[CLOSED] Bitmine CoinCraft A1 28nm chip distribution / DIY support

[Lucko]

Hi there!

We would be interested in making a board. Is there a chance to get samples and any type of unfinished documentation on chips ASAP?

Anyway we do not plan just to make a clone of what will be in documentation but add something to the product...

Let me know...

[giorgiomassa]

Hi there!

We would be interested in making a board. Is there a chance to get samples and any type of unfinished documentation on chips ASAP?

Anyway we do not plan just to make a clone of what will be in documentation but add something to the product...

Let me know...

That's what we're encouraging the users to do, there will be a best board design contest where the best 3 designs will be awarded with very interesting prizes!

Regarding the samples, I'm afraid there won't be any because we managed to pull out from the first pilot run of the foundry all the needed chips we need for the units and the chip sales of December, so you're basically getting them as soon as we get them too.

[Lucko]

That's what we're encouraging the users to do, there will be a best board design contest where the best 3 designs will be awarded with very interesting prizes!

Regarding the samples, I'm afraid there won't be any because we managed to pull out from the first pilot run of the foundry all the needed chips we need for the units and the chip sales of December, so you're basically getting them as soon as we get them too.

OK what about unfinished documentation you my have already... So we get the idea what we are dealing with...

So we need to buy chips to get samples right? What about possibility of smaller then 100 chips punches so we can buy them sooner for development?

Let say that condition would be that you need to show something you already did with any chips and you can buy less...

Thanks

[bitcoiner49er]

Following.

Looks like a good chip might be worth a DIY open source hardware project.

That will result in a  when pricing is announced.

Every new ASIC will end this way. Because no one wants to pre-order anymore (for good reason), ASIC companies will maybe give some estimate of cost. Then when the real product comes to life, the price proves to be too much.
Welcome to A-sick-bitcoin world. 

[zefir]

So we need to buy chips to get samples right? What about possibility of smaller then 100 chips punches so we can buy them sooner for development?
[...]

The minimum quantity for this offer is 50 chips, and it was chosen based on my experience with other chip distributions where I realized that serving too many small customers kills my ability to ensure the service I would like to provide. A lot of 50 chips was therefore my compromise for offering an interesting entry volume for DIY folks and serious miners, as well as being a well sized offer for manufacturers of single chip devices.

Alas, for designers I understand that it is not ideal, since they do not want to order 50 chips just to see if the chip is doing as proposed. Therefore I am evaluating an additional offer dedicated to support you developers out there, without limiting my ability to handle everything well. The problematic part with low-volume orders is the shipping and handling component, which basically is identical whether someone orders 50 or 2 chips. The 8% surplus I put on top of Bitmine's prices are covering those expenses, but won't do so if applied to very small orders.

One option I am thinking of is to add a low-volume flat fee for smaller orders of say 1 BTC for quantities below the lot size, to compensate the S&H expenses. I understand that this would cause a huge cost overhead if buying a single chip. But assuming the reasonable number of chips required to test a design to be around 4 (if you want to test the binary-tree communication you need 3 or 7) it is maybe quite acceptable. As an example, 4 chips would be charged 4.4 BTC (3.4 for the chips plus 1 BTC low-volume fee).


What do you think, is this fair and acceptable?

[topminingcontracts]

Sorry, Stupid question but....


Price for 50 chips batch?

TMC

So we need to buy chips to get samples right? What about possibility of smaller then 100 chips punches so we can buy them sooner for development?
[...]

The minimum quantity for this offer is 50 chips, and it was chosen based on my experience with other chip distributions where I realized that serving too many small customers kills my ability to ensure the service I would like to provide. A lot of 50 chips was therefore my compromise for offering an interesting entry volume for DIY folks and serious miners, as well as being a well sized offer for manufacturers of single chip devices.

Alas, for designers I understand that it is not ideal, since they do not want to order 50 chips just to see if the chip is doing as proposed. Therefore I am evaluating an additional offer dedicated to support you developers out there, without limiting my ability to handle everything well. The problematic part with low-volume orders is the shipping and handling component, which basically is identical whether someone orders 50 or 2 chips. The 8% surplus I put on top of Bitmine's prices are covering those expenses, but won't do so if applied to very small orders.

One option I am thinking of is to add a low-volume flat fee for smaller orders of say 1 BTC for quantities below the lot size, to compensate the S&H expenses. I understand that this would cause a huge cost overhead if buying a single chip. But assuming the reasonable number of chips required to test a design to be around 4 (if you want to test the binary-tree communication you need 3 or 7) it is maybe quite acceptable. As an example, 4 chips would be charged 4.4 BTC (3.4 for the chips plus 1 BTC low-volume fee).


What do you think, is this fair and acceptable?

[RoadStress]

Sorry, Stupid question but....


Price for 50 chips batch?

TMC

It's in the first post of this topic. I will help you.

With today's rate, the price is 42.5BTC per 50-chip lot (1 THps).

[marto74]

So we need to buy chips to get samples right? What about possibility of smaller then 100 chips punches so we can buy them sooner for development?
[...]

The minimum quantity for this offer is 50 chips, and it was chosen based on my experience with other chip distributions where I realized that serving too many small customers kills my ability to ensure the service I would like to provide. A lot of 50 chips was therefore my compromise for offering an interesting entry volume for DIY folks and serious miners, as well as being a well sized offer for manufacturers of single chip devices.

Alas, for designers I understand that it is not ideal, since they do not want to order 50 chips just to see if the chip is doing as proposed. Therefore I am evaluating an additional offer dedicated to support you developers out there, without limiting my ability to handle everything well. The problematic part with low-volume orders is the shipping and handling component, which basically is identical whether someone orders 50 or 2 chips. The 8% surplus I put on top of Bitmine's prices are covering those expenses, but won't do so if applied to very small orders.

One option I am thinking of is to add a low-volume flat fee for smaller orders of say 1 BTC for quantities below the lot size, to compensate the S&H expenses. I understand that this would cause a huge cost overhead if buying a single chip. But assuming the reasonable number of chips required to test a design to be around 4 (if you want to test the binary-tree communication you need 3 or 7) it is maybe quite acceptable. As an example, 4 chips would be charged 4.4 BTC (3.4 for the chips plus 1 BTC low-volume fee).


What do you think, is this fair and acceptable?

You have an order for this , if you decide to do it

[Lucko]

The minimum quantity for this offer is 50 chips, and it was chosen based on my experience with other chip distributions where I realized that serving too many small customers kills my ability to ensure the service I would like to provide. A lot of 50 chips was therefore my compromise for offering an interesting entry volume for DIY folks and serious miners, as well as being a well sized offer for manufacturers of single chip devices.

Alas, for designers I understand that it is not ideal, since they do not want to order 50 chips just to see if the chip is doing as proposed. Therefore I am evaluating an additional offer dedicated to support you developers out there, without limiting my ability to handle everything well. The problematic part with low-volume orders is the shipping and handling component, which basically is identical whether someone orders 50 or 2 chips. The 8% surplus I put on top of Bitmine's prices are covering those expenses, but won't do so if applied to very small orders.

One option I am thinking of is to add a low-volume flat fee for smaller orders of say 1 BTC for quantities below the lot size, to compensate the S&H expenses. I understand that this would cause a huge cost overhead if buying a single chip. But assuming the reasonable number of chips required to test a design to be around 4 (if you want to test the binary-tree communication you need 3 or 7) it is maybe quite acceptable. As an example, 4 chips would be charged 4.4 BTC (3.4 for the chips plus 1 BTC low-volume fee).


What do you think, is this fair and acceptable?

You have an order for this , if you decide to do it

I think two... But I can't say how many chips since at this moment it would be shout in the dark... It looks like you already have some documentation since you are talking about binary tree communication

[Xian01]

Pencil me in for 50 chips and assembly costs with a community project.

