Pages:
Author

Topic: [ANN] OpenBitASIC : The Open Source Bitcoin ASIC Initiative - page 14. (Read 50782 times)

legendary
Activity: 1099
Merit: 1000

Well that's fucked up, can't they take a "general" design and adapt it to their processes as part of the mask fee?

Yes and no. A synthesizable design, aka soft core can be ported relatively easy between different processes.  You can think of it as the source code of the chip which has to be compiled for a process.

But thats not the problem; what we call a process, consists of pre-made transistor designs and a whole set of design rules. Thats different for every process, and almost every fab. So you once you "compile" your design and generate a mask, its no longer portable and you will still need a specific mask for every process.

Since the cost of that mask is what is preventing ASICs from being available today, I dont really see what problem open sourcing a design would solve. It might be neat project for a hardware engineer, but who's going to pay the mask? And why wouldnt he use a less portable, but better optimized design? A soft core is never very optimized.


Many thanks for all the insights.
I understand that the mask cost will take a big share on the project. But what you call the soft core must carry a certain work, time and cost too. The idea is to make it public all the design processes and optimizations tasks, that are not part of the mask manufacturing itself. We will need to subcontract an "open" fab or fabless provider with no NDA issues.
Does it make any sense ?

rjk
sr. member
Activity: 448
Merit: 250
1ngldh

Well that's fucked up, can't they take a "general" design and adapt it to their processes as part of the mask fee?

Yes and no. A synthesizable design, aka soft core can be ported relatively easy between different processes.  You can think of it as the source code of the chip which has to be compiled for a process.

But thats not the problem; what we call a process, consists of pre-made transistor designs and a whole set of design rules. Thats different for every process, and almost every fab. So you once you "compile" your design and generate a mask, its no longer portable and you will still need a specific mask for every process.

Since the cost of that mask is what is preventing ASICs from being available today, I dont really see what problem open sourcing a design would solve. It might be neat project for a hardware engineer, but who's going to pay the mask? And why wouldnt he use a less portable, but better optimized design?

Well, I think the problem that was wanting to be solved was for everything to be open and accessible, but now that I understand the NDAs are to protect the fabs, and not so much for the people funding the devices, I see now that this may be unachievable. Too bad there are no current-generation "open source" foundries.
hero member
Activity: 518
Merit: 500

Well that's fucked up, can't they take a "general" design and adapt it to their processes as part of the mask fee?

Yes and no. A synthesizable design, aka soft core can be ported relatively easy between different processes.  You can think of it as the source code of the chip which has to be compiled for a process.

But thats not the problem; what we call a process, consists of pre-made transistor designs and a whole set of design rules. Thats different for every process, and almost every fab. So you once you "compile" your design and generate a mask, its no longer portable and you will still need a specific mask for every process.

Since the cost of that mask is what is preventing ASICs from being available today, I dont really see what problem open sourcing a design would solve. It might be neat project for a hardware engineer, but who's going to pay the mask? And why wouldnt he use a less portable, but better optimized design? A soft core is never very optimized.
rjk
sr. member
Activity: 448
Merit: 250
1ngldh
First, the design you make for a foundry cannot just simply be sent to another foundry. Every design is very foundry specific. In the fabless world everything is operated under NDAs. You cannot simply broadcast you design unless the foundry has checked that you didn't disclose any of the foundries classified information. You cannot even start to design without signing a NDA, because you cannot get the basic information to start with.

Well that's fucked up, can't they take a "general" design and adapt it to their processes as part of the mask fee?
newbie
Activity: 32
Merit: 0
First, the design you make for a foundry cannot just simply be sent to another foundry. Every design is very foundry specific. In the fabless world everything is operated under NDAs. You cannot simply broadcast you design unless the foundry has checked that you didn't disclose any of the foundries classified information. You cannot even start to design without signing a NDA, because you cannot get the basic information to start with.
sr. member
Activity: 325
Merit: 250
Our highest capital is the Confidence we build.
Nice, and really needed project. I hope it can be done.
legendary
Activity: 1099
Merit: 1000
what kind of open source license would the design most likely fall under, specifically what are your thoughts on commercial entities ability to use the design?

I'm thinking on GNU or MIT. Design must remain open and public, including any commercial entity interested in manufacturing.
We need ASIC's to spread around and avoid concentration.
legendary
Activity: 1099
Merit: 1000
I'm not really qualified to comment on chip design.  But this is a company I've known about for a little while that makes circuit design software.  For someone who is a circuit designer and who might be interested in this project, it might be worthwhile to contact them and try to get a license for non-profit use.

Who knows, they may be interested just for the publicity.  It could be very beneficial for them to show how their software can help trim costs and avoid errors with a bootstrappy open source project such as this one.  And obviously if an aspiring chip designer gets a successful project under his belt as well, then it's win-win.

That site is indexed in google, but seem down at this moment, from here at least.
legendary
Activity: 1099
Merit: 1000
You might have to make it a for-profit business venture.  I don't know that there are enough people willing to donate funds towards the project to make it happen otherwise.  But investors interested in a for-profit venture would likely be willing to invest far larger sums of money into the company with the potential for a return on the investment.

You are certainly right. It can be a for-profit, with a return to investors.
 
member
Activity: 84
Merit: 10
what kind of open source license would the design most likely fall under, specifically what are your thoughts on commercial entities ability to use the design?
legendary
Activity: 1330
Merit: 1000
I'm not really qualified to comment on chip design.  But this is a company I've known about for a little while that makes circuit design software.  For someone who is a circuit designer and who might be interested in this project, it might be worthwhile to contact them and try to get a license for non-profit use.

