I've been lurking and learning about bitcoin/mining for awhile now. One of the more recent advancements that I've been keeping an eye on is FPGA mining. I was hoping someone could try to straighten out some of my mental difficulties with the development process of the FPGA miners.
What part of designing those kind of mines is the the current "bottleneck" (EG hardware or software)? More so is it that the hardware is available but getting "efficient" code onto the boards that's causing the problem? Or are the FPGA board makers spending more time working on the hardware and that the software is pretty well "set" for each release?
How much effect does the software that the FPGA developers have on the hash rate? I ask because I saw a post by one of the people making the FPGA miners saying they were doing about 100 mh/s at the moment but the "corner cases" still needed to be worked out. They reference being able to hopefully get more mh/s out of it once resolved. How/what are they doing to get that extra mh/s?
I've seen something about an open source FPGA software. With that being said could a person theoretically buy a FPGA board, load that software on it, and be good to go? (I'm under the impression that the answer is no). If that is a yes, what does someone who is "selling a finished FPGA board" doing for the user that they couldn't do on their own?
Thanks for any explanations that you can provide.
Most of the cost and limitation is in the hardware (they're very expensive at the moment relative to GPU's, unless you're buying top of the line GPU's I guess). As with most things, software can be tweaked to glean a little more hashrate out of these things, but we're not talking double or anything like that. It'll be small percentage increases.