A3 has apparently been dropped. I remember one rumour (from a souce that had the money to be an Innosilicon partner) that it turned out to be BW 1401 level efficiency (or S7 level in BitMain terms).
Dunno if Innosilicon is working on a more efficient version behind the scenes, if that's the case.

Given the final results on the A4, I'm tending to believe that rumour.

The "seperate ground wire" was a recomendation BEFORE the 1996 NEC revision, it became a requirement after that.

 I'm not currently an electrician - let the license and bonding expire years back when I went into other work - but I keep up on the changes when I can.

 I DO strongly recommend that you DO NOT DO ELECTRICAL WIRING unless it's inspected afterwards by a trained electrician that DOES know the code for where you are at - amateurs make way too many mistakes that can cause shorts, fires, and overloads through ignorance.

 The stats on the CHIPS aren't even close.

KNC liked "big chip" designs, Innosilicon prefers "more chips but smaller".

 Think Spondoolies vs. Bitmain for comparison.


 I would anticipate pricing on the A2 to drop once a significant quantity of A4s start shipping - but not real fast, they're still profitable after all even at near-average Electric pricing.

 Strange, seems like most Monero miners currently are using GPUs.

Most mining program writers that support a specific LINUX generally make packages for Ubuntu 14.04 and occasionally for other Ubuntu versions.

 Those packages will generally work on most Debian-based LINUX variations with little or no additional work, if the otehr variation is recent enough.

 Otherwise, it's source code compile and that's going to be pretty much "any distribution of recent enough vintage and with the right drivers" should work.

I'm only seeing about a 2x kick. It varies depending on the motor startup circuit, load, type of motor, etc.

 You'd have to SEE the wiring setup on this place to believe it though - the actual wire used is decent, the boxes and breakers are QO but the actual INSTALLATION is a massive code issue - I suspect the wiring was all done before code was adopted here.

 Ever seen an OUTDOOR breaker box on a house, using an INDOOR panel? At least that's the neighbor's box, and whoever installed it used silicone sealer around all of the openings, but GAH.....

 Overall, I'm most worried about the use of bloody spices (they DID use the insulated screw-on connector type stuff that USED to be valid under the code but does tend to have issues at times) and the breakers overheat-derating too much come summertime (but I'm planning to be OUT of here by then, and I'm VERY well aware of required derating for higher temps).


 I was, in the past, a union-trained Journeyman Electrician, bonded and licensed for a few years (in another state, but NEC is NEC). Code hasn't changed a TON since then, other than to adopt some "best practices" recommendation type stuff that is now required (like seperate ground wire, which was only a recommendation 'till the 1996 revision), but this stuff is scary in some ways.

It could be done, but you would need multiple panels or perhaps a VERY LARGE commercial-type panel, and one heck of a large battery bank or a "grid-tie" setup to handle times when there is no sun.

Given the support for SHA-1 and SHA-2 at 256 bit, it MIGHT be possible to write a kludge program to get it to mine Bitcoin - but it's not going to be competative with a Bitcoin ASIC unit.

 So on a PRACTICAL basis, no.

My cooler is supposed to be a 1/4 HP but it draws about 500 watts at high setting (actual measured figure, higher for a fractional second when the fan motor starts up), I suspect it's actually a 1/3 HP. Electric motors are usually highly efficient, but they are NOT 100% efficient (1 HP = appx 750 watts at 100% efficiency).

 If I was willing to run every circuit at 100% of breaker limit, I'd have enough power - but the house is old enough I'm actually targeting 75% to be safe, *AND* I have to worry about the total draw on the main breaker being the REAL hard limit here.


 I AM thinking about breaking down some of the parts out of my cooler and using them on my existing fans to make impromptu evaps, it would save noticeable power.
 I'm hoping to be out of this place before I actually NEED the evep though - 6 month lease and I figure that last couple weeks to a month of that for actually moving....

Right now, for a lot of folks, the best isn't Bitcoin mining at all but GPU mining of altcoins (EHT and XMR currently are the hot coins, ZCash might get hot but now it's still in the "get it working at all" pre-release stage, couple other "dark horses" out there but seem to have quite narrow limitations).

 This may change *YET AGAIN* over the next couple months as BitFury is apparenly FINALLY actually getting something out into 3'd party hands and other shifts keep happening in cryptocoin mining in general.

 3GB cards will proably be fine for the remaining PoW lifetime of ETH - but that's older cards for the most part, I think the only current 3GB offering is one version of the "GTX 1060" that's also reduced core count and NOT a good miner card overall.

 3GB cards might get outdated eventually on some of the ETH alternate coin(s) like ETC if those stick with PoW.

Evaporative cooler aka "Swamp Cooler" - both helps keep the temps under control for a LOT less money than an A/C in low-humidity areas, and helps keep the humidity level up.


