If its pulling 1.75 watts per 3.17MH for scrypt thats a little over 2x increase in efficiency over KNC titans 3.9 watts/MH. Either way these have to be priced pretty low or be near the 10TH speed range to make any financial sense with current BTC/LTC prices.
Using the S5 as an example lets say it has 55 chips that means it runs at 8.4TH @ 1600 watts so if bitman can sell a 55 chip device for 400 then this would have to be priced somewhere around 1-2k USD to make sense.
I think 1.5k would be a sweet spot for such a device but thats probably a dream...even at 1.5k that would mean a 100 day ROI IF (and a BIG IF) difficulty continues to remain flat.
Of course this changes if they can simutainiously mine BTC/LTC ...but I'm pretty sure those stats are for single mode.
Then again people are still paying over 400 for an S5 which will never see an ROI so who knows...
They are supposed to release a 1 chip miner and a 24 chips miner
1 chip will probably be around 100 Gh stock, and maybe 2TH for the 24 chips (probably string design to feed the chips with 0.5v).
BTW, depending on what's inside the datasheet, a 20 or 22 chips string design could be a better choice, still efficient, but with a little undervolting capability for better efficiency.
Okay so our Dev Board will consist of 1 SF3301 chip. Our first miner will come in 2 forms: a single SF3301 miner and a 24 chip miner. These are still tentative plans, but be sure we'll update you when things become a little more concrete.