ASICMINER: Entering the Future of ASIC Mining by Inventing It

[KS]

My understanding is that sha256 asics are very simple and basically work in parallel*, so a 130nm design can easily be scaled to a 40nm design by adding more cores.
*I might be wrong about that, but they are connected in some simple way where you can add more cores and not have to tinker too much to make them all work.

This means that for the same sha256 asic design, you should have a performance evolution more or less equivalent to the square of the size reduction from one node to the next (physics permitting - 28nm doing some funky things AFAIK, but 40nm should still be safe).

Actually, the performance increase is even greater than size reduction, because the clock speed is also usually increased.
BUT on the other hand, the production cost for the same die area is MUCH higher for smaller nodes.  NRE costs are also MUCH higher.  Of course, the overall cost per GH/s is still lower, but not that much lower.

P.S. Also, developing smaller node design is not only more expensive, but takes much more time.  For example, at 130nm you can do a full-chip physical simulation, while for 28nm simulating the whole chip would take eternity, so they usually simulate only small parts of the chip, which increases probability of errors and degradations.  The software toolchains are also much more complex and expensive.  That's why 130nm chip can only take a few weeks to develop, while 28nm takes several months at best.  KnC managed to shortcut this by using a standard-cell design (all other existing Bitcoin ASICs are full-custom), but this means that future 28nm chips will be both more cost-efficient and more energy-efficient than KnC.

I think the important thing for AM is "hash density", since deploying new data centers is costly and time consuming. With a 40nm design they could potentially multiply their hashing power by an order of magnitude or more in the same space (cue to the potential need for special cooling solutions).

You point out the price per GH/s as being similar, but is that on a per wafer basis, i.e. without the NRE's, or do you somehow extrapolate the production numbers and include them as well?

[muyuu]

Just initiated the AM100 transfer, will report back experiences.

I transfered AM1 from Havelock no prob. It took 4-5 days.

[Strange Vlad]

I think the important thing for AM is "hash density", since deploying new data centers is costly and time consuming. With a 40nm design they could potentially multiply their hashing power by an order of magnitude or more in the same space (cue to the potential need for special cooling solutions).

They don't depend on self-mining though.  There's also franchising as well as sales.

You point out the price per GH/s as being similar, but is that on a per wafer basis, i.e. without the NRE's, or do you somehow extrapolate the production numbers and include them as well?

I didn't say the price is similar...  Look, I don't know exact numbers of course, but what I wanted to say is that if there were, say, 20x more hashpower per die, but the die costed 5x as much, then the chip would only have 4x greater price efficiency, not 20x. And yes, the costs are greater for both fixed AND recurring costs, so I'm not extrapolating anything.

[data]

For the people worried about the 40nm talk here when others are using 28nm: check this article about "real" feature size, not marketing blub.

http://spectrum.ieee.org/semiconductors/devices/the-status-of-moores-law-its-complicated

(and never forget the energy density of these chips.s..)

[HowlingMad]

How to Convert Your AM100 (or TAT.ASICMINER) Shares to Direct Shares with Friedcat
 1. Using the Transfer feature, push the AM100 (or TAT.ASICMINER) shares to the issuer account. They must be a 105:1 ratio, this means the minimum is 105 AM100 or TAT.ASICMINER shares.

Would this trade-in also apply to users on Havelock?  If so, I did not see a 'transfer' function.
Thanks in advance.

[daCoops]

G.ASICMINER & AM100 (TAT.ASICMINER) Conversion Offers
(Limited Time Only, Ends Nov. 15, 2013)
With Bitfunder blocking US citizens, requiring ID verification, and not allowing exports unless you have 250 shares, we know some of you feel stuck and are looking for options. Below are two new, temporary services we are offering.

G.ASICMINER Offer (Limited Time Only, Ends Nov. 15, 2013)
We have decided to offer a limited time service to convert your G.ASICMINER shares into AM100 (HavelockInvestments.com) fractional shares. There is no fee to convert to AM100 (fractional shares), but please note that AM100 shares carry a 5% management fee on dividends.

