My Roadmap:
end of September to November - Bitfury 55nm
from early January 2014 - CoinTerra - 28nm
I'm interested in your take on the CoinTerra chips, burnin. Is there really any point doing a board based on them? Their quoted power usage per Gh/s is only marginally less than Bitfury, and it sounds like they're probably impractical without water cooling. Sounds like a lot of hassle for no clear gain over Bitfury? Or am I missing something?
EDIT TO ADD: sure it's a shed load of GH/s per chip, but the power and cooling requirements mean your chip density is going to be much lower. By the time CoinTerra is shipping, Bitfury chips will undoubtedly have dropped in price substantially. I think a high density Bitfury design might be more interesting than a CoinTerra design...
roy
If CoinCraft pans out that seems rather more interesting than CoinTerra. Avoids the implausibly high TDP - and *way* better J/GH. Of course, always assuming they ship product in a timely manner, and that reality is anywhere near specs....
roy
EDIT: Hmm, I guess I was confusing chip efficiency with system effeciency. Still, looks like one to watch:
https://bitmine.ch/?page_id=863https://bitcointalksearch.org/topic/closed-bitmine-coincraft-a1-28nm-chip-distribution-diy-support-294235And an obvious dig at CoinTerra:
Do you think that fitting the highest possible number of hashing units into a single, 500+ GH/s silicon die is the best solution? We don’t. Real life experience have shown that the highest performance, lowest total system cost and greatest scalability are achieved by large arrays of small and cool ASICs.
