I just spoke to BFL and they agreed to demo the current development hardware/firmware tonight. However, I am otherwise already engaged tonight and they agreed to do the demo tomorrow afternoon.
Part of the agreement of me viewing the development hardware and software was that no hard numbers are released for this particular demo, although I will confirm what I actually see without providing exact numbers until they have a production version of the software available for release.
Not exactly what I know we all want to see, but at least they are willing to provide something. Having developed both hardware and software products in the past, I am inclined to be a little lenient with expected production issues to a moderate degree.
So anyway, I will report back what I see, just don't expect hard numbers for the demo tomorrow.
Rock on, thanks. Even though you can't give hard numbers can you atleast give us a thumbs up or something if it is within, say, 20% of the stated 1.05GHs? ;p
So this goes from $500 1GH/s to a $700 800MH/s ?
Interesting.
Actually, these prices being raised and numbers being lowered make sense. While I question the legitimacy of this, I think that the prices being raised could mean that production costs, and actual numbers they intended to reach aren't as easy to reach (with early software) as they intended. My understanding is that the hardware is quite capable, but the software isn't yet. I'm open to criticism on that point, but my overall statement I think makes sense.
Well mining software is pretty trivial (relatively speaking no offense intended mining software authors), and open source. My understanding is they are simply taking an open source mining software and modifying it to feed the exact same data in a format understood by their board.
Most of the mining software would be identical to existing software (long polling, pool configuration, share submission, etc) the only part which would be modified it talking to the hashing engine and it is as "simple" as "engine here is a block header. Complete nonces 0 to xxx using this header and return any shares".
Oh, well see I was wrong. I thought FPGA mining software was still very inefficient (and not optimized) I assumed this was the issue. The price increase, and hash decrease could come from a hardware limitation they did not foresee in the past.