Pages:
Author

Topic: Bitfury: "16nm... sales to public start shortly" - page 41. (Read 108588 times)

member
Activity: 92
Merit: 10
I am under NDA .
So just take my word for it.


How much do you think a board would cost to run 20 chips per board?  Subtract the price of the BF chips so I am just talking about the board.

20 chips should run 2TH at .11w/gh so 220W per board.  How big would the board need to be? 10" X 10"?

Qty 500 boards.

I am just curious as to the cost of the board in qty. 

I don't have a ballpark price on hand but that's not too far from what we're looking to build. A pair of my boards would do 2TH/220W in approximately 6"x9" of PCB and screw onto an S1 heatsink.

If you make one of those 2T at  220w I will buy that for sure.
legendary
Activity: 1600
Merit: 1014
Between I got a mail from bitfury.



Have a look.

Thanks for making this available! It's unfortunate he doesn't indicate how many chips you get if you put up $1M.

Do you get 100, 1000, 500,000, what?

Nice way of saying "Don't talk to us unless you have serious money". Not a surprise.....

Punin himself stated exactly that in the russian part of the forum.
alh
legendary
Activity: 1846
Merit: 1052
Between I got a mail from bitfury.



Have a look.

Thanks for making this available! It's unfortunate he doesn't indicate how many chips you get if you put up $1M.

Do you get 100, 1000, 500,000, what?

Nice way of saying "Don't talk to us unless you have serious money". Not a surprise.....
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Anyone have an online class they recommend for designing chips?
You need to learn prerequisites first:

1) digital logic design (combinatorial and sequential logic, synchronous and asynchronous circuits, automata theory, etc.)

2) elements of circuit theory (nodal analysis, mesh analysis, you'll need to understand what SPICE will do for you and what it won't do)

Only after some understanding of the above you will be able to comprehend the CMOS ASIC design textbooks.

For somebody who is a self-starter the best way is to get some low end FPGA kit from e.g. Digilent or Avnet or Terasic. Low-end because those are supported by the free editions of the respective design toolchains. This is the main objective: to actually understand the methodology and limitations. When FPGA's were competitive for Bitcoin mining several members of this forum learned everything from scratch in about 3 months.

The classroom environment becomes important only later. You want to choose a school that has access to the https://www.mosis.com/ prototype fabrication service.

Bitfury actually did his first chip in partnership with some Polish research institute when I pointed to him that he will then qualify for http://www.europractice-ic.com/ which is an European equivalent of USA's MOSIS.

Thanks for the info and recommendations.

I teach telephone technicians how to use their meter to identify and locate troubles on wire line telephone service; not wireless.  I also teach transmission trouble shooting and some outside plant engineering design.  ADSL is asynchronous but I'm very limited on my knowledge of it.  My training focuses more on maintaining the wire and/or fiber [In the middle] connecting networks together rather than each end of the network with its equipment.  

Telephony training in the telephone industry has taken a hit lately with the economy being so terrible.  When the stocks of CenturyLink and AT&T take a hit, one of the first things to go in their budget cuts is training.  I used to be on the road 42 to 45 weeks annually.  Now, I'm fortunate to get 6 to 10 weeks of training classes annually after more than 18 years in training.  So, i turned to mining to supplement income I had lost from less training.  

Now, it looks like I might have to open up some e-cig stores to supplement my income from training.  I've been thinking about opening up six to ten e-ciggerette stores in my area.  I may end up going that route to make the funds.  It's quite possible to clear $5K+ monthly from each store after taxes and expenses.  A buddy of mine does this on average from each of his 14 stores scattered over about a 40 mile radius.

I'll take your advice to heart, weigh my options and decide what to do with my time.  Obtaining the funds is the biggest obstacle in my opinion.  I don't see where it profits me much to learn the stuff if I don't have the funds to use it.  I may end up continuing the way I'm going with buying the best gear I can get my hands on when it's available.  The block halving approaching doesn't help at the moment either.

So, more than likely, I'll go the e-cig route to obtain more funds to have more options when it comes to mining.  If a 0.06 watts per GH/s rig comes on the market, I'm quite certain it would be able to mine with profits for quite a while.  Maybe for the lifetime of the rig.  I hope someone comes out with one soon that's available to us all.  I'm not giving up on mining.  That's for sure.  I'm now in wait and see mode regarding mining.

