But all of you missed the IMPORTANT part of that post.

 Is that picture really Iz, or is it just a picture of some random pretty lady?

 9-)




 BitMain doesn't care about maxing efficiency on their chips - never has. They want to max their profits, so their recent batches with fewer chips per miner actually make THEM more money, as long as they can sell enough of them.
 
I was VERY suprised when the initial batches did NOT come out running the BM1385 at .8 volts per chip or close, they actually set them at the mid-point (more or less) of the specified operating range.
 Later batches corrected that "oversight", or they might have figured out that the chip actually does NOT work reliably at the lower voltage figures of it's specs in a real-world string situation....

 Yes. Ethermine is PPLNS, and I think 2 hours is their window.

 Where?

I'd still like to see what F@H PPD are like out of a GTX 1080 and a GTX 1070.

 I'd say probability ZERO it will be a 2KW unit.

 They want to sell more of their PS after all.



 On that subject, did they ever get their QC issues straightened out on those 1600 watt PS they sell?

X11 diff has barely moved for months - not enough ASIC have actually deployed yet to matter.
 That COULD change any day now, if the BeMiner folks or the other "no actual reported sightings from a REPUTABLE source" folks actually start deploying them in quantity.

 I think it's more about "Ethereum is more profitable, and folks have started HEARING about that" - just like any rush into any other coin that has happened.

 For now and probably for a few more months. I'm looking further ahead though, when the mass influx + the announced "exponential diff mod" stuff will probably drive Ethereum to marginal on profitability - and of course the switch to PoS will kill it outright eventually (sometime next year IIRC).


 ethpool is "semi-solo" payout method, ethmine is standard PPLNS type pool.

 As far as I know any ethereum address should work with it - mine is pointing to the exchange I normally use for now, don't even have a wallet set up.

 I don't think it waits on a block to be found to show your shares, but you DO have to have submitted at least one share before you start seeing stats - and it takes a while for the graphs to start showing anything as they're based on the 24-hour average (Ethmine anyway, I never tried Ethpool).

 SP20 was more efficient than the S5 was when hashing at the S5's hashrate - though it was VERY close, close enough that manufacturing tolerances probably had a few of the worse efficiency SP20s not managing that against a few of the most efficient S5s.

 Bitmain will probably already HAVE competition from Bitfury on efficiency, and that LK1402 might be as well depending on how close BW.com/LKetc hit the announced specs on it.

 Right now, though, it's still looking like Bitfury is in the lead, as they've SHOWN working silicon for the full-custom 14/16nm generation and nobody else has (I discount KNC's "Solar" announcements, they've not shown any indication they have fielded that FOR REAL in any quantity worth mention IF AT ALL and KNC has a long history of paper announcements with VERY VERY long delays before delivering actual product).

(edit)
 And KNC is bankrupt now and likely to soon be dead anyway, RIP no big loss....


 I doubt that the S9 will be 1 BTC/TH - though I won't be shocked at .6-.8BTC/TH range. Consider that the S7 is down to around .3/TH lately - the S9 would have to be a .10w/GH unit AT THE MINER (not just at the chip) to be worth over *3 TIMES* the price/TH.

Temperature DOES affect semiconductor power draw, hotter = more, this is actually very old news and is the cause of "thermal runaway" situations.

 On realistic terms though, it's not a BIG effect. I'd worry more about "keep them cool to keep them lasting longer".


 And like I've already mentions, GOOD fans don't die when run at 100% of design speed for quite a few YEARS - a good ball-bearing design should still be alive by the time the GPU it's on is almost worthless from being too old and outdated. It's the CHEAP sleeve bearing fans the DIE YOUNG (I count the derivitives as well, like "hyrdo" and such - they're stlll SLEEVE bering and reliant on being lubricated to not sieze up over time - ball bearing fans run FINE for long time with no lubrication at all, just don't let dust jam them up.

 Isn't the peninsula that Singapore is on now part of Indonesia (except for Singapore itself)?
 That area WAS British dominated - most of the rest as I recall was Netherlands "controlled".



 Not all cloud mining is a scam - some of it is just overpriced (there is a difference between scam, where you DON'T get what you pay for, and overpriced/ripoff where you DO get what you pay for but pay too much for it).
 



 If you're going to move into that general area, might as well target Wenatchee/Moses Lake areas - Avista probably buys power from one of the PUC in "those three counties" and possibly 2 or all 3 of them.

 AFB aren't focused flow, they're FFB something-or-the-other or one of the FFB derivatives.

 Bitmain announced IN THE SAME ANNOUNCEMENT where they announced the S7 that they were working on another chip - I forget if it was 14nm or 16nm offhand. They have also in the last couple days mentioned that they would have information on the "upcomming S9" soon - but since they're not selling off "used" S7 units yet I'd guess at least another month perhaps 2 before they even think about starting sales on the S9, except perhaps on a pre-order basis.

 Given their roll-back on the efficiency spec of the A4 and the miner based on it from their original announcement, and that they were working on it at the same time as the A3, I suspect it turned out to have similar efficiency to the BW.COM LK-1401 chip rather than being a "new ground breaking" level of efficiency like the BitFury chip.

 It's also possible that the LK-1401 is actually the Innosilocon A3 design, given the long-standing relationship between Innosilocon and LKetc (which owns part of BW.COM), but that is pure speculation.

12.5 isn't impressive at all though - now that I've got my HD 7870 running, it's pulling very close to that (12.05 with occasional bumps up to 12.3) on Ethereum - on the other hand, my best estimate is that it's eating about 100 watts at the wall (massively overkill Seasonic X1250 (gold) running the system right now) which isn't competative.

 Pretty sad that a 2 Gig 5ish year old design card is pretty much matching the throughput of NVidia's latest and greatest - though could be early days pre-release drivers aren't optimised yet.


