot at Samsung (CPU Lead Architect at Samsung Austin Research Center). The star power is even greater on CoinTerra's advisory board - their chairman, Naveed Sherwani, is also the chair of the Global Semiconductor Association Technical Steering Committee.

http://www.anandtech.com/show/7246/the-rush-to-bitcoin-asics-ravi-iyengar-launches-cointerra

http://arstechnica.com/business/2013/08/chip-veterans-form-new-startup-will-sell-high-end-bitcoin-miner-for-16000/

Even more important than the experience is the industry connections. Nothing gets done without knowing the right people.

There is no need to have so many big shots in bitcoin mining industry, which only increases the cost. Several smart heads can do the job pretty well.

Firedcat has proved himself and his teammates to be smart enough to do the job well. No other bureaucrats are needed.

Depends what "do the job well" means.

These guys are going straight for 28nm.

28nm is cool and I would admire them if they make a straight success.

What I mean here is those big shots are very fancy and they have big appetite. I am not sure if they would like to take or even understand the risks related to the bitcoin related business. This is something Friedcat has stated in his IPO post why he would not source funds from VC and would rather make it in the community.

As long as some times in future the BTC/USD price diverged from the difficulty, these companies and their customers will experience pressure, if they make such a huge investment in such a short time. That would be the real challenge. However, maybe I am wrong and bitcoin has evolved into a matured state.

[101111]

Being the chief architect on CPU mid-cores at one of the biggest consumer electronics companies in the world makes this guy far more qualified than a team of guys in China that made one simple chip at an archaic process node.

Rubbish.

[stripykitteh]

Point is - you should recognize FC has one year of direct experience and this guy has decades of related experience.

That also ties in to a wider point; i.e., there are *hundreds* of vastly experienced chip designers with experience working for Intel, AMD, Samsung, et cetera who are capable of designing a 28nm ASIC. Acquiring the talent is not a big hurdle and there are probably a few more Cointerras lurking in the wings right now.

Exactly. Money buys talent. AM has money. I don't see how one name attached to cointerra threatens AM so significantly.

AM has a massive head start and funds to buy whatever skillset or expertise they may be lacking

They do, but they won't be able to lock up all the talent. Nor will having money beyond a certain point make any difference, and it looks as though the cost of bringing a 28nm ASIC to market isn't that high a hurdle either. I think you are mistaken in thinking AM have a big lead over their competition. They will have a place in the market in the short to medium term, but there's no 'moat' around their business, unlike the ones Microsoft and Oracle enjoy. Anyone with a few million and the right contacts in the semiconductor industry can start eating into AsicMiner's market share within a few months. Jeepers, bitfury isn't even a qualified EE apparently and he's eating friedcat's lunch right now using 65nm.

[Rival]

I would not characterize it as FC getting his lunch eaten when FC sells out of everything he ever offers. Even with populist suggestions that his products are overpriced. Such remarks lead me to believe that people think FC is losing potential profits on products he has not yet produced, which could be said for pretty much anyone else in the marketplace.

Continuously selling out of everything you produce is a sign of success, not weakness.

[BitThink]

I would not characterize it as FC getting his lunch eaten when FC sells out of everything he ever offers. Even with populist suggestions that his products are overpriced. Such remarks lead me to believe that people think FC is losing potential profits on products he has not yet produced, which could be said for pretty much anyone else in the marketplace.

Continuously selling out of everything you produce is a sign of success, not weakness.

Agree with you.

 Having a reliable product line to continuously produce chips and boards is definitely the strength of AM, which is often be overlooked. One of the reasons AM choose 130nm in the beginning was that there were plenty of idle 130nm product lines in Shenzhen that time because exactly the technology is ancient, so AM could find one with very low price and did not need to compete with big players in EE world.

Now some startups are developing 22nm chips, which is the mainstream now. As a result, they have to try their best to secure the product lines. That means to compete with Intel, AMD, SamSung, .... Even if they win, the cost cannot be low for sure. In short, I believe they have the ability to design a chip, but how many they can produce and what is the cost is another thing.

[Mabsark]

I would not characterize it as FC getting his lunch eaten when FC sells out of everything he ever offers. Even with populist suggestions that his products are overpriced. Such remarks lead me to believe that people think FC is losing potential profits on products he has not yet produced, which could be said for pretty much anyone else in the marketplace.

Continuously selling out of everything you produce is a sign of success, not weakness.

Agree with you.

 Having a reliable product line to continuously produce chips and boards is definitely the strength of AM, which is often be overlooked. One of the reasons AM choose 130nm in the beginning was that there were plenty of idle 130nm product lines in Shenzhen that time because exactly the technology is ancient, so AM could find one with very low price and did not need to compete with big players in EE world.

Now some startups are developing 22nm chips, which is the mainstream now. As a result, they have to try their best to secure the product lines. That means to compete with Intel, AMD, SamSung, .... Even if they win, the cost cannot be low for sure. In short, I believe they have the ability to design a chip, but how many they can produce and what is the cost is another thing.

22nm is only mainstream for Intel. Mainstream for everyone else is 28nm. 20nm will become mainstream for them in 2014/2015 while Intel will be moving to 14nm in 2014.