20nm won't provide much of an increase in efficiency, and in this miners are about to hit a hardware wall. 14 or 16 nm will provide even less of an efficiency bump at one hell of a tab for NRE. As we're not even at 20nm as the standard chip size, we're a long ways from 16nm and 14nm. 28-40 will be the norm for a while to come yet.
From AsicMINER Gen 1 140nm to KnC 28nm the difference is huge, but that difference is now less and less and more and more expense to develop and produce. $/hash should now effectively stabilize since there isn't much lower they can go, and the benefits of doing so are nill. Whatever efficiency bonus gained is instantly wiped out by extremely high cost per chip.
Agreed, the chase for performance.
In the meantime Bitmain is killing it with their 55nm chips because they can design and go to market faster than anyone else. In 5 months they are on second product cycle already and have a 20+ and soon to be 30+ % hash network.
Operatr is very correct inasmuch; the differences in efficiency, are minuscule. What we should be seeing of course, will be larger rigs at a massively reduced $/GH/s (or even cents/GH/s). I would imagine any 16nm rig would need to be sold at <$0.40/Gh/s and provide >10TH/s to accomplish genuine viability in an increasing network hash. We are already seeing more efficient 55nm chips and even though KnC made the bold move to pre-sale the Neptune, by the time they are shipped, we are going to see a like-for-like efficiency in larger chips, comparable to that of the 20nm.
With regards to that increasing network difficulty, if one is able to get ~5TH/s running in a home environment from just a couple of rigs before; say autumn (fall), then you may break even on these, depending on the prices you have paid for such effect. I think there is one h/w manufacturer currently offering bundles to this scale - at a price which is clearly selling in the marketplace. Forgive me, i haven't done the mathematics on ROI for any of these, numbers isn't my strong point. However, the next step up is going to have-to-be a game-changer, and my only concern is how we are going to be able to spread the hash-power when it arrives. Somehow the gentle art of persuasion (to steer manufacturers away from selling rigs to big buyers in bulk) may simply fall to those with the funds to invest and even if some kind of WASP Collective can be established, trust in partners is going to be key to the success for h/w companies who have integrity at their heart. Those who do not, and are not so careful, will continue to screw us over, such as the case with KnC; building their own farms - to date, with 13% of the entire network.
I note a post on the spondoolies-tech thread which remarks upon WASP; perhaps this can work? I'll be watching that one attentively.