 I was just looking at my local Craigslist, see a 7870 listed for $90 and a pair of R9 270 (same except single 6-pin connector instead of 2 + updated BIOS) for $160. I'm being tempted......

Bitfury might have miners hit the market May 21.

I doubt BitMain will have an S9 out by then - though they might announce initial pre-order of the first batch of the S9 on the 21'st, for delivery probably a month or two later.

 I'm not seeing signs of them selling off a bunch of their used S7 having replaced them with S9s internally yet, unless that's what the "new 2 fan" batch is about.


 If I had to guess, I'd guess ballpark S9 will end up around 9-10 TH at 1200-1400 watt range when it shows up, and will be the same form factor as the S7 or very very close (like the S7 was very very close to the S5+ individual units, it keeps looking like the S5+ was a deliberate proving ground step to the S7 more than anything else).

 It's very iffy to even buy used cards NOW just for Ethereum - even the more efficient options might not manage to ROI.

 Now, if you have other uses for those cards after Ethereum gets to be unprofitable for you, or when it goes PoS, THEN it might be worth buying them and mining with them for a while first.

 I just wish I could get my HD 7870 working on Ethereum - trying one last possibility tonight, but so far it's looking to be giving me the same complaints at the same places.

 (Edit)

 Eueka, I found the clue I needed, now hashing with my 7870 FINALLY!

 Turns out I didn't have enough SYSTEM ram.
 2 Gig wasn't QUITE enough, 4 Gig is working.

 Sadly, I'm not seeing anywhere near the "reported" hashrates on other folks 7870s - but I suspect that is the "gradual deterioration in hashrate as the DAG files get bigger" issue at work, 12 Mhash at about 100 watts card usage isn't TOO shabby though.

 One thing I have figured out, this card is VERY VERY picky about memory speed setting - too high is actually WORSE than too low, optimal is just a hair above "stock" at which point it's all about the core clock. Dunno if that applies to all 7870s or just certain models though.

Unless you can manage to get the card down to 0C ballpark (perhaps lower), there is no such thing as "too cold" - at worst, it's a tradeoff between "keep the fan alive" and "keep the card as cool as possible to improve card longevity".

 Fans, relatively, are cheap - cards are NOT.



 And yes, I've done the "twist tie a GOOD fan to a card with dead junk fans" trick a few times, most recently on most of my HIS HD 7750 cards (many of THOSE fans didn't last 6 months, but leftover Delta 60mm "CPU cooler for Athlon Thunderbird/XP" fans not only keep the cards noticeably cooler than the crap STOCK cooling ever managed, they should also last a VERY VERY long time even at the 100% can't control them speed they're running at.

 The one HIS card I did NOT do that trick to died outright about 2 months after I bought the bloody thing - and it's still in the RMA process, HIS is BLOODY SLOW about that.
 

BW's current chip is not - their announced LK-1402 chip should be a lot closer, as it's specced in the same ballpark as the "on video" Bitfury chip.

 I suspect there can be one more round of optimisation past those (and any similar chips showing up from BitMain and Avalon and Innosilicon etc) as folks get better aquainted with the 14/16nm FinFet type node and process, but I doubt it would be a massive huge improvement.

 At this point, Bitcoin mining chips are about to reach the current semiconductor "state of the art" past which there won't be any major improvements for a few years (when the NEXT node reaches production stage).



 Might not happen then, given the "almost no improvement" that happened at the 20/22nm node vs 28nm.

 GTX 1080 was demonstrated to overclock to over 2 Ghz at the Nvidia demo on stock air cooling in the reference design.
 Also, while it has fewer shaders, they're supposed to be more efficient so it's not a straight up comparison.

 More importantly, some of my web digging over the last few days (slow at work) came up with references to some architecture changes that might help it noticeably on Ethereum, as they're targeted specifically at improving random access some AND they left the register count alone per processing unit while halfing the Cuda units per, effectively doubling the registers per processing unit - which should help a LOT if TLB register limits are what's holding the higher-end cards back.


 2048 per a couple of sites, which matches up well to the comparative benchmarks vs. the 1080, though not OFFICIALLY specified so far.





 Ethereum using qtminer
 X11 but that's off topic, so post it in "the other thread"?

Issue the first - on a fresh Ubuntu 14.04.4 install, aptitude is NOT installed.

Easy fix, just do apt-get install aptitude before you start the rest of the guide.



Unfortunately, the guide is COMPLETELY broken in Ubuntu 14.04.4 at step 12 due to an issue where installing MESA drivers *AFTER* you have installed AMD drivers causes your xorg.conf file to be deleted every time you reboot.

 Ubuntu folks are aware of the issue, no timeframe or mention of when/if they will have a fix - and since most later Ubuntu versions don't get along with fglrx AT ALL "upgrading" to a newer version isn't an option - not to mention certain other programs I run require 14.04 for compatability.


 I'm going to try installing fglrx AFTER the step where you install the mesa stuff, see if that's a viable workaround, but due to time limits it might be a few days before I can report back on success or failure.


 BTW - why is installing the BOINC stuff mentioned in your guide? I am not aware that SGMiner is a BOINC application or has anything to do with BOINC?

Have you tried Googling or asking in the Ubuntu forums to find a working install guide to USB for your version of Ubuntu?

It depends a TON on the fan itself. If it's a good ball-bearing design like all of my EVGA and Sapphire cards have had, or the blower in my HIS IceQ or ANY Delta fan I know of, it should last a very very long time even at 80-90% (my sapphire HD7750 has run 80 or 85% for almost 3 YEARS now no issues).

 If it's a cheap non-ball-bearing fan like all of my other HIS cards used, 60% for less than a year might be enough to kill the bloody worthless things.