Thanks again for your recommendations.
legendary
Activity: 2128
Merit: 1073
Anyone have an online class they recommend for designing chips?
You need to learn prerequisites first:

1) digital logic design (combinatorial and sequential logic, synchronous and asynchronous circuits, automata theory, etc.)

2) elements of circuit theory (nodal analysis, mesh analysis, you'll need to understand what SPICE will do for you and what it won't do)

Only after some understanding of the above you will be able to comprehend the CMOS ASIC design textbooks.

For somebody who is a self-starter the best way is to get some low end FPGA kit from e.g. Digilent or Avnet or Terasic. Low-end because those are supported by the free editions of the respective design toolchains. This is the main objective: to actually understand the methodology and limitations. When FPGA's were competitive for Bitcoin mining several members of this forum learned everything from scratch in about 3 months.

The classroom environment becomes important only later. You want to choose a school that has access to the https://www.mosis.com/ prototype fabrication service.

Bitfury actually did his first chip in partnership with some Polish research institute when I pointed to him that he will then qualify for http://www.europractice-ic.com/ which is an European equivalent of USA's MOSIS.


member
Activity: 99
Merit: 10
Between I got a mail from bitfury.

Have a look.


Good to see and there always has to be the possibility that a big purchaser can also be a Distributor. Much as I like the idea of an own design ASIC I feel buying chips has got to be the way froward.

Rich

You bet. Wink
hero member
Activity: 588
Merit: 500
Between I got a mail from bitfury.

Have a look.


Good to see and there always has to be the possibility that a big purchaser can also be a Distributor. Much as I like the idea of an own design ASIC I feel buying chips has got to be the way froward.

Rich
member
Activity: 99
Merit: 10
Between I got a mail from bitfury.



Have a look.
legendary
Activity: 2128
Merit: 1073
Interesting idea, but logistically this would be really challenging to do - although one way you COULD do it would be if you developed a miner, and didn't release the driver as open source, and basically built a special version of the cgminer/bfgminer that was locked to a particular pool - and just make the pool fees that fund reinvestment.  Then there's no bother trying to enforce it, and even if you sold the gear, the new owner would still be contributing to the 'cause' so to speak.  This would also solve the issues of predictable cashflow for the parent company...  Hell, the pool could even be public, and people could mine into 'equity' positions even using other gear.  Use the infrastructure that already exists...
Somewhat naïve, but at least doable. Essentially 21 Inc's mining chip works like that, but designed by somebody more experienced in DRM (digital rights management).
legendary
Activity: 980
Merit: 1001
aka "whocares"
I would love to be involved with getting an ASIC built, but I am realistic about the chance of being successful- I would say it is about the similar chance as finding a block with a Compac. 

There are so many other viable options for far less money and far less risk.  In my humble opinion if you are unable to speculate on the difficulty and network hash at the time you are done with the project then how would you know if you will not end up making an ASIC with no where to play.

I would suggest fund raising and getting about $250-500,000 and then approaching Bitfury or 21inc and buying chips.  Sure they asks for $1 million but if you wave $500,000 it will certainly catch some attention.  A lot of great things can be done with others chips and often better than the manufacturer.  Look at the S5 efficiency vs the Compac efficiency.

Cheers
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Perhaps it's time to start a new thread for this line of discussion.

I haven't said I won't learn how to do it. I'm just saying that's not in my skill set. Given a chip, I can build a miner around it. Literally every part of a miner except the silicon design (including power, controls, driver code, what have you) is something I already know I can do. But I'm not going to assume there isn't someone who already knows what he's doing and can do it better - and certainly in less time, which matters a heck of a lot in the bitcoin world. I intend someday on working on silicon design, and I already have some ideas what I'd try and make. But I'd prefer to start on something quite a bit less intense than something 16nm running 500MHz.

