Thanks LR for your hard work and many successes in this ultra-competitive field!
Nevertheless, I'd like to play devil's advocate for a minute and cool the optimism around this announcement.
First, LR has announced a coming BitFury deal for 100-200 TH/s well over a month ago, in mid-December, so this isn't really news.
There are only two aspects to this which are new: the new boards and the collaboration with CoinSeed.
As for the former, the new boards may be somewhat better but I wouldn't call this a game-changer.
As for the latter, I'm still not sure how much we have to gain from economies of scale by working with CoinSeed. What percentage of the planned CoinSeed mine will belong to LRM? If it's a large percentage, than we gain little from economies of scale by collaborating. If it's small, than it might be in our best interest to compete with CoinSeed, by using LR's existing relationship with Dave to try and obtain as much gear as possible before CoinSeed comes online (which would probably be a gradual process). Just hypothesizing...
So, the only question that really matters about this BitFury deal is when will the gear arrive and start hashing. This question has yet to be answered, even though the deal has been tentatively announced over a month ago. Since the difficulty is still rising at a double-exponential rate, and will probably continue doing so in the near future, we are not-too-many-months away from the point at which 156 TH/s will yield just about the same dividends that we're getting now.
By the way, is this what LR meant by "ASIC project"? Or is that a separate thing? Because putting the same 55nm BitFury chips on a new board doesn't fit my definition of an "ASIC project"... Are we going to be involved in making our own chips?
+1
sounds good, but more info and clarification is needed
BTW: I do not think those are BitFury chips - at least not the original one. They have BioInfoBank stamped on them, and they are hashing at apparently 10GH/chip. However, they certainly look like BitFury chips. If they are BitFury, as the news article suggested, then does this really count as a new ASIC project with LRM partnering in it's development?
Maybe it's the original chip? 720 of them to an M-board. That means just over 2GH/s/chip which is about what the BF chip does. If they are the original chip, then I think I read somewhere that the chips do about 0.8W/GH/s, which would mean this rig can do around 1150W, call it 1200W, in minimal configuration at about 1440GH/s (yes, straight from my arse).
I've also read that these chips can do up to 2.7GH/s, but have no idea about efficiency at that level. I'm not sure, but if the clock in the chip is voltage sensitive, an increase in voltage would increase the clock speed, but also increase the I²R losses (which would go up with both current and clock for a double whammy) and greatly increase the heat that needs to be dissipated - but since heat transfer rate is proportional to heat difference (why the r9 290x can run at furnace temp) there may be a sweet spot, even with a passive cooler. It will certainly be interesting to see how fast the chips can be pushed -- my guess is LR will get them up to 1.7TH/s, maybe over 1.8TH/s with some active cooling.
Either way, exciting times... and this chip design could undoubtedly be ported to a smaller process node, which I wouldn't be surprised to find out will happen this year! All speculation...
Now we just need BFL to deliver the 25TH/s and get our 156TH/s deal installed, and... happy days.
