Crowd funded ASIC miner device | Bitcointalksearch.org

lame.duck

legendary

Activity: 1270

Merit: 1000

Quote from: bitfreak91 on April 07, 2013, 11:44:36 AM

Quote from: lame.duck on April 03, 2013, 01:53:36 PM

Can you post a picture of pcb with the Virtex6 on it without the heatsink please. Frankly, i doubt it will work to have 25A @ 1V over a cable connection for a device with high speed switching. Also the 'power supply boards' look for me like some random surplus boards.

I will not post anymore photo.

I am not surprised.

Quote from: bitfreak91 on April 07, 2013, 11:44:36 AM

Believe what you want but I will give you technical details. The green board is capable of delivering 30A and because of those cable that you mention the voltage that reaches the FPGAs is 0.90-0.91 V (read using the JTAG). So because of those cables and high current voltage drops but even so the FPGAs work stable. Yes I know, the minimum voltage specified in the FPGA's documentation is 0.95V but they are still stable at 0.90V. About the high speed switching again there is no problem: the FPGA board has it's own capacitors for DC filtering. If you have EE background/knowledge I guess you understand all this.

What a lot of Buzzword bingo in the EE flavor.

What about those Power supply modules, i wonder where the high current connectors are? Have you ever seens the connectors on the mod miner? And the were only for less that the half current. Not to talk about the inductance of the power cable. And in combinations with the 2 high current switching devices it would give the funniest results rather than a stable power supply.

And then just for curiosity, how do you read the voltage level over jtag?

wtman

legendary

Activity: 1030

Merit: 1000

Quote from: bitfreak91 on April 07, 2013, 11:50:29 AM

Quote from: ProfMac on April 03, 2013, 10:58:32 AM

I'm trying to grok this.

Are there a total of 36 Virtex 6 devices in your 2 rig farm?

Is this 20 GH/s / 36 Virtex 6 = 500 MH/s per Virtex 6

Producing perhaps 1.5 BTC / day

You are right sir. To be more accurate each FPGA has two fully unrolled SHA256d cores working at 300MHz. That is 600MHash/s/FPGA and the total hashrate is nearing to 21.6MHash/s.

With the last diff increase I am getting around ~1.3BTC/day.

What did your 2 rig farm cost?

Also, mind telling which country you are based in? I am in India

You can pm me

bitfreak91

newbie

Activity: 18

Merit: 0

Quote from: ProfMac on April 03, 2013, 10:58:32 AM

I'm trying to grok this.

Are there a total of 36 Virtex 6 devices in your 2 rig farm?

Is this 20 GH/s / 36 Virtex 6 = 500 MH/s per Virtex 6

Producing perhaps 1.5 BTC / day

You are right sir. To be more accurate each FPGA has two fully unrolled SHA256d cores working at 300MHz. That is 600MHash/s/FPGA and the total hashrate is nearing to 21.6MHash/s.

With the last diff increase I am getting around ~1.3BTC/day.

bitfreak91

newbie

Activity: 18

Merit: 0

Quote from: lame.duck on April 03, 2013, 01:53:36 PM

Can you post a picture of pcb with the Virtex6 on it without the heatsink please. Frankly, i doubt it will work to have 25A @ 1V over a cable connection for a device with high speed switching. Also the 'power supply boards' look for me like some random surplus boards.

I will not post anymore photo. Believe what you want but I will give you technical details. The green board is capable of delivering 30A and because of those cable that you mention the voltage that reaches the FPGAs is 0.90-0.91 V (read using the JTAG). So because of those cables and high current voltage drops but even so the FPGAs work stable. Yes I know, the minimum voltage specified in the FPGA's documentation is 0.95V but they are still stable at 0.90V. About the high speed switching again there is no problem: the FPGA board has it's own capacitors for DC filtering. If you have EE background/knowledge I guess you understand all this.

lame.duck

legendary

Activity: 1270

Merit: 1000

Can you post a picture of pcb with the Virtex6 on it without the heatsink please. Frankly, i doubt it will work to have 25A @ 1V over a cable connection for a device with high speed switching. Also the 'power supply boards' look for me like some random surplus boards.

ProfMac

legendary

Activity: 1246

Merit: 1002

Quote from: bitfreak91 on March 31, 2013, 09:29:22 PM

@tbd: really I optimized the existing open source code to be able to squeeze two fully unrolled SHA256d cores in one big FPGA. About the ASIC: the chip is the expensive and the most time consuming part of the project and as I propose that the public fund the project I think is miners what they want and not chips.

@witherworth: there is no company. With me everything must start from zero as that is exactly what I have in terms of money. I know that the community lost money and hopes as many failed to keep their promises but I cannot do anything without your financial support. And also any technical support will be welcomed.

Here is one of the two rigs: under each heatsink there is one Virtex 6. The green PCB next to each heatsink is a power supply with 25A@1V for the FPGA core. One level below the top you can see a Digilent Nexys3 board. That board acts as a remote (over TCP) multiple serial port. Hope you enjoy.

I'm trying to grok this.

Are there a total of 36 Virtex 6 devices in your 2 rig farm?

Is this 20 GH/s / 36 Virtex 6 = 500 MH/s per Virtex 6

Producing perhaps 1.5 BTC / day

toffoo

sr. member

Activity: 408

Merit: 261

I think you could argue that ASICMINER was "crowdfunded", and successful, so there you go. Best of luck with the project.

mobodick

hero member

Activity: 840

Merit: 1000

Quote from: bitfreak91 on April 01, 2013, 11:06:44 AM

Quote from: mobodick on April 01, 2013, 10:33:13 AM

I can imagine that, since the SHA256 algorith used in bitcoin is pretty well defined, that pooping out RTLs won't be very hard. I think that the functional design cannot become much more optimized than it is now.
The only place where you will be able to compete is at the physical design stage.

You are right about that, physical optimization will bring such an ASIC to a better performance vs. actual miners. But that is true only when using the same node as the other manufacturer (I consider Avalon the only at this moment!).

In this project we can use smaller technology node, and as more people fund this project the better will be the ASIC.

You can have a lot of advantages over the current market offer: small price, miner performance, eliminate the possibility of the manufacturer mining in the backyard etc. What other proposal can be better than this when I have no money to do this project? OK, maybe another person with an ASIC background, but the funding scheme is good I think.

This is what i actually mean by physical design.
The only way to get better than competition is to win on the physical part, no matter the node.
Everyone will converge to current node technology so this win will decrease in the future.

In a year other ASICS will be on the market baked with smaller masks. You will need to make sure now that you will be on top then because otherwise your (or in this case, our) investment will go down the drain.

This is why i'm very sceptical about crowd funding someone who can't guarantee a certain product in a year.
This project looks way too risky unless you already have a complete plan on what you need to achieve in your product a year from now.

bitfreak91

newbie

Activity: 18

Merit: 0

Quote from: ShadesOfMarble on April 01, 2013, 09:26:12 AM

What about using existing chips? I.e. Avalon.

For the moment they are the best public solution in the bitcoin network. But they can be taken out of the game soon if they don't come up with a better solution.

Maybe this is a strategy of theirs: to start selling chips using old technology and then giving better solutions when the market is hungry again.

bitfreak91

newbie

Activity: 18

Merit: 0

Quote from: mobodick on April 01, 2013, 10:33:13 AM

I can imagine that, since the SHA256 algorith used in bitcoin is pretty well defined, that pooping out RTLs won't be very hard. I think that the functional design cannot become much more optimized than it is now.
The only place where you will be able to compete is at the physical design stage.

You are right about that, physical optimization will bring such an ASIC to a better performance vs. actual miners. But that is true only when using the same node as the other manufacturer (I consider Avalon the only at this moment!).

In this project we can use smaller technology node, and as more people fund this project the better will be the ASIC.

You can have a lot of advantages over the current market offer: small price, miner performance, eliminate the possibility of the manufacturer mining in the backyard etc. What other proposal can be better than this when I have no money to do this project? OK, maybe another person with an ASIC background, but the funding scheme is good I think.

aTg

legendary

Activity: 1358

Merit: 1000

You mean those PCBs work with Virtex6? Because we can not see a photo of the FPGA without heatsink and a photo of hashing on PC?

mobodick

hero member

Activity: 840

Merit: 1000

Quote from: bitfreak91 on April 01, 2013, 09:56:52 AM

Quote from: mobodick on April 01, 2013, 09:32:01 AM

Quote from: bitfreak91 on April 01, 2013, 09:00:24 AM

Quote from: mobodick on April 01, 2013, 07:41:10 AM

If your ASICS won't outcompete whatever is on the market by then noone will invest in you.
So, what can you say to assure us that you are even capable of designing an ASIC?
Playing around with FPGAs and designing boards is not going to be enough.

In my proposal the buyers/investors have powers to take decisions if the project should go on at any point, or if strategy should be changed (in case performance is not considered satisfactory).
There is somebody selling ASICs nowadays and he started from selling FPGA-based devices. I think it is a good point to start and you can add to that a lot of time that I can dedicate to this project and people like you giving advices and making decisions so that things finish good for everybody.

So now you need us for the advice as well as for the funding.
What do we need you for then?
Seems much more like your only goal is to capitalize on the crowd.

So let me ask you again.
What can you say to assure us that you are even capable of designing an ASIC?

I can establish this company and do a lot of technical stuff. I can prepare a bunch of RTL variants to be tested in an ASIC test run and from this select a finalist/a few finalists. I can design the system architecture. Of course there is a gap between FPGA and ASIC but there is also time as I proposed.

I can imagine that, since the SHA256 algorith used in bitcoin is pretty well defined, that pooping out RTLs won't be very hard. I think that the functional design cannot become much more optimized than it is now.
The only place where you will be able to compete is at the physical design stage.

bitfreak91

newbie

Activity: 18

Merit: 0

Quote from: mobodick on April 01, 2013, 09:32:01 AM

Quote from: bitfreak91 on April 01, 2013, 09:00:24 AM

Quote from: mobodick on April 01, 2013, 07:41:10 AM

If your ASICS won't outcompete whatever is on the market by then noone will invest in you.
So, what can you say to assure us that you are even capable of designing an ASIC?
Playing around with FPGAs and designing boards is not going to be enough.

In my proposal the buyers/investors have powers to take decisions if the project should go on at any point, or if strategy should be changed (in case performance is not considered satisfactory).
There is somebody selling ASICs nowadays and he started from selling FPGA-based devices. I think it is a good point to start and you can add to that a lot of time that I can dedicate to this project and people like you giving advices and making decisions so that things finish good for everybody.

So now you need us for the advice as well as for the funding.
What do we need you for then?
Seems much more like your only goal is to capitalize on the crowd.

So let me ask you again.
What can you say to assure us that you are even capable of designing an ASIC?

I can establish this company and do a lot of technical stuff. I can prepare a bunch of RTL variants to be tested in an ASIC test run and from this select a finalist/a few finalists. I can design the system architecture. Of course there is a gap between FPGA and ASIC but there is also time as I proposed.

mobodick

hero member

Activity: 840

Merit: 1000

Quote from: bitfreak91 on April 01, 2013, 09:00:24 AM

Quote from: mobodick on April 01, 2013, 07:41:10 AM

If your ASICS won't outcompete whatever is on the market by then noone will invest in you.
So, what can you say to assure us that you are even capable of designing an ASIC?
Playing around with FPGAs and designing boards is not going to be enough.

In my proposal the buyers/investors have powers to take decisions if the project should go on at any point, or if strategy should be changed (in case performance is not considered satisfactory).
There is somebody selling ASICs nowadays and he started from selling FPGA-based devices. I think it is a good point to start and you can add to that a lot of time that I can dedicate to this project and people like you giving advices and making decisions so that things finish good for everybody.

So now you need us for the advice as well as for the funding.
What do we need you for then?
Seems much more like your only goal is to capitalize on the crowd.

So let me ask you again.
What can you say to assure us that you are even capable of designing an ASIC?

ShadesOfMarble

donator

Activity: 543

Merit: 500

What about using existing chips? I.e. Avalon.

bitfreak91

newbie

Activity: 18

Merit: 0

Quote from: mobodick on April 01, 2013, 07:41:10 AM

If your ASICS won't outcompete whatever is on the market by then noone will invest in you.
So, what can you say to assure us that you are even capable of designing an ASIC?
Playing around with FPGAs and designing boards is not going to be enough.

In my proposal the buyers/investors have powers to take decisions if the project should go on at any point, or if strategy should be changed (in case performance is not considered satisfactory).
There is somebody selling ASICs nowadays and he started from selling FPGA-based devices. I think it is a good point to start and you can add to that a lot of time that I can dedicate to this project and people like you giving advices and making decisions so that things finish good for everybody.

papamoi

sr. member

Activity: 378

Merit: 250

Quote from: bitfreak91 on April 01, 2013, 06:28:34 AM

Quote from: tbd on March 31, 2013, 10:43:06 PM

Ultimately, yes, people would like devices and not chips. However, there have been some challenges and many apparent scams of companies promising to deliver said devices, so the community is understandably wary. I think getting funding would be difficult.

Another option would be to get funding from a smaller number of investors/partners for the chip development. There's clearly a demand for chips, both for a small ASIC device (https://bitcointalksearch.org/topic/diy-pcb-with-avalon-the-quarter-stick-needs-help-161715) and others interested in bulk purchases (https://bitcointalksearch.org/topic/companies-who-have-asic-bitcoin-chips-and-who-are-willing-to-sell-162613, https://bitcointalksearch.org/topic/avalon-chip-orders-159257). You can consider me part of this group.

I see people are desperate to buy these chips/miners these days. It's because we are in the days where difficulty is relatively low. As I proposed there is one year needed for this project to have a consumer device and by that time difficulty will be very high. Unlike ASIC-based miner manufacturers who forget to mention that difficulty goes very high when a lot of ASIC miners enter the bitcoin network, I am warning everybody about this fact.