AM100 & TAT.ASICMINER Offer (Limited Time Only, Ends Nov. 15, 2013)
Additionally, we will be offering AM100 or TAT.ASICMINER shareholders a temporary service to convert their shares into AM1 or direct shares for a fee of 5 (five) TAT.ASICMINER shares per whole share. That is a 105:1 ratio.

Hi TAT,

This is an offer that I'd like to take up - but some advice would be great. I have 5174 TAT.ASICMINER shares on BitFunder, and 150 AM100 shares on HaveLock. Would like to convert these both to AM1 just on Havelock, and have them all in one Exchange.

Whats best way to do? Looks like I'd have to transfer 4200 TAT.ASICMINER into 4 AM1, and then 105 AM100 into 1 AM1.

This correct? Any ideas what I can do with the left-over shares..?

[ThickAsThieves]

How to Convert Your AM100 (or TAT.ASICMINER) Shares to Direct Shares with Friedcat
 1. Using the Transfer feature, push the AM100 (or TAT.ASICMINER) shares to the issuer account. They must be a 105:1 ratio, this means the minimum is 105 AM100 or TAT.ASICMINER shares.

Would this trade-in also apply to users on Havelock?  If so, I did not see a 'transfer' function.
Thanks in advance.

Yes, this is for Havelock AM100 and Bitfunder TAT.ASICMINER shares.

Please check your Havelock Account Portfolio Page for a link that says "Xfr" by your shares. If it's not there, email us.

[ThickAsThieves]

G.ASICMINER & AM100 (TAT.ASICMINER) Conversion Offers
(Limited Time Only, Ends Nov. 15, 2013)
With Bitfunder blocking US citizens, requiring ID verification, and not allowing exports unless you have 250 shares, we know some of you feel stuck and are looking for options. Below are two new, temporary services we are offering.

G.ASICMINER Offer (Limited Time Only, Ends Nov. 15, 2013)
We have decided to offer a limited time service to convert your G.ASICMINER shares into AM100 (HavelockInvestments.com) fractional shares. There is no fee to convert to AM100 (fractional shares), but please note that AM100 shares carry a 5% management fee on dividends.

AM100 & TAT.ASICMINER Offer (Limited Time Only, Ends Nov. 15, 2013)
Additionally, we will be offering AM100 or TAT.ASICMINER shareholders a temporary service to convert their shares into AM1 or direct shares for a fee of 5 (five) TAT.ASICMINER shares per whole share. That is a 105:1 ratio.

Hi TAT,

This is an offer that I'd like to take up - but some advice would be great. I have 5174 TAT.ASICMINER shares on BitFunder, and 150 AM100 shares on HaveLock. Would like to convert these both to AM1 just on Havelock, and have them all in one Exchange.

Whats best way to do? Looks like I'd have to transfer 4200 TAT.ASICMINER into 4 AM1, and then 105 AM100 into 1 AM1.

This correct? Any ideas what I can do with the left-over shares..?

Simply email these details and we'll sort it out with you, but generally, any leftover shares will need to be held as-is or sold at market.

[webbrowser]

https://blockchain.info/block-height/266903

AsicMiners slogan should be "King of Orphaned Blocks" after the last 2 weeks they had.

I wonder whats the reason for that. Maybe asicminer is too poorly connected to other wallets? I mean a miner that is connected to as many wallets as possible and those wallets are again connected to as many miners as possible should have an advantage against a miner that is poorly connected because the block that is accepted by most miners wins. When i understand it correctly. So i only see that as an explaination. I doubt a internet connection can be that bad to have such a big effect.

I guess this is relevant: http://bitcoin.stackexchange.com/questions/8399/how-can-pool-operators-reduce-the-number-of-orphaned-blocks-that-they-mine

We've seen AM's "data center", so who knows what ISPs were chosen. Running services out of mainland china is not as simple as you'd think:

https://bitcointalk.org/index.php?topic=99497.msg2733783#msg2733783

And this may be relevant for the question within that reference, asking why more transactions are not included.
https://gist.github.com/gavinandresen/5044482

[keewee]

How to Convert Your AM100 (or TAT.ASICMINER) Shares to Direct Shares with Friedcat
 1. Using the Transfer feature, push the AM100 (or TAT.ASICMINER) shares to the issuer account. They must be a 105:1 ratio, this means the minimum is 105 AM100 or TAT.ASICMINER shares.

Would this trade-in also apply to users on Havelock?  If so, I did not see a 'transfer' function.
Thanks in advance.

Yes, this is for Havelock AM100 and Bitfunder TAT.ASICMINER shares.

Please check your Havelock Account Portfolio Page for a link that says "Xfr" by your shares. If it's not there, email us.

When transferring AM100 shares to AM1 on Havelock we click on the Xfr link and it directs us to transfer the shares to another user by email address. Do we use the same email address (tat.investments@gmail.com) for the share transfer as we do for the email we use to send you the details?

[bitcoin.newsfeed]

Its November.

"We made an order of another brand of power chips which are about two weeks late. In October this will provide another 500TH/s (Gen1)."

Failed.

"ASICMINER intends to grow the franshising program at a faster pace in October, so please participate in this process if you are capable of hosting blades and looking to mitigate some of your risks in buying hardware."

Failed.

"Because we are collecting funds to get ready for the exponentially increased devices to be assembled in September and October."

Failed.

"We hope to share a good video or exhibition materials by the end of the month."

Failed.

"The board has committed to meeting even more often than we have been, and intentionally gleaning whatever info can be shared from each meeting, and posting it to public."

Failed.

Underpromise and overdeliver ?  Not anymore.

[jimmothy]

Its November.

"We made an order of another brand of power chips which are about two weeks late. In October this will provide another 500TH/s (Gen1)."

Failed.

Never did he say 500TH/s would be added to their solomining operation. From what I and many others have taken from it is that the 500TH/s will be available for sales.

"ASICMINER intends to grow the franshising program at a faster pace in October, so please participate in this process if you are capable of hosting blades and looking to mitigate some of your risks in buying hardware."

Failed.

How is this failed? AM has a significant amount of TH/s from franchising..

"We hope to share a good video or exhibition materials by the end of the month."

Failed.

"We hope to" not "we will guaranteed".

[shmoula]

Are there similar plans with TAT.VIRTUALMINE on bitfunder? To change it for something on havelock...

G.ASICMINER & AM100 (TAT.ASICMINER) Conversion Offers
(Limited Time Only, Ends Nov. 15, 2013)
With Bitfunder blocking US citizens, requiring ID verification, and not allowing exports unless you have 250 shares, we know some of you feel stuck and are looking for options. Below are two new, temporary services we are offering.

G.ASICMINER Offer (Limited Time Only, Ends Nov. 15, 2013)
We have decided to offer a limited time service to convert your G.ASICMINER shares into AM100 (HavelockInvestments.com) fractional shares. There is no fee to convert to AM100 (fractional shares), but please note that AM100 shares carry a 5% management fee on dividends.



AM100 & TAT.ASICMINER Offer (Limited Time Only, Ends Nov. 15, 2013)
Additionally, we will be offering AM100 or TAT.ASICMINER shareholders a temporary service to convert their shares into AM1 or direct shares for a fee of 5 (five) TAT.ASICMINER shares per whole share. That is a 105:1 ratio.

NOTE: If you want to go direct from G.ASICMINER to AM1 or direct shares, you may, but you must pay the 5 (five) TAT.ASICMINER-shares-per-whole-share fee still.



Example #1: You have 10 G.ASICMINER shares, but you want to convert them to AM100 fractional shares. Pay no fees, simply push your G.ASICMINER shares to TATInvestments and email us according to the instructions below.

Example #2: You have 10 G.ASICMINER shares, but you want to convert them to direct shares. Push the 10 G.ASICMINER shares plus 50 TAT.ASICMINER shares to TATInvestments, and follow the instructions below.

