[Coincraft or Minion 28nm] - Meeting for Group Buy Chip to Miner (DIY: THE WASP)

[Bicknellski]

Situation:

Designed from scratch, Bitmine’s Coincraft A1 is a third generation Bitcoin Mining IC developed by Bitmine in co-operation with a team of expert engineers from Innosilicon. Targeting the highest possible power efficiency, the A1 is made to be deployed in huge binary trees structures within large scale private or public mining pools.

Available in mass scale production starting from the last week of November 2013, the A1 employs cutting edge 28nm HPP technology from GlobalFoundries and ultra low voltage transistors, packaged into a custom engineered IC package that features our proprietary PowerBar technology for very low voltage, high current applications. Whether you want low power usage or high performances, the A1 fits perfectly both.

Our ASIC is the result of more than half a year of engineering and analysis of current systems limitations and predicted future of Bitcoin mining. That’s why we are confident it will be the best ASIC Bitcoin Miner design ever to be available on 28nm technology.

Detailed datasheet with the specifications of the package outline and the interface protocol will be released to buyers during the month of September.

https://bitmine.ch/?product=coincraft-ai-asic

Focus

Looking for hardware / software designers and miners willing to work together on a DIY Open Source Hardware / Software project that will develop a smaller 4 chip board that can turn 80 to 160 Gh/s for the group members and also the rest of those in the community who would be interested. Open Source would be released only after things are fully tested and working to the whole community. Production to take place upon delivery of the chips could take place at various fabricator worldwide.

Response

The group buy would be a start to finish development not just the chips.

Evalution

Miners would be shipped to members of the group buy after production is complete. Some members can do the fabrication as well and produce the miners. It would be up to each member of the Group Buy to decide on their preference for production.

[1714866694]

1714866694

[1714866694]

1714866694

[Bicknellski]

[RESERVED]

Code:

Product Description

Developed on 28nm HPP process from Global Foundries
Custom IC package with power bars for low voltage, high current feeding
Configurable in binary tree or daisy chain mode for distributed work with up to 4095 ASICs.
Standard I2C interface
Hashing power* of 20 GH/s in nominal and 40 GH/s in Turbo mode
Power usage* of 0.35 W/GH in low power, 0.6 W/GH in nominal and 1 W/GH in Turbo mode
Supply voltage* of 0.5V in low power, 0.65 V in nominal and 0.75 V in Turbo mode
Mass production available starting from the last week of November 2013

* Please note that while we took the highest possible care in doing accurate and conservative 
simulations, the real performances of the final product could vary by a +/- 20% tolerance.

How I can miss that
End of October. In this case, we have very limited time to design.

No, end of October is where the support material is guaranteed to be complete. This implies that it will be made available as open source as early as Bitmine has it - essentially DIY developers will have identical access to the relevant documents.

Since chip sales are part of their business, Bitmine has a vital interest to provide all available support to developers. That's why I am confident in this commitment enough to include the guarantee as part of the offer.

[Bicknellski]

Gotta have a mascot for a group buy like this right?

[darkfriend77]

Interesting ... what's the financial plan?

[Bicknellski]

Not definite on numbers that can be determined more accurately later but something like this has worked.

Suggested plan:

1. Purchase 500 chips = $50,000 (Depending on numbers of people in the group buy this could be a larger number of chips)
2. Prototype design boards and firmware etc = $5000 not certain but costs for design could be quite low per board and covers EE, SE and LEAD fab in finished chips / boards costs.
3. Members pay for the finished production boards with chips = 150$ or less would need to see the BOM / Gerbers to get the real costs.
4. Shipping out units from various fabrication points = cost of shipping / duties.
5. Gerbers and BOM released for everyone as open source hardware / software.

EE's putting in the design work etc is given percentage of chips / finished boards as sweat equity contribution.
SE's putting in the firmware / software work etc debugging work are given percentage of chips / finished boards as sweat equity contribution.
Lead on fabricators is given percentage of chips / finished boards as sweat equity contribution or slight mark up for time and effort to oversee production.

------

What would happen is people would be in a group buy that starts with a chip and ends with a shipped board this way all the costs are covered and shared proportionally by all the members in the Group Buy without the mark up that normal group buys have as everyone is responsible to share the load as it were depending on your initial chip investment. The EE's, SE's and Lead on Fabrication should also be invested to some degree in chips.

10 chips $1000
10 boards $50 / board
Prototype development costs $10 per board.
Shipping $50

----------------

10 boards $1700 (20 Gh/s x 10 = 200 Gh/s and potentially 400 Gh/s)


Note of course some people just want the DIY experience unpopulated boards and parts and chips can be shipped and you can go to town as well. And given the support from the EE people can send fubar boards back to them for repairs in the event of a "meltdown". Still others might want to do their own fabrication and get more chips and resell to the community so they do not need to pay for production form other members. With some minor alterations this is how the BitFury 1 USB made it to production as well from Big Picture Mining Cooperative it works very well as a model for production.

[darkfriend77]

This sounds like a interesting opportunity, to get ready for the future mining market ...

I would be willing to be part of it ... also for organizing stuff and such ... I'm from Switzerland, so might be useful for the communication or other stuff with bitmine.ch ... !

[Stack]

Could be a winner, watching.

[Bicknellski]

Feel free to PM me or post more questions.

Basically we are working on some concepts for the miner based on what we know of the chips from this point obviously early days but there seems to be some headroom and longer term ROI for this chip. Potential for alternative cooling as well. We will be working out more details on the plan but feedback here is also welcome so we can work out a fair equity for those doing the hard lifting for design and fab. We are hoping to make a design that will also be able to upcycle so that parts can be reconstituted should there be a new revised chip.

[nachius]

Definitely something to follow.  Looks like details are coming out this month on the chips pertaining to a data sheet if it hasn't already?

At those numbers (8.5$ per Ghash) and a quick swag at the term it looks like these would need to be ready to ship in around the next 60 days at $1700 per 200 Ghash to make it towards a ROI projection.  Looking at getting them out in November at 200 Ghash or possibly December if they can hit 400 GHash (4.25$ per Ghash) should be a target.  If that isn't realistic or at least close it may be a tough sell!

[bobsag3]

For such a community project, I will donate my hosting for the cost for power/cooling. No profit/parts, I would be honored to help out

[KS]

Crazy Canadian....

Good luck, hoping for some W1/W16 for Christmas!

[Bicknellski]

Definitely something to follow.  Looks like details are coming out this month on the chips pertaining to a data sheet if it hasn't already?

At those numbers (8.5$ per Ghash) and a quick swag at the term it looks like these would need to be ready to ship in around the next 60 days at $1700 per 200 Ghash to make it towards a ROI projection.  Looking at getting them out in November at 200 Ghash or possibly December if they can hit 400 GHash (4.25$ per Ghash) should be a target.  If that isn't realistic or at least close it may be a tough sell!

Pretty rough rough numbers so hard to really say yet.

The point of this is cutting out the "middlemen" to certain degree and going at this from a more collective perspective, so in terms of costs it be a lot more reasonable to do a DIYer or even the novice miner using this methodology, than say purchasing a similar ready made "off-the-shelf" unit from someone else and given that the group buy members are the owners of the project there won't be the same sort of lack of information for the members that can occur. I would hope also that we can look at this rig in some way to upcycle or reuse it for other purposes rather than just a dead miner. There needs to be some sort of way to give ASICs a secondary value if they become redundant as miners. Vanity Address Generating Engine? But yes we have to look very closely at the numbers what sort of target price do we need to make it viable?

[Bicknellski]

For such a community project, I will donate my hosting for the cost for power/cooling. No profit/parts, I would be honored to help out

Wow! That be great but I think we should value your contribution a bit more than that, so I  am sure a collective of members such as this would be willing to work out something that at least covers your costs and a bit to maintain and upkeep the systems? I am thinking cooling is going to be an important part of these newer units where a "turbo" mode doubles the hash.

[bobsag3]

For such a community project, I will donate my hosting for the cost for power/cooling. No profit/parts, I would be honored to help out

Wow! That be great but I think we should value your contribution a bit more than that so, I  am sure a collective of members such as this would be willing to work out something that at least covers your costs and a bit to maintain and upkeep the systems? I am thinking cooling is going to be an important part of these newer units where a "turbo" mode doubles the hash.

Assuming a dec-jan delivery, My cooling will provided by the 2 4 ft exhaust fans bringing in the 10f outside air .

[Bicknellski]

Wow.

That is definitely nice in the winter... come summer?

I am kinda keen on oil submersion have components priced out going and have space on the 3rd floor of my building for a test tank something that could hold 100 to 200 of these units and maintain constant temperature year round. PM me or email we can bounce some ideas back and forth.

------

Thinking about the "hive".

These units could be racked together so that you can expand a farm pretty quickly. Start with one and add more as you expand. 1 hive can be stacked on another hive for even greater density.

[nachius]

Definitely something to follow.  Looks like details are coming out this month on the chips pertaining to a data sheet if it hasn't already?

At those numbers (8.5$ per Ghash) and a quick swag at the term it looks like these would need to be ready to ship in around the next 60 days at $1700 per 200 Ghash to make it towards a ROI projection.  Looking at getting them out in November at 200 Ghash or possibly December if they can hit 400 GHash (4.25$ per Ghash) should be a target.  If that isn't realistic or at least close it may be a tough sell!

Pretty rough rough numbers so hard to really say yet.

The point of this is cutting out the "middlemen" to certain degree and going at this from a more collective perspective, so in terms of costs it be a lot more reasonable to do a DIYer or even the novice miner using this methodology, than say purchasing a similar ready made "off-the-shelf" unit from someone else and given that the group buy members are the owners of the project there won't be the same sort of lack of information for the members that can occur. I would hope also that we can look at this rig in some way to upcycle or reuse it for other purposes rather than just a dead miner. There needs to be some sort of way to give ASICs a secondary value if they become redundant as miners. Vanity Address Generating Engine? But yes we have to look very closely at the numbers what sort of target price do we need to make it viable?

I fully support the idea and believe in the community collective with detailed DIY service or some of the community members stepping up to provide a low cost assembly for those that can't or won't chance assembly.  As long as it is expandable as mentioned similar to the hive reference that would allow scalability.  I have not done enough research to even throw my hat in the ring as to "What else" ASICs can do to comment intelligently on that front.  As far as the price goes That's a probability and statistics professors dream problem.  There are inverse relationships that are interdependent on each other as you know:

Price
Time to market
Difficulty

It's a balancing act but a simple economics equation.  If they can be created and shipped while the projections show a probable ROI they will sell plain and simple.  If they are close to ROI, but can be expanded with expected dropping prices for each batch that then project ROI the original batch has a higher chance of selling than being a stand alone product.

At $1700 as I mentioned doing 200Ghash they would need to be out in November and at 400Ghash December.  That's using genesis block projections, but there are people that have more in depth projections as well beyond the readily available sites.  The price point is determined by the time and difficulty projections of that time unfortunately not the hardware costs themselves to assure they will sell!

I could envision a controller board that could either daisy chain other boards together or provide a backplane for hashing boards.  Yes this currently exists, but what I see is chip sockets that are similar to motherboard CPU slots insert chip snap down chip, heat sink, cooling and go.  Controller sees it configures and hashing begins!  This expandable board system and controller would allow for continued growth with changing technology smaller faster chips etc.  This then at that point is just a matter of device power consumption as long as each board released projects towards ROI and you could continue to use the original controller with firmware updates etc.  Just how I'm seeing it as instead of replacing devices as they get beyond useful life you can continue to use the same rig system just replacing or adding hashing boards indefinitely.  Maybe even once the additional use of ASICs is found there could be a trade in program or something similar.

That's a lot of tangents, but I think the power beyond ROI projections is in a future proof controller and expansion system if possible.

[thehun]

Following with interest 

[bobsag3]

Wow.

That is definitely nice in the winter... come summer?

