BFL - 65 nm. KNC, 28 nm. Conclusion--exactly the opposite of yours here.
I still highly doubt they can have access to the 28 nm factories. It's still cutting-edge technology used mostly by big players with a lot of hard issues to solve.
It's been in use since 2011. And you doubt that Orsoc has access to them? An ASIC/FPGA design-house with over a decade of experience and 10+ employees, most of whom are chip designers?
Yes but BFL was initially slated to use 5 x less power than they do now they are shipping. They advertised and took large pre orders based on 1w a G/H. They are shipping at 5w G/H or slightly less for the bigger rigs.
I was not aware of that. It's very interesting if you consider the very long-term usage of ASICs: even them could become unprofitable to run given power costs!
BLF outsourced their chip design to who knows who and ran into problems. Because you simply can't simulate a working ASIC design on an FPGA and copy it directly into a chip. You have to know the little tricks of the trade of turning an FPGA design simulation into a functioning ASIC. If anyone should know how to do this, it would be a long-standing design-house like Orsoc.
It's not really surprising that BFL ran into problems, problems that it seems Avalon narrowly avoided and fore-saw, but it would be surprising if Orsoc runs into this problem, considering that they've been producing ASICs for years and claim that a hashing chip isn't even close to the most complex design they've completed successfully.
I think too many of people's expectations have been colored by the difficulties of BFL and the like. These guys are pros.
Anyway I'll stop here as this is off-topic.
