From what I've seen reported, Polaris will start out in the low-to-mid range (more or less replacing the current R7s and perhaps the bottom end R9) and won't move up-range for a bit.

 I've not seen ANY report of it showing up next month or even July - might get an "official announcement" in June, but actual "release for sale" date last I've seen was still estimated for mid-to-late 3Q 2016 at the earliest - with the TSMC damage from the big earthquake adding months to a LOT of estimates.

Just don't spend a ton building ETH rigs from new parts, it is very iffy if you'll manage to ROI if you try that at this point.

 (Unless, like me, you were planning to build the rigs anyway for other usage and are just mining ETH short-term with them to help defray part of the cost).

How are we suppposed to give a pre-order quantity when we have ZERO CLUE what the price is on the miners?

 This is TOTALLY BACKWARDS.

Every sleeve-bearing fan with a Cooler Master logo on it I've ever bought died within less than a year - specifically including a pair of Gemini coolers (nice coolers, work well once I put REAL fans on them).

 On the other hand, I DO have a few fans with Cooler Master logos on them that appear to actually be Delta fans that have lasted a long time.


 

I'm not quite as big as Phil.

 My numbers are (right now) electric rate at almost exactly 7 cents/kwh 'till middle of June (by which point I hope to be in process of moving, to a land of VERY VERY CHEAP electric). Base rate goes way down after the first couple thousand KWH, but "transportation fee" and "energy cost surchage" actually cost me MORE than the bloody base rate.


 Appx. $3700 invested into my A2 LTC farm, a fair bit of that is paid off already. Been earning 4+ LTC a day reliably, the last couple weeks closer to 4.5 - most of that time with LTC in the 3.25 ballpark but the recent runup pushed it close to $4.
 Power usage total (at the wall, measured) right at 5200 watts, so appx. $8.50 in electric usage per day for that farm - consistant 9 months of the year (I've not owned the A2s quite that long).

 Long-time numbers look like $13ish a day earned for a net of about $4.50, last couple week numbers $16-$18 a day earned for a net of $7.50-$9.50 a day.


 Eth farm - I have $80 invested, the rest of the hardware I already had or was in process of buying for other intended usage when I started working on Ethereum.

 Ballpark 80Mh total - most of the time that worked out to about .8 ETH/day or ballpark $8/day earned - recent has been more like $10/day earned with the price runup.
 Ballpark 1400 watts used (and this is nothing CLOSE to an optimised setup), so under $2.5/day.

 Net $5.5 a day most of the month, up to $7.5 a day - and if I count what I paid for this gear NEW, then subtract out how much I net earned on some of it mining Litecoin and DarkCoin/Dash, I estimate the TOTAL cost of the gear to be a little over $2000.


 I priced out a 2-card rig on Newegg this past week - fairly conservative estimate 48MH/s (I don't trust most of the posted MH figures, they seem to be fairly old and bigger DAG files slow cards down) at a little over $1000, would earn ballpark .48ETH/day or $6ish right now on an electric cost of ballpark $1.20/day
 I had the ROI figured on more exact "current" figures at 255 days.

 I did the same on a 1-card and on a 4-card rig (3 cards takes an EXPEN$IVE motherboard to get proper spacing, or you end up crowding one card on any 3-slot-only board).
 The 4-card rig was deliberatly designed using blower-type cards for heat management.
 ROI worked out to higher on the 1-card rig despite being able to go VERY cheap on MB/RAM and such (mini-ITC AMD A1 slot stuff, but gold power supplies don't EXIST below about 450 watts so couldn't get REAL cheap there).
 4-card rig ROI worked out to be 254 days, but most likely the cards wouldn't be able to be pushed as hard as in the 2-card rig so I suspect the ACTUAL ROI would be a bit longer.

 VERY VERY iffy on being able to achieve ROI, as the current price bump isn't looking real stable, hashrate is GUARENTTEED to drop over time as the DAG file keeps growing, I just couldn't justify building a complete rig out of NEW parts for ETH mining.

 I can *mabye* justify it on the basis of "mine ETH to recoup some of the cost, the do my other stuff afterwards" if I set it up with NVidia cards instead of the AMD ones I specified - but then it earns less ETH and probably becomes unprofitable sooner, or fails to achieve ROI before ETH goes POS, so I'd have to justify such a rig on the basis of "going to buy it ANYWAY" not on "I can pay for this with reasonable confidence before ETH goes POS/unprofitable".



 Still, for the short term for those of us with hardware that CAN mine ETH already on hand (or MOST of the parts already on hand), ETH is certainly a good option over a mythical miner that might not show up for several months IF ever.