I totally agree with you, Sir.  I was simply trying to help with marketing and coordinating to try to get things moving with others who already have the know how.  Instead, I'm told by someone who appears to have the knowhow to do one of 3 things.   lol   Cheesy

At least I know NOW not to waste my time with them.  I would probably end up getting taken advantage of out of spite now.  I was simply trying to offer my time, energy and money to help them and others get something done.  Plain and simple...
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Anyone have an online class they recommend for designing chips?  If there are certain things I cannot learn unless I'm in a classroom environment, let me know and I won't waste my time trying to learn online but do my research of a college here locally in Birmingham to take specific classes needed for what I want to accomplish.  I don't need a degree; only the knowledge with some experience.

It sounds like an ASIC for bitcoin mining is relatively simple compared to other applications after reading this subject in other threads.  I just want to take the time to learn it myself.
legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
Perhaps it's time to start a new thread for this line of discussion.

I haven't said I won't learn how to do it. I'm just saying that's not in my skill set. Given a chip, I can build a miner around it. Literally every part of a miner except the silicon design (including power, controls, driver code, what have you) is something I already know I can do. But I'm not going to assume there isn't someone who already knows what he's doing and can do it better - and certainly in less time, which matters a heck of a lot in the bitcoin world. I intend someday on working on silicon design, and I already have some ideas what I'd try and make. But I'd prefer to start on something quite a bit less intense than something 16nm running 500MHz.
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
I still would like to know how much would be required to get a 16nm design to engineering samples.

same here.
All of it, I mean all that you can come up with. You have this big target painted over your faces: I want to buy Bitcoin turboencabulators, I'm ignorant of the technology, please rip me off.

Cypherdoc was exactly like you, he was ripped off on his Bitcoin turboencabulators by Hashfast, but actually made money by shilling and roping more naïfs.

There are three ways of getting something done:

1) learn how to do it and do it yourselves;
2) learn who can do it for you and for how much, then pay it;
3) prohibit your teenage children from doing it, and hope they will turn like all teenagers and do what's prohibited.

What I see in both of you is the strong desire to not learn anything about logic design and the semiconductor industry. Sidehack is one step ahead, because he is at least openly willing to admit that he won't do it.

You are like two guys who want to win competitive dance or acrobatic event, but don't want to get their asses off the lazy chair. Learn how to walk and tap your feet with the music before you sign up in a competitive event.

Fine, I can take that medicine.  Sounds like it's time to do research and homework.  Then try to come up with the funds one way or another.
legendary
Activity: 2128
Merit: 1073
I still would like to know how much would be required to get a 16nm design to engineering samples.

same here.
All of it, I mean all that you can come up with. You have this big target painted over your faces: I want to buy Bitcoin turboencabulators, I'm ignorant of the technology, please rip me off.

Cypherdoc was exactly like you, he was ripped off on his Bitcoin turboencabulators by Hashfast, but actually made money by shilling and roping more naïfs.

There are three ways of getting something done:

1) learn how to do it and do it yourselves;
2) learn who can do it for you and for how much, then pay it;
3) prohibit your teenage children from doing it, and hope they will turn like all teenagers and do what's prohibited.

What I see in both of you is the strong desire to not learn anything about logic design and the semiconductor industry. Sidehack is one step ahead, because he is at least openly willing to admit that he won't do it.

You are like two guys who want to win competitive dance or acrobatic event, but don't want to get their asses off the lazy chair. Learn how to walk and tap your feet with the music before you sign up in a competitive event.

Edit: anyone here remembers that guy who was claiming to develop ASIC simultaneously with BFL? I forgot his handle on this forum. He was hiring people for the assembly and shipping departments, rented the premises, but his RTL wasn't even through the simulation stage.

Edit2: All right, I've found it: it was cablepair. He successfully shipped ModMiner (somebody's else FPGA design) and failed to deliver bASIC, but returned at least some of the collected money. That was about 2012. Now we have 2016 and another experienced telephone technician is trying to start another fabless ASIC boutique.

