Question Please

[barngrover]

Hello Kramble. I am looking for someone to do some developing for a FPGA project for MaxCoin. I was told by the MAX devs that MAX is forked from Blakecoin and that you are a particularly talented programmer that might be able to help us out. Is there a way I can contact you privately about this issue without me posting my email? Thank you for your time.

[kramble]

You can contact me privately via a Personal Message on bitcointalk or litecointalk.

That said, I don't think I'm going to be able to help with a MaxCoin FPGA since (from a quick read of their wiki) it uses the keccak algorithm. I did some preliminary work on kekkac a few months back and was able to coax a few MHash/sec out of my DE0-Nano setup, however I TOTALLY FAILED to get it to build on the spartan6 LX150 chip that the old bitcoin FPGA miners use. Given the relatively poor performance, and my general frustration with the Xilinx ISE toolchain, I pretty much gave up at that point.

IMHO FPGA for crypto-currency hashing is pretty much dead now. GPUs beat them hands down on price per hash rate, though power consumption may be lower with FPGA. The only reason I can see for doing a FPGA port is as a predule to an ASIC project (don't go there, really).

[kramble]

No messages received, but I did find this https://www.startjoin.com/comments/project/maxmining

I've watched your videos on that thread, and I most definitely DO NOT want to get involved.

Closing thread.

[barngrover]

Sorry Kramble, I've been so frantic trying to find my best options I've forgotten who I have and haven't replied to. No disrespect intended.

Thanks, -Chris

On Thu, Jul 31, 2014 at 8:00 AM, kramble notifications@github.com wrote:

Closed #1 https://github.com/kramble/FPGA-Blakecoin-Miner/issues/1.

— Reply to this email directly or view it on GitHub https://github.com/kramble/FPGA-Blakecoin-Miner/issues/1#event-147665068 .

Chris Barngrover

[kramble]

OK. Thanks for the reply (I thought I may have upset you a little by my rather frank dismissal of the FPGA concept).

Anyway, all the best (and somehow I can't quite get the image out of my head of you sitting there in the video, stroking your white cat, planning...)

(The cat was heard but not seen)

[barngrover]

That's interesting, because my first project funding target was for -ONE MILLION DOLLARS!-

On Thu, Jul 31, 2014 at 9:45 AM, kramble notifications@github.com wrote:

OK. Thanks for the reply (I thought I may have upset you a little by my rather frank dismissal of the FPGA concept).

Anyway, all the best (and somehow I can't quite get the image out of my head of you sitting there in the video, stroking your white cat, planning...)

(The cat was heard but not seen)

— Reply to this email directly or view it on GitHub https://github.com/kramble/FPGA-Blakecoin-Miner/issues/1#issuecomment-50760163 .

Chris Barngrover

[barngrover]

Seriously though, it was the main developers who pointed me in your direction and I have been working with them on trying to get something good going for the MAX community.

I actually appreciate the frankness. I'm working with other peoples money and I need to know the best way to use it, not sugar-coated promises.

The project was born because one of the devs got his hands on some free FPGA equipment and it was slow going to get something done for it. I started the project independently to help out and experiment with a reward system that would keep people coming back for more. Turns out the boards were for bioinformatics and weren't going to do us any good, but by then I was committed to the project with money flowing and I needed to get some results.

Anyway, thanks for the input, and I know who to contact when we go into development of our quantum computing mining rig!

-Chris

On Thu, Jul 31, 2014 at 9:58 AM, Chris Barngrover barngrover@gmail.com wrote:

That's interesting, because my first project funding target was for -ONE MILLION DOLLARS!-

On Thu, Jul 31, 2014 at 9:45 AM, kramble notifications@github.com wrote:

OK. Thanks for the reply (I thought I may have upset you a little by my rather frank dismissal of the FPGA concept).

Anyway, all the best (and somehow I can't quite get the image out of my head of you sitting there in the video, stroking your white cat, planning...)

(The cat was heard but not seen)

— Reply to this email directly or view it on GitHub https://github.com/kramble/FPGA-Blakecoin-Miner/issues/1#issuecomment-50760163 .

Chris Barngrover

Chris Barngrover

[kramble]

FPGA is rather a strange beast as applied to crypto currency mining. I suspect there is some "magical thinking" going on in some people's minds, with the expectation that is must somehow be better than GPU (perhaps midway to an ASIC) in performance.

In practice, for most algorithms, GPU turns out to be optimal, short of a full ASIC implementation. The reason for this is that GPUs are bleeding-edge high performance, high volume (hence low cost) products, while FPGA is a much more niche product, hence behind the curve technology-wise and much more expensive.

If it was not for the application of FPGA to bitcoin mining then we probably wouldn't even be having this conversation now. Even then, the cost per hash performance was somewhat worse than GPU and ROI relied on the huge jump in the value of BTC. SHA256 was actually quite a good fit onto FPGA, other algorithms vary quite a bit. Scrypt is very poor, blake was good (hence my involvement in that project). Keccak ought to be fairly good, but in practice does not seem to work well (this could just be my relative inexperience here, I'm not a professional FPGA designer, and the device I was targeting, the LX150, is rather old technology now). But none of this overturns the generic disadvantage FPGA has relative to a GPU miner. There is no way that new-build FPGA can compete. The only hope for ROI is with reusing cheap secondhand boards (ie those old bitcoin miners, or maybe even your bioinformatic boards). But I don't think there is the return needed to justify the cost/effort to develop the VHDL code for the mining algorithm (professional developers don't come cheap). And keccak probably isn't the algorithm to target given the low value of the coins currently using it. X11/X13 would be nice, but it's a huge amount of work, and not a good fit onto those older devices (back to my original point, no way FPGA can compete with GPU).

Anyway, I'm just rambling on and repeating myself, so I'll shut up now.

PS. I'll believe in quantum when I see it. Too much hype right now, D-Wave especially.