Stratix IV 180k - Possibilities for Mining - 25BTC bounty plus tips

[yjacket]

Since i don't know bunch about FPGAs and I know that BFL uses Stratix III 150K's in their singles, i was wondering if anyone could could provide some numbers as to what a Stratix IV 180k could do in terms of MH/s.

I may have a way of getting some on the cheap, and I'm wondering if you guys had any ideas.

Edit-

Maybe this will work better:

I'm pledging 25 BTC(I will increase if necessary) for a board design for the Stratix IV 180k featuring usb and capable of mining(i assume all this really means is supplying adequate power). Maybe this will be the seed of open source Stratrix-based mining.

The 25 BTC goes to the person that delivers the complete design. This design can be posted in the forum, I will not claim ownership. If they have help, the onus will be on then to distribute the BTC fairly. I don't think I will be able to manage dividing it between people fairly unless they all agree on the split.

I will also deposit tips(.5 - 3btc) from a separate pot for insightful posts.

I'm 100% serious about all of this. These stratix chips really seem like beasts considering how much BFL got out of them.


If anyone has a better idea of how to crowd-source this, please let me know. I'm open towards structuring it in a different way.

[yjacket]

No one has a guess? No one wants to think about it out loud?

[rjk]

No one has a guess? No one wants to think about it out loud?

Maybe get a few and see if you can get a bitstream running on them? It sounds interesting, but I doubt anyone has developed any firmware for them.

[shad]

just a guess, but i think it should be possible to achive the same speed as on an stratix III 150k with less power consumption

stratix III is 65nm
stratix IV is 40nm

also you have more space, it could be possible to fit an additional core into it

but the BFL is not open source afaik
so there is development needed

short: 400mhs maybe

EP3SL150 142.5K
EP4SGX180 175,750
23% more LEs

[yjacket]

No one has a guess? No one wants to think about it out loud?

Maybe get a few and see if you can get a bitstream running on them? It sounds interesting, but I doubt anyone has developed any firmware for them.

Although it's a good price it's still a large amount of money. I have no experience in board design or even in loading a bitstream so that makes it all the more difficult.

FYI these are used so they will have to be reballed as well.

[yjacket]

just a guess, but i think it should be possible to achive the same speed as on an stratix III 150k with less power consumption

stratix III is 65nm
stratix IV is 40nm

also you have more space, it could be possible to fit an additional core into it

but the BFL is not open source afaik
so there is development needed

short: 400mhs maybe

EP3SL150 142.5K
EP4SGX180 175,750
23% more LEs

That's what i'm thinking. Also, wouldn't it run faster as well?

[yjacket]

Maybe this will work better:

I'm pledging 25 BTC(I will increase if necessary) for a board design for the Stratix IV 180k featuring usb and capable of mining(i assume all this really means is supplying adequate power). Maybe this will be the seed of open source Stratrix-based mining.

The 25 BTC goes to the person that delivers the complete design. This design can be posted in the forum, I will not claim ownership. If they have help, the onus will be on then to distribute the BTC fairly. I don't think I will be able to manage dividing it between people fairly unless they all agree on the split.

I will also deposit tips(.5 - 3btc) from a separate pot for insightful posts.

I'm 100% serious about all of this. These stratix chips really seem like beasts considering how much BFL got out of them.


If anyone has a better idea of how to crowd-source this, please let me know. I'm open towards structuring it in a different way.

Edit: Moved to title post

[ice_chill]

Seeing how many FPGA companies are popping up, I think someone would do this already if they thought it was profitable and possible.

[lame.duck]

How many FPGAs are 'some'?

While you could produce a pcb artwork using open source tools you probably will produce one or two boards ($100 each) for testing before producing a (small) series which is  quite costly. Then you need a ful version of quartus(42500) to produce the bitstream. Then you need someone whos assemles the boards for you which isn't (as far i know) so trivial due to high ball count and the 'heavy' metal heat spreader.

[yjacket]

How many FPGAs are 'some'?

While you could produce a pcb artwork using open source tools you probably will produce one or two boards ($100 each) for testing before producing a (small) series which is  quite costly. Then you need a ful version of quartus(42500) to produce the bitstream. Then you need someone whos assemles the boards for you which isn't (as far i know) so trivial due to high ball count and the 'heavy' metal heat spreader.

5 of them. This may have been a mistake, but that's alright. What I really need is a PCB design. I've read all the specs for these and I really don't have the technical capabilities to design the PCB. I just just don't know what a platform for mining really requires with regards to components. Can someone help me out with that?

[rjk]

How many FPGAs are 'some'?

While you could produce a pcb artwork using open source tools you probably will produce one or two boards ($100 each) for testing before producing a (small) series which is  quite costly. Then you need a ful version of quartus(42500) to produce the bitstream. Then you need someone whos assemles the boards for you which isn't (as far i know) so trivial due to high ball count and the 'heavy' metal heat spreader.

5 of them. This may have been a mistake, but that's alright. What I really need is a PCB design. I've read all the specs for these and I really don't have the technical capabilities to design the PCB. I just just don't know what a platform for mining really requires with regards to components. Can someone help me out with that?

Eheh, damn. I was hoping you came across a truckload for $200 each, because you could sell out quick. Oh well.