Maybe from this point of view the perfect investor/buyer would be the category of people who own a moderate to high amount of BTC: they will initially fund this project and this will attract a lot of more people who are not a part of this category. This way more and more people will have access to ASIC based miner and the BTC will gain strength. Also the competition would learn a lesson (hopefully ASIC monopoly/oligopoly would go down) and stop selling their ASIC-based miner at these prohibitive prices.

if asic itself is time consuming,maybe there will be some way to do something in between

i ve pmed you

mobodick

hero member

Activity: 840

Merit: 1000

Quote from: bitfreak91 on April 01, 2013, 06:28:34 AM

Quote from: tbd on March 31, 2013, 10:43:06 PM

Ultimately, yes, people would like devices and not chips. However, there have been some challenges and many apparent scams of companies promising to deliver said devices, so the community is understandably wary. I think getting funding would be difficult.

Another option would be to get funding from a smaller number of investors/partners for the chip development. There's clearly a demand for chips, both for a small ASIC device (https://bitcointalksearch.org/topic/diy-pcb-with-avalon-the-quarter-stick-needs-help-161715) and others interested in bulk purchases (https://bitcointalksearch.org/topic/companies-who-have-asic-bitcoin-chips-and-who-are-willing-to-sell-162613, https://bitcointalksearch.org/topic/avalon-chip-orders-159257). You can consider me part of this group.

I see people are desperate to buy these chips/miners these days. It's because we are in the days where difficulty is relatively low. As I proposed there is one year needed for this project to have a consumer device and by that time difficulty will be very high. Unlike ASIC-based miner manufacturers who forget to mention that difficulty goes very high when a lot of ASIC miners enter the bitcoin network, I am warning everybody about this fact.

Maybe from this point of view the perfect investor/buyer would be the category of people who own a moderate to high amount of BTC: they will initially fund this project and this will attract a lot of more people who are not a part of this category. This way more and more people will have access to ASIC based miner and the BTC will gain strength. Also the competition would learn a lesson (hopefully ASIC monopoly/oligopoly would go down) and stop selling their ASIC-based miner at these prohibitive prices.

If your ASICS won't outcompete whatever is on the market by then noone will invest in you.
So, what can you say to assure us that you are even capable of designing an ASIC?
Playing around with FPGAs and designing boards is not going to be enough.

bitfreak91

newbie

Activity: 18

Merit: 0

Quote from: tbd on March 31, 2013, 10:43:06 PM

Ultimately, yes, people would like devices and not chips. However, there have been some challenges and many apparent scams of companies promising to deliver said devices, so the community is understandably wary. I think getting funding would be difficult.

Another option would be to get funding from a smaller number of investors/partners for the chip development. There's clearly a demand for chips, both for a small ASIC device (https://bitcointalksearch.org/topic/diy-pcb-with-avalon-the-quarter-stick-needs-help-161715) and others interested in bulk purchases (https://bitcointalksearch.org/topic/companies-who-have-asic-bitcoin-chips-and-who-are-willing-to-sell-162613, https://bitcointalksearch.org/topic/avalon-chip-orders-159257). You can consider me part of this group.

I see people are desperate to buy these chips/miners these days. It's because we are in the days where difficulty is relatively low. As I proposed there is one year needed for this project to have a consumer device and by that time difficulty will be very high. Unlike ASIC-based miner manufacturers who forget to mention that difficulty goes very high when a lot of ASIC miners enter the bitcoin network, I am warning everybody about this fact.

Maybe from this point of view the perfect investor/buyer would be the category of people who own a moderate to high amount of BTC: they will initially fund this project and this will attract a lot of more people who are not a part of this category. This way more and more people will have access to ASIC based miner and the BTC will gain strength. Also the competition would learn a lesson (hopefully ASIC monopoly/oligopoly would go down) and stop selling their ASIC-based miner at these prohibitive prices.

tbd

newbie

Activity: 45

Merit: 0

Ultimately, yes, people would like devices and not chips. However, there have been some challenges and many apparent scams of companies promising to deliver said devices, so the community is understandably wary. I think getting funding would be difficult.

Another option would be to get funding from a smaller number of investors/partners for the chip development. There's clearly a demand for chips, both for a small ASIC device (https://bitcointalksearch.org/topic/diy-pcb-with-avalon-the-quarter-stick-needs-help-161715) and others interested in bulk purchases (https://bitcointalksearch.org/topic/companies-who-have-asic-bitcoin-chips-and-who-are-willing-to-sell-162613, https://bitcointalksearch.org/topic/avalon-chip-orders-159257). You can consider me part of this group.

Topic: Crowd funded ASIC miner device (Read 4895 times)