Eyes to the horizon: A bitcoin ASIC project will be announced in June

[jayeeyee]

Leave it to the Chinese to make something in 27 days while BFL takes 300+ days to come out with whatever it is they came out with.  ..... sigh T_T

[1715308860]

1715308860

[1715308860]

1715308860

[1715308860]

1715308860

[Wayne_Chang]

Finally we see your thread here after your pre-announcement on Bitman.
I just can't wait to join the project.

[wrenchmonkey]

Ok fair enough, my mistake.  I just don't understand why they would go with 130 considering power would play such an important role in their ROI.  That just lead me to believe they are using the same fab and the chips are easier to acquire.  I mean, if you're going to make your own, why 130?  Cheap?  

Because it's not just twice as expensive to manufacture a chip on half the die size. It's ORDERS OF MAGNITUDE more expensive to get a die made in 65nm than to manufacture a die for 120-130nm.

Designing and producing a 130nm chip can be done for a few tens of thousands, or maybe low hundreds of thousand of dollars (depending on how motivated and talented the engineers are, and whether they're working on spec or for salary).

Going to 65nm is probably potentially half a million up to a million dollars (or more). Anybody who wants to get into the ASIC game is pretty much going to have to start in the 120-130nm die size range, and then build enough capital in order to fund research and development into the 65nm or smaller. If it was as easy as just jumping right on the latest die size, why didn't you suggest they got straight to 45nm die size?  

It's always entertaining when people who don't know anything about the industry purport to tell people like electrical engineers what's "current technology" and what isn't.

[ecliptic]

130nm process

hmm

Avalon is 110nm?
ASICMiner is 130nm?
BFL claims 65nm?

Is this right?

Could be the same chip as Avalon, but in a better package, with more VCC/GND connections and decoupling caps.  Avalon chips are hardly temp limited.  Putting it in a different package (lqfp64 for this vs QFN48 for Avalon), adding more power connections, and decreasing thermal resistance to ambient could very likely get 400MH/sec out of the same 130nm chip.  

Or perhaps they used the same public/published core design so it's basically identical regardless.  

Avalon chip buyers may very well be able to push their chips to mid 300MH/sec hashrate.

Your main problem is going to be this : Avalon is not closing orders on their chip sales until the end of 2013.  That leaves 7 months people can order chips and compete directly, toe-to-toe with you on mining.  Over 200 TH/sec have already been ordered from Avalon.  I would not be suprised to see many of the BTC produced by these miners to go back into Avalon, buying more chips, acquiring more hashing power.  Rather than trying to deal with the very ugly legal market of bitcoin mining -> USD (Mining = Money transmitter.  You have to get a license or it is illegal.  Nobody knows what will happen if you try to report your mining income of 5,6,7 figures per year and pay IRS tax on it.  They could arrest you.)

When avalon chips become to come online, expect ASICMiner shares to  go back to their true value

edit : I got avalon & AM node size mixed up i think.  this would be the same chip as ASICMiner then.  Is ASICMiner out of China as well?

[creativex]

Correct.

[ecliptic]

Also, companies like KnCMiner claim 28nm ASIC.  I recall another person (i cannot recall the name) mentioning they were in the works for creating 28nm asics, but would only sell to companies (like a real semiconductor company usually does..)

Also, Bitfury is 28nm i think?

[RoadStress]

Bitfury is 65nm, helveticoin claimed to have some 28nm available and kncminer wish they had 28nm

[Wayne_Chang]

Also, companies like KnCMiner claim 28nm ASIC.  I recall another person (i cannot recall the name) mentioning they were in the works for creating 28nm asics, but would only sell to companies (like a real semiconductor company usually does..)

Also, Bitfury is 28nm i think?

Base on TSMC's capacity as I know, it will take a long time for KnCMiner to wait in the queue of manufacture.
As we all know that Qualcomm, Nvidia and AMD are using 28um tech for their chips. TSMC definitely will feed these companies first.
Therefore, I think using 130um to start the game is wise.

[newtothescene]

Thanks for the announcement - very interested, leaning toward option A.

[dan99]

Butterfly wants to be perfect in making a 65nm with low power consumption but takes forever to get them produce, where as BTCGarden and others goes for 110nm or 130nm etc, so they can get up and running fast .. in China they probably work 24/7 to get their lines up and running. Low production costs + labor you win hands down

[aiwill]

So MODE B is just a hosted situation, meaning they don't ship you the unit correct?

correct

[aiwill]

CHINESE ARE DOMINATING ASIC MINING!!!

I hardly think so( although I dont think its bad if it comes true ), by far, BFL still occupys an absolute advantage percentage of market shares.

[aiwill]

can i ask how much the project cost you ?

Since our team is made up of associate/colleague/classmate ，all of who are fully experienced in this field, it is not a huge amount   . It is a little weird if I told you the exect number here,but once again, to compare with our former jobs (mainly on aerospace),it is easy.

[aiwill]

have you already arranged for the infrastructure to deploy your hash rate? the power and physical space needed for such a deployment is one of the biggest concerns for an effective rollout.

yes ,we have . and yes you are right,thanks for reminding.

[aiwill]

hi

are you planning to sell your hardware directly?

what is the hashing speed of your chip?

you know that 130 nm is an outdated tech and asic from bfl or else are working on 65 nm?

let me know

thanks

1. please re-read my thread.
2. same as above
3. it is not.    

[iikun]

CHINESE ARE DOMINATING ASIC MINING!!!

I hardly think so( although I dont think its bad if it comes true ), by far, BFL still occupys an absolute advantage percentage of market shares.

Umm...BFL certainly occupies an absolute advantage of pre-orders.  Market share, I dont think so as that would require actually shipping a product

[YipYip]

have you already arranged for the infrastructure to deploy your hash rate? the power and physical space needed for such a deployment is one of the biggest concerns for an effective rollout.

yes ,we have . and yes you are right,thanks for reminding.

Like WOW ..... bring it on like donkey kong !

[dan99]

hi

are you planning to sell your hardware directly?

what is the hashing speed of your chip?

you know that 130 nm is an outdated tech and asic from bfl or else are working on 65 nm?

let me know

thanks

1. please re-read my thread.
2. same as above
3. it is not.    

You should have make your thread more clear ... and if you could first answer some of the question on the 1st to 2rd  page
than it will be more clearer... and remember you only have 16 post, you may have thousand of postings in http://bbs.btcman.com or other
Chinese forum in china, but here is an international forum.

[wrenchmonkey]

hi

are you planning to sell your hardware directly?

what is the hashing speed of your chip?

you know that 130 nm is an outdated tech and asic from bfl or else are working on 65 nm?

let me know

thanks

1. please re-read my thread.
2. same as above
3. it is not.    

You should have make your thread more clear ... and if you could first answer some of the question on the 1st to 2rd  page
than it will be more clearer... and remember you only have 16 post, you may have thousand of postings in http://bbs.btcman.com or other
Chinese forum in china, but here is an international forum.

What questions do you have that weren't already answered in the original post? Everything that was asked has been answered, if people will but take the time to READ.

[Franktank]

Posted to follow...