I am kinda keen on oil submersion have components priced out going and have space on the 3rd floor of my building for a test tank something that could hold 100 to 200 of these units and maintain constant temperature year round. PM me or email we can bounce some ideas back and forth.

------

Thinking about the "hive".

These units could be racked together so that you can expand a farm pretty quickly. Start with one and add more as you expand. 1 hive can be stacked on another hive for even greater density.

I would totally do a submersion oil build. I have extra radiators and pumps from old Watercooling systems I can use to cool the oil already.

[Bicknellski]

Definitely something to follow.  Looks like details are coming out this month on the chips pertaining to a data sheet if it hasn't already?

At those numbers (8.5$ per Ghash) and a quick swag at the term it looks like these would need to be ready to ship in around the next 60 days at $1700 per 200 Ghash to make it towards a ROI projection.  Looking at getting them out in November at 200 Ghash or possibly December if they can hit 400 GHash (4.25$ per Ghash) should be a target.  If that isn't realistic or at least close it may be a tough sell!

Pretty rough rough numbers so hard to really say yet.

The point of this is cutting out the "middlemen" to certain degree and going at this from a more collective perspective, so in terms of costs it be a lot more reasonable to do a DIYer or even the novice miner using this methodology, than say purchasing a similar ready made "off-the-shelf" unit from someone else and given that the group buy members are the owners of the project there won't be the same sort of lack of information for the members that can occur. I would hope also that we can look at this rig in some way to upcycle or reuse it for other purposes rather than just a dead miner. There needs to be some sort of way to give ASICs a secondary value if they become redundant as miners. Vanity Address Generating Engine? But yes we have to look very closely at the numbers what sort of target price do we need to make it viable?

I fully support the idea and believe in the community collective with detailed DIY service or some of the community members stepping up to provide a low cost assembly for those that can't or won't chance assembly.  As long as it is expandable as mentioned similar to the hive reference that would allow scalability.  I have not done enough research to even throw my hat in the ring as to "What else" ASICs can do to comment intelligently on that front.  As far as the price goes That's a probability and statistics professors dream problem.  There are inverse relationships that are interdependent on each other as you know:

Price
Time to market
Difficulty

It's a balancing act but a simple economics equation.  If they can be created and shipped while the projections show a probable ROI they will sell plain and simple.  If they are close to ROI, but can be expanded with expected dropping prices for each batch that then project ROI the original batch has a higher chance of selling than being a stand alone product.

At $1700 as I mentioned doing 200Ghash they would need to be out in November and at 400Ghash December.  That's using genesis block projections, but there are people that have more in depth projections as well beyond the readily available sites.  The price point is determined by the time and difficulty projections of that time unfortunately not the hardware costs themselves to assure they will sell!

I could envision a controller board that could either daisy chain other boards together or provide a backplane for hashing boards.  Yes this currently exists, but what I see is chip sockets that are similar to motherboard CPU slots insert chip snap down chip, heat sink, cooling and go.  Controller sees it configures and hashing begins!  This expandable board system and controller would allow for continued growth with changing technology smaller faster chips etc.  This then at that point is just a matter of device power consumption as long as each board released projects towards ROI and you could continue to use the original controller with firmware updates etc.  Just how I'm seeing it as instead of replacing devices as they get beyond useful life you can continue to use the same rig system just replacing or adding hashing boards indefinitely.  Maybe even once the additional use of ASICs is found there could be a trade in program or something similar.

That's a lot of tangents, but I think the power beyond ROI projections is in a future proof controller and expansion system if possible.

Yes in total agreement. Now how to move this forward taking those excellent points in mind?

I think given the density that can be achieved a run of 125 of these 4 chip boards might be something to go for now and work concurrently on alternative uses for them firmware wise as they are being designed. Also a controller board is the right way to go I believe and makes these units modular but very manageable given they will not need to be all that large. Guessing something like 9x9cm. The other possibility is FPGAs could also be interchangeable in this system should scrypt coins miners become available soon.

[darkfriend77]

... we should form a group of interested person ... get a communication plattform like IRC ... and move this project forward ... .-)