The normal ETA for a new Claymore version is "when I think it's ready".

 Reference/blower style cards help a lot, since they blow the air AWAY from the cards instead of just right around them.



 Per readings elsewhere *ALL* of the Z170 boards seem to have some serious limitations past 4 or 5 GPUs - that chipset doesn't support many PCI-E lanes to transfer data on (relies almost entirely on the ones on the CPU itself) and some of them end up assigned to stuff like M.2 SSD slots and the like.



 BTW - I finally got around to doing some tweeking on my my current 3 card 1070 builds, currenty pulling 86-88 MH/s on ETH running around 460 watts at the wall with a Seasonic X850 gold PS *WITH* the GPU on the A10 working MooWrapper - a lower-power CPU/APU would probably drop the system draw 30-50 watts.
 Stock BIOS, no undervolt, +550 memory -200 core, 75% power limit in Afterburner - and seems like there's still some room for improvement.

 Down side is the cost of the 1070s....

 [edit] was playing with a USB-drive based Win10 version, lost over 10% of the hashrate with no other changes. THAT experiment quickly got reverted [/edit]

 That BIOSTAR board seems to go on sale at $120 or so every 2-3 months for a bit.

 Flypool has done that at least once, I'd guess they'll do it again at some point.

 That's tempting numbers on that 7870 since I've got 2 of those (one reference, one HIS IceQ)....

 But GAH have to install Windoze for that, SO STUPID.

You get quite a bit less hashrate that way - might be worth it if your electric cost is higher than mine though.

 Same mods tend to also help ETH/ETC et cetera - but ZEC does seem to use less power.
 I was "only" noticing about a 20% improvement going from ETH to ZEC on my 1070s though.

https://bitcointalk.org/index.php?topic=1584973.0

 Phil likes the BIOSTAR Z170 boards you can put 4 cards on with NO risers needed.

Why do you insist on posting your speeds with a GTX 660?

 Those things are ancient, extremely inefficient, and NOT relevant to mining in the last few years.

 One of TheStilt bios for the R9 290 should let you kick the temps down quite a bit while still bumping the clocks up some.
 On ETH I'm routinely seeing 80C more or less at 1100/1250 under LINUX though I'm running the fans a bit higher than you are.
 

Appx. 40 h/s on a HD 7870 (HIS IceQ)

 950 core 1450 memory seemed to be optimal on clocks (less than those started dropping hashrate a little, can't clock the RAM higher and higher on the core actually seemed to SLOW the hashrate down a hair).

 Ubuntu 14.04 / Catalyst 15.12 / SDK 3.0

 Had to modify Makefile on the line where it asks for the path to the OpenCL directory to point at the APPSDK directory instead of that amd-gpu stuff that doesn't work with the older cards.

 Same setup on ETH using qtminer pulls about 10.5 Mh/s with same clocks.


 I was just bringing that rig back online (it used to have a R9 280x in it but that card appears to have gone bad during my last move so I swapped in the lower card just to get SOMETHING out of it), figured it would be a good rig to do some testing on before I put it back into production.



 A quick test on one of my R9 290 + R9 280x rigs gave about 160 h/s once it ramped up, split about 90/70 - but I didn't do a long run on that rig as those numbers aren't even close on profitability at this point to the same rig pulling 46ish Mh/s on ETH (16/30 or so) and that rig IS a long-time production rig.



 GTX 1070s are easy 28Mh/s on ETH with just some memory overclock and power limit / core clocks dropped to keep the consumption down, realistically 170h/s on ZEC out of one would be near-identical profitability on ZEC give-or-take the power cost where you live.

Even if they DID allow that (I am pretty sure someone is working on or has done LINUX for those), the APU in them is VERY similar on specifications to the A10-7890k overall (except for 8 CPU cores instead of 4 and lower clocks) and is NOT going to be a viable mining machine for any significant amount of mining done.

 The PS4 might have a slight advantage due to the GDDR5 memory, but it's still a VERY SMALL GPU at rather low clock speed by current standards (probably a tossup to a hair better than the HD 7750 which had the SAME number of GCN cores, GDDR5 ram though probably clocked a bit lower, and the same to a bit higher GPU core clock speed - the APU might gain a bit due to more recent GCN implimentation, but it'll not overclock nearly as well due to the TDP limitations and the many CPU cores as part of the same silicon).


 It's the XBox 1 that has DDR3, not the PS4 - but that's the ONLY significant difference between the APU in the 2 units (and is why the XBox 1 loses on benchmarks, slower memory matters).



 Keep in mind that back in the day that the PS3 could mine Bitcoin effectively, there was *NO* GPU compute capability at all so it was only competing with CPUs, and the CPU in the PS3 was one of the most powerfull available for ANYTHING at the time.



 As of the last time I looked at pricing for the two, I figured out I could build a computer with the top-end A10 APU for about the same price or a little LESS than a Xbox1 and definitely less than a PS4 with the same amount of RAM and bigger HD (or smaller but much faster SSD).

 Even with undervolting they STILL eat a ton of power.
 I run my R9 290s with BIOS from TheStilt and they STILL eat close to 300 watts EACH - up side, they also can overclock without thermal limiting + memory timings are a lot faster + they run a lot cooler since the BIOS upgrade, which makes them capable of pulling about 30% more hashrate then without the BIOS change - all at about the same power draw.