Edit3: Maybe somebody remembers who was that Arab guy living in England with experience in fashion for Islamic women? He was also trying to do the same.

legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Interesting idea, but logistically this would be really challenging to do - although one way you COULD do it would be if you developed a miner, and didn't release the driver as open source, and basically built a special version of the cgminer/bfgminer that was locked to a particular pool - and just make the pool fees that fund reinvestment.  Then there's no bother trying to enforce it, and even if you sold the gear, the new owner would still be contributing to the 'cause' so to speak.  This would also solve the issues of predictable cashflow for the parent company...  Hell, the pool could even be public, and people could mine into 'equity' positions even using other gear.  Use the infrastructure that already exists...

I like that idea too.  I would make a nomination for CK/Kano Pool to be that pool and they would probably be wiling to develop that special version of cgminer just for their pool at no cost.  They would be getting paid for their contribution by having more rigs on their pool finding more blocks, etc...  Hell, I would even send them a contribution for their time and effort.
hero member
Activity: 687
Merit: 511
Also, we would need to create something where those who invest MUST continue to put in a certain percentage of their earnings from mining back into the fund each month.  For example, maybe 3 to 5 percent should go back into the fund.  If they invested $100, they need to put back $3 to $5 every month into the fund every month for continued product development and to continue manufacturing.  It may be that 10% would need to go back into the fund every month.  I'm simply trying to throw out a number to for those who purchased ownership of a rig to continue their ownership in the company.  They can sell their ownership to someone else if they like.  If someone gets behind on their monthly dues for product development and manufacturing, they can no longer expect to have access to future gear that is created by the company of the crowd fund.

Interesting idea, but logistically this would be really challenging to do - although one way you COULD do it would be if you developed a miner, and didn't release the driver as open source, and basically built a special version of the cgminer/bfgminer that was locked to a particular pool - and just make the pool fees that fund reinvestment.  Then there's no bother trying to enforce it, and even if you sold the gear, the new owner would still be contributing to the 'cause' so to speak.  This would also solve the issues of predictable cashflow for the parent company...  Hell, the pool could even be public, and people could mine into 'equity' positions even using other gear.  Use the infrastructure that already exists...
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Now, I'm only throwing ideas out for people to brainstorm.

We could make it where ALL investors are owners of the crowd fund.  The percentage one owns depends on the amount they invest.  Rules need to be set up so that NO ONE ever has 51% or higher ownership.

Also, we would need to create something where those who invest MUST continue to put in a certain percentage of their earnings from mining back into the fund each month.  For example, maybe 3 to 5 percent should go back into the fund.  If they invested $100, they need to put back $3 to $5 every month into the fund every month for continued product development and to continue manufacturing.  It may be that 10% would need to go back into the fund every month.  I'm simply trying to throw out a number to for those who purchased ownership of a rig to continue their ownership in the company.  They can sell their ownership to someone else if they like.  If someone gets behind on their monthly dues for product development and manufacturing, they can no longer expect to have access to future gear that is created by the company of the crowd fund.

I'm throwing out ideas to try to keep something going if we are able to start it.  I don't want it to simply be a one time thing.  I want it to keep going and grow.

This would also mean people would have to be chosen to run "support" if issues should arise.  There needs to be a place to house the gear before shipment; for support; for repairs; etc...  Quite a bit to consider.  I would like to get something like this going if many with the contacts and the knowhow can work together to try to get this done.

EDIT:  The minimum investment might have to be like $500 or $800.  I'm not sure.  We would have to determine that number.  I simply used $100 for easy math.  If we could create a 16nm rig with 10 TH/s at 0.06 watts per GH/s at $800 cost per rig, it would ROI relatively quickly at current price of bitcoin and difficulty.

So, we would basically have to come up with a ball park figure of how much a 8 to 10 TH/s rig would be at cost for development, parts and manufacturing.  Each individual would be responsible for shipping to their location, any customs charges, VAT, etc...

I'm basically saying the minimum investment would be the cost for one rig.  The individual who made that investment also owns part of the crowd fund AS LONG AS they continue paying monthly dues for continued product development and manufacturing.  It may be we could make it 20% of their initial investment.  If they want to invest more as time goes along to get more rigs, that is fine too.  I'm just brain storming trying to figure out how to do this.
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
I still would like to know how much would be required to get a 16nm design to engineering samples.

same here.
legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
I still would like to know how much would be required to get a 16nm design to engineering samples.
Pages:
Jump to: