With the last few diff jumps, my 11GH FPGA rig is becoming an expensive heater. But cycles are cycles, and the power of distributed computing is an undeniable lesson we can take from the mining explosion. I've been writing some alternate bitstreams to play around with repurposing these things, and it got me wondering what kind of FPGA hashpower is really out there.
I'm considering a kind of cloud computing center where we run alternate bitstreams, and are paid (in BTC, of course) for having our FPGA's do other chores on a massive scale. The mechanics are all familiar - load bitstream and do work in a pool. Pool pays per share on a shift basis. Only difference is the work done is not simply hashing sha256, but whatever work people are willing to pay for. You load the bitstream you want to do work for (or is most profitable), and the server doles out data and keeps the troops honest.
I'd like to answer a couple initial questions with an informal survey. Please give a shout if you can.
1) Miners: If you own an fpga rig, what's your hashpower? I'd love to know #of LX150's, but MH or GH of sha256 is good enough since other devices abound. Is it online or off? Would you still run your rig if it were kept above water by other income?
Me? I have acquired 62 LX150's in ZTEX and X6500's. Not all in the best of shape, but alive. They are up, and I'd be happy to run them until they die.
2) Manufacturers: How much hashpower have you sold? I'm running under the assumption that this endeavor could soon pay more than mining btc, so I'd like to know the estimated pool of resources that could be captured if properly motivated. I guess we'll estimate +20% clones, unless anyone has a better idea of what the factories have been churning out after dark.
3) Developers: Anyone like the idea of distributed computing just for the sake of doing it? At ~3PH, I think the BTC network is safe. If you read some of the journal papers on sequencing DNA and other jobs that would be deemed trivial by the hashpower we're discussing, would you consider jumping in on these goals? Porting to other platforms is an especially important task that's too big for one small team.
4) Idealists: Sure, we could "test" the security of any number of algorithms or protocols. And there might be a fair bit of that. But if you consider how many networked FPGAs are out there, the distributed computing power is simply staggering. What would you use it for? Throw out your wildest ideas, let's see what sticks.
Overall, I envision this project to be: voluntary, open-source, and cooperative. This bitcoin thing grew from a bunch of people sucking up spare CPU cycles to taping out 28nm ASICs in just a handful of years. What else could be done?