 For perspective, the GTX 1070 *usually* has a TDP of 151 watts (I have a couple Gigabytes that TDP at 180ish per nvidia-smi but they're high-end factory OC models), and I believe the 1080 has a TDP of around 250 watts - but outside of the crypto and cryptocoin mining worlds both of those cards blow the R9 290 completely out of the water.

 ZEC also burns through power - just not quite as much of it.

 For those with higher power costs I'm sure that's a significant factor in their profitability.

 It depends quite a bit on your cost of power, and on WHAT you are mining with.

 It's not a "it's more" or "it's less" flat out truth for all, especially since the profitability on all 3 has been pretty close for the last week or so.



 As for my power - it's all (or nearly all) hydro, and 100% renewable, where I'm at.



 BTW - I'm still giggling occasionally at the idea of "AMD" or "NVidia" propping up the ZEC market. Mining cards might be a noticeable percentage of their market, but AT THIS TIME they're probably moving about every chip they CAN build as it is - the extra *maybe* 1% additional cards sales due to mining taking a big (by mining standards) jump isn't at all likely to be worthwhile to them when they're already selling bloody near every chip they can get out of available foundry capacity.

 Keep in mind that NEITHER have put out any new "budget" level cards yet - that's where the BIG number of sales happen, along with "on-board" chipsets for low-level laptops and entry-level machines (though APUs seem to be taking over most of that market of late).
 AMD also has next year's launch of the "Zen" series 14/16nm CPUs and APUs to keep the foundry capacity available to them overloaded.

(1) Don't expect ROI of less than a year. The market hasn't supported THAT for a while (unless you got VERY lucky as an early ZEC adopter or a fairly early ETH adopter or a VERY early X11 ASIC adopter).

 (2) Location location location. Locate somewhere with VERY VERY low electric cost if you except to have a chance at long term profitability.

 (3) Live with the fact you're late to market, there is NOTHING you can do about that (unless another new coin shows up that managed to turn into another ETH or at least another ZEC).

 (4) X11 / X13 / X14 / X15 / QUARK / Qubit like SHA256 and Scrypt before them are dead for GPU mining, the Baikal ASIC mines ALL of those.

R9 290 and R9 3xx series would also be marginal to money-losing on ETH at 20c/KWH at current pricing and difficulty on ZEC.

 On the other hand, at THAT high of electric cost it's real iffy to mine anything at all.

 Curecoin value isn't even back down to where it was 4 months ago - though it's still sliding from the recent spike up into the 6+ cent range, it's not down into the 2.5 c ballpark that was the norm most of this year.


 Yet.


 On the positive side - we've repassed EVGA as the top producer team the last week and a bit, but that's more the result of them losing a ton of "short term for that early November competition" PPD than us picking up much.

 I can see the possibility of somewhat better than the current BM1387 being possible, but I doubt it's going to be a major improvement like Bitmain pulled off on the 28nm generation.
 30-50% better efficiency should be possible, but given that the 14/16 nm node chips aren't running at much lower voltages than 28nm was capable of, there's some serious limit on how much more efficiency is available to achieve.

 We've already had at least one early 14/16nm design that proved to be little or no better than the best 28nm designs (LK-1401, the apparently-abandoned Innosilicon A3 might be another given the final efficiency ratings on the A4) and BW already claimed some major improvements going to the LK-1402.

 Under 0.10 watts/GH? Possibly. Under 0.06? Looking MIGHTY iffy at best, even the Bitfury demo chip didn't manage that at it's lowest power setting.

 I'm a bit supprised we haven't seen SOI move down to the newer node range before now - did I miss something, or did SoI completely skip the 22nm genration?

 Check your wiring - mining hardware doesn't use enough throughput to overload a 10 Mb Ethernet connection, much less 100 Mb or Gigabit.
 You could also have a port going/gone bad - did you try using the machine on a different port of the SAME switch?


 Also, I strongly recommend that you don't use the "silver case" Netgear stuff. It's "consumer-grade" cheap and does NOT stand up well or long term.
 Upgrade to their "blue box" professional gear, THAT stuff works well for very long time at a time and the price isn't much more.

 I've got FS108 switches that are 15 years or more old that are still working well (one is a bit over 20 years old).
 ALL of their GS-series junk I ever bought died in less than 5 years and one of them died about 2 months after the warrentee ran out.
 I've NEVER had a blue-box Netgear switch die to date.

 (I did have a blue-box 10 Mb HUB die, but it was used when I got it and it lasted over 10 years in MY usage).

 Electric rates in China VARY - they might see more variation than the USA or Canada sees depending on where in the country someone is at.

 A very large part of the "super cheap hydropower" for China is in areas that are low on population, and they don't have anywhere near the transmission infrastructure that the USA has to distribute it.

 The 3 Gorges Dam complex in particular seems to have a lot of it's power go to waste 'cause they can't get enough of it to anywhere useful (yet), and they've not gotten enough "close enough TO use it" industry built up to date to fully utilise the capasity of those generators.