 I've not looked into undervolting my GPUs yet - PrecisionX won't let me undervolt my NVidia cards (and research indicates I'd need to mod the BIOS on them voiding the warrenttee), and I've not looked hard at undervolting the AMD cards yet (I know the tools exist, just haven't had time to dig up the right ones yet).

For home miners the last few months, BTC has been no competition at all to ETH.

 For the big farms, it's a lot more complicated - but I'd bet at least a FEW of them are doing SOME ETH mining.

 
 The big questions still remain though. When will there be an S9, what will it cost, how much hash will it do, how efficient will it be, and will it continue to be string-design unadjustable (by practical means) efficiency or will it go user-settable efficiency / max hashspeed tradeoff?


 Oh, wait, I see six sigma certainty it won't be adjustable-efficiency - so one less question to worry about!

Not all up cost at 2 c / KWH (if you know of such an area, I WANT TO KNOW before I commit to actually moving) - though there is one small area that can hit sub-3 on some classes of service.

 7 cents/kwh in the US isn't real hard to achieve - but most of those areas are climbing as the EPA keeps hammering on coal power, especially under Obama.

 It's too bad that there are so few places that are well suited to hydropower - and that so many of those places are inaccessable, or heavy populated to the point you can't get a dam built due to massive lawsuit issues.


 I'll point out that KNC is Scandanavian-based - no import fee when you're BUILDING the stuff right there - though they might have an import fee on the chips themselves, dunno if anyone has a current-art or close-to-current-art fab in that area.

 Less on a couple of my ETH systems - though not by a lot.

 If AMD made a 754-based CPU on a semi-current process I could probably drop the 190ish watts my pair of 7870 single-card Semperon 3000+ systems and the 170ish the 7850/Semperon 3000 systam is using by 30+ watts each.

 On the other hand - existing hardware, ZERO out of pocket for the most part (I did buy ONE HD 7870 locally for $80, but I figure it'll make THAT back by the time that system becomes unprofitable - and if it misses by a bit, no big deal it turns into a very cheap serious RC5-72 cruncher system).


 Unless someone (Bitfury-based most likely) gets a 14/16nm custom-gen Bitmine miner out before the halfing, it's looking very iffy on my returning to Bitcoin mining at all this year - though I'm going to have to do some serious number crunching when the A4 shows up before I decide.


 As much as I'd like to see a sub-$400 Bitcoin miner show up in that generation, I'm betting it doesn't happen this year IF at all. Takes about as much time and effort to design a small miner, the infrastructure costs on it are HIGHER per MH, making it that much harder to achieve ROI on it - and I suspect the market for such a "small" miner isn't all that big.
There's a REASON Bitmain has never bothered releasing a BM1385-based "U-series" miner design, much less an "R1" upgrade to the next-gen chip.

 290 did get to 30MH though, so it was still scaling somewhat - just not 100% scaling vs. shader count.
 Limit is definitely somewhere in the memory system though - and it's NOT just bus width, Fury/Nano have a 4096 bit wide memory bus (due to the structure of HBM) yet they're as fast as the 290 or about the SAME hash.

 Genoil IIRC was speculating it was a limit in the TLB table hardware, but I dunno how far their research into that has gotten.

 I don't see anything close to 20 MH out of my 7870s - the 270 has the SAME shader count, but it does have somewhat faster memory and IIRC 2x the memory bus width, but while it's certainly a good bit faster than the 7870 it's nowhere near 2X as fast, much less 2x PLUS the memory speed ratio.


 TDP target on Polaris should be a LOT lower - the smaller node is a lot more efficient, even if you puch more transistors AND kick the clock rate up quite a bit.
 On the other hand, it won't be out in time to do much Eth mining with, unless AMD pushes the release dates up a good bit.

Ethereum mining is more profitable at this time, and probably will remain that way for at least a couple more months.

Mostly it comes down to what the price is at right now, and the slow difficulty cree p built into DAG file growth.

 The big hit to many current Ethereum miners will happen when the DAG file gets too big for a 2 Gig GPU to be able to mine any more, assuming Ethereum is still profitable by then - on the plus side, folks with 3Gig+ cards should show a jump in profitability for a while after that.

 Not a big deal to folks like Phil and I that are mostly or entirely mining on existing left-over-from-Litecoin (or left-over-from-X11 etc) hardware - it's just extra profit on hardware that was already paid for or mostly so while it still works. It's the folks spending $$$ on new gear that might never achieve ROI that are gonna hurt, though they should be able to recoup most of their losses or even end up profitable by selling off some of their used gear when it's no longer profitable.