Who knows, they may be interested just for the publicity.  It could be very beneficial for them to show how their software can help trim costs and avoid errors with a bootstrappy open source project such as this one.  And obviously if an aspiring chip designer gets a successful project under his belt as well, then it's win-win.
legendary
Activity: 1400
Merit: 1005
You might have to make it a for-profit business venture.  I don't know that there are enough people willing to donate funds towards the project to make it happen otherwise.  But investors interested in a for-profit venture would likely be willing to invest far larger sums of money into the company with the potential for a return on the investment.
legendary
Activity: 1372
Merit: 1003
Just registered domain openbitasic.com for future use.
One thing I would like to hear opinions is if the company to be registered, must be a non-profit or a for-profit.

Non profit would be good yes people need good salary's and stuff but any potential profits could be ploughed back into the company or community.
legendary
Activity: 1099
Merit: 1000
Just registered domain openbitasic.com for future use.
One thing I would like to hear opinions is if the company to be registered, must be a non-profit or a for-profit.
legendary
Activity: 1099
Merit: 1000
I support this initiative.

One question though. Say an ASIC design is finalized and manufacturing commences. If there are changes to the hashing algorithm, how extensive must they be for the design and already manufactured units to be deprecated and not work with the modified algorithm?

Many thanks for all support, guys.

AFAIK, algorithm is hardcoded in the ASIC design, so any change in hashing algorithm (or protocol) will make them obsolete.
That's a great economic risk if this change happens to be mandatory, and backward compatibility cannot be accomplished.

Anyway, the product selling to a major audience, will assure that any change to protocol will be approved by miners if, and only if, it is absolutely necessary.   
sr. member
Activity: 242
Merit: 251
I support this initiative.

One question though. Say an ASIC design is finalized and manufacturing commences. If there are changes to the hashing algorithm, how extensive must they be for the design and already manufactured units to be deprecated and not work with the modified algorithm?
hero member
Activity: 714
Merit: 500
I like this idea!
rjk
sr. member
Activity: 448
Merit: 250
1ngldh
Very good. If we can get as far as decent designs, I'm sure that's when the money will flow in to create a product.
legendary
Activity: 1099
Merit: 1000
Frequently Asked Questions

1) What is this OpenBitAsic project?

The OpenBitAsic project is about producing an ASIC-based bitcoin miner. It is open because
the code and design will be published free under the terms of the GPL.

The OpenBitAsic miner will consist of a custom-designed structured ASIC based on Altera
Hardcopy technology. The ASIC will be mounted on a PC board along with other components
necessary to interface to a USB bus. Naturally, the protocol will be published allowing anyone
to write their own miner front-end for any operating system. We also hope to be compatible with
cgminer in time for the initial release of the hardware.

The miner will be sold as plug-and-play hardware with the PC board already mounted in an
enclosure with necessary cooling and power supply included.

2) What is the difference between OpenBitAsic and other ASIC projects presented
before ?


The OpenBitAsic project will be an open design. The hardware will be sold to all interested
parties. The company intends to make money by selling mining hardware, not by mining itself.

3) Can I use this product with any pool I choose?

Yes, there are no restrictions on the use of the hardware. You can use the product with any
compatible mining software (initially we hope to have support added to cgminer, but we will work
with other mining software authors if they would like to add support for our hardware as well).

4) Will it be a company behind this project?

Yes, a company is in process to be incorporated in Hong Kong. This company will manage all
aspects of the development, manufacturing, and sales of the ASIC bitcoin miner.

5) Is the company being operated as a non-profit?

We believe everyone deserves to be rewarded for their hard work, especially when it is of
benefit to the community. The company will be run on a for-profit basis.

6) What is the product that will emerge from this project ?

The device will be small (we have not selected an enclosure yet). Inside it will be a structured
ASIC (Altera Hardcopy) with an estimated hashing power of 8 GH/s (meaning we estimate the
device will be able to scan 8 billion Bitcoin nonce's per second).

Our preliminary worst-case power consumption estimates show around 75 watts of power
dissipation. However, we do not know how accurate the software that made our estimate is
(Quartus II), so we must emphasize that this is a preliminary spec and we cannot promise that it
will not be higher.

7) What is the timeline for this product to come to market ?

We expect the hardware will be available for sale in the fourth quarter of 2012.

8 ) Will it be possible to pre order ?

After prototype is built, tested and fine tuned, pre-sales will be opened to reserve a device.

legendary
Activity: 1099
Merit: 1000
This project aims to the design, manufacture and selling, of an open source ASIC for bitcoin mining.

Roadmap proposal :

1. Registering company in a business friendly and open jurisdiction, with near access to ASIC manufacturers (e.g. HK, Singapore). The company will be in charge of comissioning the manufacture and further selling of ASIC's.

2. Open calling to ASIC designers and ASIC design companies. All the design steps will be open sourced. Design will be peer reviewed and proposed improvements will be made before next step.

3. Manufacture testing, thru a small quantity manufacture. Testing and improving design.

4. Production manufacture, first round.

5. Selling to miners.

6. Improvements, changes, etc.

7. Further manufacture rounds.


UPDATE April 26 : Company registration in process in Hong Kong. Jason is working on the preliminary design.
  
Pages:
Jump to: