FPGA Miner Design for Sale to Someone willing to Market It

[TeraMiner]

A couple of engineers with day jobs have been working on an FPGA miner with the intention of creating a product that would change the mining equation. We formed a small company that we call TeraMiner. We've developed a hardware and software solution that raises the bar on mining performance while lowering power significantly. We have a prototype running in our homes today using an off-the-shelf FPGA development kit. As it turns out, we've become very busy in our day jobs and have decided that it would be best for us to see if anyone in the BitCoin community has the time to take this product to market. Here are the highlights of what a "productized" version of our design would look like.

• PCI Express connectivity for high density installation into industry standard servers
• Entry-Level board achieves 750 MH/s @ 15W for $370.
• High-End board achieves 3,000 MH/s @ 60W for $1,050.

 
    
These manufacturing numbers are based on actual pricing we secured from Xilinx for their Virtex-6 product line. Selling prices would certainly be higher, to whatever the market will bear. We have also negotiated future technology improvements and price reductions through 2012. Details of the major components used in this design will be provided to interested parties.

We are prepared to discuss the details with anyone interested in creating their own line of BitCoin mining products. The level of engagement may vary from simply providing the source code of our current IP all the way to the delivery of a production-ready prototype. The details of these deliverables and associated costs will be provided to those who contact us at teraminer.llc@gmail.com.

Keep in mind that we started this business with the intention of going to market with our own set of products. But now that our day jobs have become too demanding, we're just looking to pass our design on to someone prepared to take it to market. Please contact us at teraminer.llc@gmail.com for any further information.

Thanks for reading.

[1714194836]

1714194836

[eldentyrell_old]

Entry-Level board achieves 750 MH/s @ 15W for $370

Sounds like your company's main asset is the ability to buy chips from Xilinx at an absurd discount.

You sure they'll let you "sell" that ability to somebody else?

If so, the buyer would make more money forgetting bitcoin and just make a killing reselling Xilinx parts via this deal.

  - e

[TeraMiner]

I do have many years of working with Xilinx in my day job. I have a commitment from them to sell enough chips to build 1.5 TH/s worth of boards. We also have a pretty optimized design that allows us to make the most of the Virtex-6 technology.

[eldentyrell_old]

I do have many years of working with Xilinx in my day job.

So do I, and so do a lot of other people around here.

The only way your numbers add up is if (1) Xilinx is losing enormous amounts of money on you (if so, congratulations!) or (2) you have found some way around the laws of thermodynamics and the DC Switching Characteristics data sheet.

I call scam unless you match some Xilinx part numbers to those hashspeed claims.

  - e

[TeraMiner]

I can assure you that there is no scam here. I did manage to get "large company" pricing on these parts because I do so much business with Xilinx. The low pricing is really what made our design so promising. You must know that companies like Xilinx and Altera treat their component pricing as confidential information. While I feel comfortable listing the full BOM costs on this public forum, certain details need to be discussed more discretely. Also, I did spent several months designing and optimizing my own hash algorithm from scratch. I've been designing chips for over 20 years so I really enjoyed the challenge.

I understand that there will be doubters but we have a development kit already running (with a smaller version of the FPGA) and have ISE runs that show the target device will hit the stated performance. I don't mind if there are doubters as that reinforces the advantage this solution offers. If you have a serious interest in making a product out of this, go ahead and email me at teraminer.llc@gmail.com to arrange further talks. You might be surprised that we're just a few normal engineers who don't have the bandwidth to build a company around this cool product.

[btcbaby]

I want one!

[film2240]

I like the sound of the entry and higher end FPGA boards.Is there an academic discount for those? As I go to uni sometimes. Still a good price for the power use/performance.Can you ship to UK? I'm interested as long as the prices stay good and that they live up to the claims made on here.Are those boards multi-FPGAs that are also overclocked? Because the pricing seems rather too good to be true.1 of those FPGAs would the equivalent of several GPUs if I had several HD6950s all OC'd (the higher end board would be equivalent to several of those).

Amazing if true.What payments do you take?

[P4man]

The only way your numbers add up is if (1) Xilinx is losing enormous amounts of money on you

Doubtful. I cant find precise die sizes anywhere, but looking at pictures from various Virtix 6 chips, Im guessing they range somewhere between <200mm and 500m2 (better estimates welcome). At 40nm they arent going to cost xilinx anywhere near those quoted prices to produce them. THe die size and process used match that of typical range from lowend to highend GPU's, and given the nature of FGPA's its almost certainly cheaper to produce them as GPUs (higher yields, due to redudancy, low power etc). List prices may be high, but thats just due to the relatively low volume and high fixed costs.

[norulezapply]

Sounds like a complete scam.
Only 7 posts and ridiculously exaggerated figures.

I'd stay well away from this guy.

[sgravina]

Me and my wizard buddies have done the same thing.  We are very smart and we have some powerful magic that makes our price point even better.

We only have one board but it does 900 MH/s @ 10 Watts.  Our cost is only $29.95 per board, if we buy two or more.

We are looking for someone to market our boards because we don't like to work too hard.  Wizardry isn't about money anyway.

We can't share anymore details because we haven't made them up yet.  Just send us some money and we will help you put together a good marketing story.

Sam

[worldinacoin]

It is possible to have huge discounts if there is volume, otherwise it is quite doubtful, maybe the OP can sell a few of their products, I don't mind paying a premium as I do understand that engineering samples cost quite a bit more.  I am willing to buy, test and give feedback here on the forum.

[ArtForz]

Piling on top of the "These guys are getting completely insane discounts from xilinx or this is BS" crowd.

[xanadu]

Your numbers are so far out of line from what others have previously scoped out using the same hardware platform that I am dubious at best. The Carl Sagan quote "Extraordinary claims require extraordinary evidence" comes into play here.

Post some verifiable proof, then folks might actually consider investing.

[Reckman]

ill pay 3000 for the high end board =p

[defxor]

I go to uni sometimes. UK

You might want to re-read the OP with that UK university background in mind.

we're just looking to pass our design on to someone prepared to take it to market

I'd suggest contacting ztex who's already making and reselling a similar product:

https://bitcointalk.org/index.php?topic=40047.0

[deslok]

i'd dump all my radeons right now if those figures are half right but some proof of existence would help

[marcus_of_augustus]

  ... sniff, sniff.

[eldentyrell_old]

The only way your numbers add up is if (1) Xilinx is losing enormous amounts of money on you

.... and high fixed costs.

Uh, yeah, that was my point.  If they sold parts to everybody at the price he claims to be getting, Xilinx would be losing enormous amounts of money.

[deslok]

are any of the virtex 6 fpga's actually capable of this? i thought they might be using stuff from here but it's not even avlable yet http://www.achronix.com/

[ArtForz]

Yup, 750Mh/s on a 240 is pretty easy. On a 195... tight. on a 130 or 75? Doubt it.