 (No, my 5 NVidia cards weren't left-overs - but I was in process of buying them ANYWAY, so they are kinda-sorta that way by my figuring. ALL of the rest of the hardware I'm mining on at this point except for an $80 HD 7870 were Litecoin/X11 left-overs that I was using in my DNet/GIMPS efforts, bought specifically FOR those efforts and have been re-purposed for a while).

Ether right now is the "hot" coin, though Litecoin has been a lot more stable since it's halfing last year.

 I suspect the "hot" of Ethereum will dry up big time once it goes PoS.

I'd actually prefer it be higher = fewer machines sold = less hashrate growth = longer timeframe for my existing A2 farm to make more money.

 9-)

Only if the machine never goes down - and with all your cards on one machine, it goes down at all you lose ALL your hash.

 That's not even factoring in the pain of getting such a machine built in the first place, the higher costs of building such a machine and getting everything coordinated, and getting it to even start mining at all.

 Perhaps quote the R9 280 (Newegg has them on sale right now) - claiming 30 MH at 300 watts for almost identical up-front cost.


 *HOWEVER*

 It's not a fair comparison to just quote the cost of the CARD vs the cost of a complete ASIC - a FAIR comparison would be "cost of a multi-card system" vs that "ASIC system".

X11 really is too small a market to support multiple ASIC makers - I'm more than a bit supprised to see 2 CLAIMED ones right now (though only one that has shown actual product to anyone reputable, and THAT product seems to be in very short supply).


ASIC for ETH = suicide, ETH moving to POS will kill that market too soon.

It's a limit of the AMD drivers - which are NOT used to run non-AMD video devices, so trying to count your Intel and NVidia cards into the total is a false argument.

 Why do you want to run that crazy number of cards on one machine in the first place?

The Block Erupter was NOT a LKetc design, NOR was it a Scrypt-based miner.
Block Erupters hashed at more like 300 MEGAhash/s not KILOhash.

 That sounds suspiciously like one of the USB sticks that was based on the GC3355 chip from Gridseed - might try talking to the DualMIner folks as they're still selling similar sticks.


 Do keep in mind that a lot of auction sellers do NOT make their postings particularly complete and are not always ACCURATE on their postings.
 The quoted listing contradicts itself and is a PRIME example.


 BTW - as far as I know, LKETC never made a USB-based device of ANY type, much less anything based on a ZEUS chip (LKEtc was an A1/A2 (Innosilicon) BIG miner making company prior to gettng involved as a partner in BW.COM).

ETH last words I saw is good to somewhere in 2017 (date uncertain) when it goes POS.
They were ORIGINALLY talking POS in 2016, but appear to have delayed the planned date due to developmental delays/issues of some sort and decided they needed more time to get the issues sorted out.


 Intel is actually doing 10nm in their lab - they've had 14nm in production for a while (Skylake).

 Nvidia is releasing their first 14/16nm gear this month
 (I forget for sure but I think they said somewhere that they had qualified or were working to qualify the GTX 1080 and 1070 on both Global Foundries AND on TSMC processes).

 AMD is talking very late this year or early 2017 for their initial 14/16nm releases.

 IIRC Samsung was the first "big" manufacturer to release products at that node - portable gear like Smartphones NEED the power efficiency of the smaller node more than anything else.

Yeah, been seeing the posted 1080 benchmarks - very underwhelmed, it SHOULD do better than a 960 at least!

 I suspect the drivers need some serious optimisation, but I'm sure NVidia is already working on "in general" optimisation for Pascal in their drivers.
 Just gotta hope the "in general" stuff makes them work better on Ethereum SOON.

 240 vs 120 on most gold power supplies is only a 1-2% gain in efficiency. Not exactly a big drop in electric cost, though every bit can help sometimes.

 $1200 with good part selection would probably get you a rig doing 80-90 MH easily, with VERY carefull part selection on parts on sale might get you to 120 - assuming new parts and from-scratch build.
 Better to upgrade existing machines on a hahs/$ basis as much as possible.


 I expect GTX970 prices to collapse over the next month, as the GTX 1070 is due out June 10 and blows away the 970 hands-down for most usage - most of the benchmarks I've seen put the 1070 in Titan performance territory or a little higher.

 Now, if someone could get an ETH miner optimised to work WELL on the 1070.......


 Dropping the voltage by 10%, assuming no change in resistance, actually drops the power draw to 81% - but pretty close to that 20% quoted.