Example #3: You have 1000 AM100 shares, but you want to convert them to AM1 shares. Push 1050 AM100 shares back to us, and email us according to the directions below.



The following limitations apply:
1. You must initiate each conversion request before November 15, 2013.
2. Due to many requests, your migration might take anywhere from 1-14 days to process.
3. Beginning Nov. 15th, these temporary services will be discontinued.
4. We can make no guarantees that Bitfunder or Havelock will not impose additional limitations on trading in the future.



How to Convert G.ASICMINER Shares to AM100:
 1. Push the amount of shares to the issuer account on Bitfunder: TATInvestments
 2. Make the subject of your email: G.ASIC
 3. Send an e-mail to tat.investments@gmail.com from your account email address registered with HavelockInvestments.com and include the following info:
  - Bitfunder account name (not email),
  - Bitfunder public wallet address
  - Quantity of shares pushed
  - Havelock registered email address


How to Convert Your AM100 (or TAT.ASICMINER) Shares to Direct Shares with Friedcat
 1. Using the Transfer feature, push the AM100 (or TAT.ASICMINER) shares to the issuer account. They must be a 105:1 ratio, this means the minimum is 105 AM100 or TAT.ASICMINER shares.
 2. Make the subject of your email: AM100
 3. Send an e-mail to tat.investments@gmail.com from your account email address registered with HavelockInvestments.com and include the following info:
  - Account name
  - Quantity of shares pushed
  - Wallet address that you want the direct shares attached to
  - Email address you want the shares attached to

Do NOT send your request more than once, and do NOT include multiple requests in one email.

Conversions will be transferred in batches, and confirmations will ONLY be sent after each batch. Do NOT send emails requesting updates.

NOTE: TAT Investments is not responsible for any lost value, arbitrage, or trading opportunities due to delays in processing transfer requests. TAT Investments is not responsible for any limitations or discontinuations of services or access to liquidity imposed by the host exchanges.

[Rannasha]

Are there similar plans with TAT.VIRTUALMINE on bitfunder? To change it for something on havelock...

Could've checked the TAT.VIRTUALMINE thread ...

TAT,

Seeing as Bitfunder will be freezing out all US-based customers from exiting positions starting Nov 1, is there any thought into a process for moving our TAT.VIRTUALMINE shares over to Havelock (similar to the process that happened for TAT securities over on BTCT?)

TAT.VM does not, and will not, exist on Havelock. Migrations to Bitfunder will be completed.

[Zubilica]

OK and  G.ASICMINER-PT - > AM1 , how to ?

[SebastianJu]

My understanding is that sha256 asics are very simple and basically work in parallel*, so a 130nm design can easily be scaled to a 40nm design by adding more cores.
*I might be wrong about that, but they are connected in some simple way where you can add more cores and not have to tinker too much to make them all work.

This means that for the same sha256 asic design, you should have a performance evolution more or less equivalent to the square of the size reduction from one node to the next (physics permitting - 28nm doing some funky things AFAIK, but 40nm should still be safe).

Actually, the performance increase is even greater than size reduction, because the clock speed is also usually increased.
BUT on the other hand, the production cost for the same die area is MUCH higher for smaller nodes.  NRE costs are also MUCH higher.  Of course, the overall cost per GH/s is still lower, but not that much lower.

P.S. Also, developing smaller node design is not only more expensive, but takes much more time.  For example, at 130nm you can do a full-chip physical simulation, while for 28nm simulating the whole chip would take eternity, so they usually simulate only small parts of the chip, which increases probability of errors and degradations.  The software toolchains are also much more complex and expensive.  That's why 130nm chip can only take a few weeks to develop, while 28nm takes several months at best.  KnC managed to shortcut this by using a standard-cell design (all other existing Bitcoin ASICs are full-custom), but this means that future 28nm chips will be both more cost-efficient and more energy-efficient than KnC.

I think the important thing for AM is "hash density", since deploying new data centers is costly and time consuming. With a 40nm design they could potentially multiply their hashing power by an order of magnitude or more in the same space (cue to the potential need for special cooling solutions).

You point out the price per GH/s as being similar, but is that on a per wafer basis, i.e. without the NRE's, or do you somehow extrapolate the production numbers and include them as well?

Its correct that the advantage of lower process nodes arent so high in cost like it looks on the first view since a chip is way more expensive. But a big factor to keep in mind is that you save a lot money by assembly, you need less miner boards, you need less work to do for the same TH and you need less space. Not to forget the higher speed to bring up TH.
So as long as you dont plan to sell... for mining lower process nodes have some good advantages.

[weaknesswaran]

Its correct that the advantage of lower process nodes arent so high in cost like it looks on the first view since a chip is way more expensive. But a big factor to keep in mind is that you save a lot money by assembly, you need less miner boards, you need less work to do for the same TH and you need less space. Not to forget the higher speed to bring up TH.
So as long as you dont plan to sell... for mining lower process nodes have some good advantages.

The NRE costs for all 28nm were payed buy customers preorder funds that were willing to dig their own grave.
Not an option for Asicminer.

[JimiQ84]

It looks like USB miners are done and gone. There are only few left in hands of resellers and they won't be made again. Let's wait for gen2

[SebastianJu]

https://blockchain.info/block-height/266903

AsicMiners slogan should be "King of Orphaned Blocks" after the last 2 weeks they had.

I wonder whats the reason for that. Maybe asicminer is too poorly connected to other wallets? I mean a miner that is connected to as many wallets as possible and those wallets are again connected to as many miners as possible should have an advantage against a miner that is poorly connected because the block that is accepted by most miners wins. When i understand it correctly. So i only see that as an explaination. I doubt a internet connection can be that bad to have such a big effect.

I guess this is relevant: http://bitcoin.stackexchange.com/questions/8399/how-can-pool-operators-reduce-the-number-of-orphaned-blocks-that-they-mine

We've seen AM's "data center", so who knows what ISPs were chosen. Running services out of mainland china is not as simple as you'd think:

https://bitcointalk.org/index.php?topic=99497.msg2733783#msg2733783

And this may be relevant for the question within that reference, asking why more transactions are not included.
https://gist.github.com/gavinandresen/5044482

Interesting points though i believe only point 2 could and should be used. Or has point 3 really a chance of being useful? There is no moral downside in using point 3 too then.

The biggest factor probably would be point 2. Not only because your found block would be propagated way faster but you would be informed faster about found blocks from others too. I think here lies the problem.

Yesterday i though it might depend on the getwork-proxy... is it still used for solomining? It made only sense using it for poolmining right?

If point 2 cant be used really good because of a slow internet connection... means propagating to many good nodes fast, then friedcat should think about getting some more connections. Not because the ping is too low but because the transferable data might be too low.

What i wonder is... can it be that a found block is proofed by the wallets after they checked it fully? I mean in case asicminer sends a big block of 500KB to a node then he downloads it, checks it and sees its valid. When after this block data started to download another 25kB block comes in then its downloaded faster and the proof is done faster too. What matters then? The time a block was proofed? If so i think the wallets should note a timestamp when a block was started to download. This way the network could be made more safe so that blocks include transactions.
Though this probably wont help much when the wallet only starts to propagate the big block after it proofed it fully. The propagation in total would be very slow then.

[SebastianJu]

Its correct that the advantage of lower process nodes arent so high in cost like it looks on the first view since a chip is way more expensive. But a big factor to keep in mind is that you save a lot money by assembly, you need less miner boards, you need less work to do for the same TH and you need less space. Not to forget the higher speed to bring up TH.
So as long as you dont plan to sell... for mining lower process nodes have some good advantages.

The NRE costs for all 28nm were payed buy customers preorder funds that were willing to dig their own grave.
Not an option for Asicminer.

I didnt speak about NRE-Costs at all. Only production costs. NRE-Costs are higher of course too and the higher the diff is rising the more impact has it on future chip production. Thats correct.

But isnt asicminer already developing 28nm? I though they developed a 40nm or 65nm chip that didnt work good and now work on 28nm. Or was it another process node?