[zefir]

I think two... But I can't say how many chips since at this moment it would be shout in the dark... It looks like you already have some documentation since you are talking about binary tree communication

No, really there is no documentation yet available. The binary tree communication protocol is described in the technical specs overview section on the Bitmine webpage I linked in the OP:

Tree or daisy chain configuration ready

Larger scale designs can see their performances degraded by the propagation time needed for a new Bitcoin job to walk through the daisy chain of chips till the last IC, wasting hundreds or even thousands of clock cycles. The A1 addresses this issue in a completely new way by using a binary tree structure where each chip propagates the work to two children nodes in the tree, reducing the propagation time to just 8 clock cycles for a tree of 511 ICs! The A1 can also be configured to work in the traditional daisy chain mode where smaller arrays or simpler design is required.

[zefir]

Update: Special low-quantity Conditions for DIY Designers added

To meet the specific demands of DIY folks and board designers, the following conditions have been added to the OP:

Sample Chips
The chips offered here are from the pilot-run and are exactly meant to be sample-chips, i.e. they are the first available ones. Obviously ordering a full lot of 50 chips only to test board designs is a barrier for DIY hobbyists. Therefore the following special conditions to support the DIY scene are added:

1) chips for design verification can be ordered in smaller quantities (1-31 chips, 0.85 BTC / chip)
2) an additional low-quantity fee of 1 BTC per order is charged to make up for S&H expenses
3) down-payment is at least 1 BTC
4) all other rules stated above for lot orders apply

[Lucko]

Just a question. What will be time difference between "samples" and real chips?

And did you change minimal order to 50? Or is this a mistake...

So for lack of a batter name at this moment Cointamination 2.0 is coming

[zefir]

Just a question. What will be time difference between "samples" and real chips?

And did you change minimal order to 50? Or is this a mistake...

So for lack of a batter name at this moment Cointamination 2.0 is coming

Have you read the post above yours? It was also added to the OP and answers both your questions.

[Lucko]

Forgive me I'm dyslexic. I was all ready to tell you that you are saying 100

Availability
I have 100 lots available (5000 chips). Chips are distributed in the order down-payment is paid (see below).

But I can see timing. OK pilot run but not sure if this is different. You are saying you are getting first batch. Is that pilot run? If not what is the time difference. Unless you are saying it all be at the same time. I would suggest to add that this orders will be send out first but then again we are probably talking about max day difference... And that will not make much of a difference...

Would it be possible to get FPGA program to test ahead of time? If not I can't see boards made at same time as chips came... And it is too risky to make them ahead of time...

So best case scenario:
Low numbers of test boards ready at the time chips arrive
2 days for test
10 days for boards(hi cost rush-job) if all checks out in tests
or
20 days for boards(normal cost) if all checks out...
1 day testrun
1 day shipping

So it would be great if we can get something to test ahead of time.

[intron]

Is there a datasheet available already? Just a
pinout and package details would also be helpful.

intron

[GandalfG]

zefir, any update for chip delivery date, specification ?

[Bicknellski]

Update: Special low-quantity Conditions for DIY Designers added

To meet the specific demands of DIY folks and board designers, the following conditions have been added to the OP:

Sample Chips
The chips offered here are from the pilot-run and are exactly meant to be sample-chips, i.e. they are the first available ones. Obviously ordering a full lot of 50 chips only to test board designs is a barrier for DIY hobbyists. Therefore the following special conditions to support the DIY scene are added:

1) chips for design verification can be ordered in smaller quantities (1-31 chips, 0.85 BTC / chip)
2) an additional low-quantity fee of 1 BTC per order is charged to make up for S&H expenses
3) down-payment is at least 1 BTC
4) all other rules stated above for lot orders apply

That is excellent news. Thanks for this concession.

[Bicknellski]

Is the 19th still a hard deadline?

[zefir]

Is the 19th still a hard deadline?

The deadline was set to leave me enough time to negotiate assembly contracts for the unsold chips. After clarifying with Bitmine, this can be done quite ad-hoc and does not need to be decided beforehand. Therefore, the deadline is extended to past the delivery date (OP was modified accordingly).

As for the documentation: I am currently reviewing the datasheet draft for the A1, which will be made available to the public before October 15th.