I don't worry too much about humidity levels though, the relatively dry air where I'm at now seems to be HELPING my sinuses a lot.

 There's a reason I've been building my recent rigs for what I've been doing with them - max flexibility and close to optimal for my PLANNED long-term usage of them (which the 1070 rigs have been doing for a month or so now, and the next 2 planned rigs will go straight into).

 Up side - other than the 1070 rigs, I think pretty much all the hardware I've used on ETH is now paid for or VERY close.
 Down side - I'm having to scramble more than I'd prefer to keep enough income going with changes overall to cryptocoin mining the last few months *AND* my own move complicating things.

I've already got my swamp cooler in the house, just not hooked up.

 Unfortunately, I'll have to shut down SOMETHING to power it when I need it.



 On a side note - I had quite a few of my DNet machines (the Distributed net client isn't mining as such, but puts very similar loads on a system for the same reasons) last ballpark 20 years before I finally retired most of the oldest ones when I moved this summer. The primary issue I ever had with those was FANS dying (and often causing overheat-related other dead parts), followed by motherboards dying.

 You CAN mine without cheap electric.
 
 You are going to be hard put to mine PROFITABLY for very long without it, and if you want to ever earn your money back on the mining gear you buy you better have VERY cheap electric or cheap electric and got one heck of a good deal on used gear (or price-war gear).


 I doubt there are ANY large farms in the UK, and not a lot of miners in general.

I ended up sticking some 60mm x 25mm Delta fans on my "fan died" HIS HD 7750 cards - they were left-over from CPU cooler days, I KNEW they blew plenty of air and had GOOD bearings.

 Turns out the cards run about 20c COOLER now, despite no shroud or any fancy mount directing the airflow - but they DID get quite a bit noisier.
 Thank goodness none of them were the 8000RPM screamers....


 The IceQ HD7870 has been a nice solid cool-running card, only issue with it is the fan sticks out so far it's a 2.5 space card.



 Don't get me started on reaching "max safe power limits" on my current place. Got everything traced out, and figured, and I'm going to have to shut down the A2 88Mh/s cause I don't have enough power for all 5 of my A2 units, and it's the lowest income generator I have left right now.

 10nm pure silicon or 7nm silicon/germanium alloy/matrix, depending on if the Intel vision or the IBM vision makes it into mass production first.

 BOTH have said "10nm is end of the road for pure silicon" - which implies a rather long gap to get the infrastructure built up to handle making wafers that are NOT pure silicon in sufficient quantity past 10nm.

 The other issue with ETH mining the last month is that the profitability has gone back into "major drop" mode due to the huge number of folks moving into ETH mining + dag file size growth dropping hashrates a little across the board for existing rigs + price seems to be stagnating at best.
 I suspect there will be a short "bump" in a few months, when the DAG file gets so large that 2GB cards can't mine ETH any more, then back to dropping profitability 'till the big POS switch finally happens.

 X11 is NOT a GPU option any more, can't even break even at 2c/KWH with ANY gpu due to the widespread deployment of multiple ASIC that handle X11. Baikal also handles X13/X15 and a couple other algorythms that don't have any "large scale" coins at all, write those off too.

 XMR might be an option - but watch it tank on profitability hard if a bunch of ex-ETH mining cards suddenly shift when they can't do the DAG any more.



 It does look like the ASIC wars are about to reopen, after too many months of BitMain having the only small-miner available new option. 8-)

 Miner sold with no PS is common - get used to it.

 Miner that INCLUDED a power supply was never particularly common, and the trend has been away from that for a while now for many reasons.


 One interesting concept for those of us that ran Gridseed "80 blades" - the same PS setup for one "blade" rig could easily run a pair of Baikals.

 All of the gold power supplies I've worked with are 1-2% more efficient near 50% than at 80-90%. It's not that big of a difference - you could save MORE by going to a same-size Platinum-rated PS (or comparable, Seasonic X-1200 Plat instead of the X-1250 Gold for example) and would probably pay LESS than by going to a 1600 gold from the same manufacturer.


 I DO recommend SOME "overhead" space for potential overclocking, manufacturing variences, and better reliability - but running at 50% is totally a waste of a lot of cost on the power supply.



 There seemed to be a bit of variation on the A2 Mega hash boards over time, but the pair I put X1250s into used normal 8-pin PCI-E connections on the hash boards, and a standard ATX-24 on the fan/control board.
 *ALL* of my A2s are 5-fan model, though one is a "long case" 88 and one a "short case" 110 with thin back fans.
 I have seen mention of some hash boards that used a "CPU-type" layout on the power connector - those had to have PCI-E plugs put in backwards to use a standard ATX PS.



 1600 is NOT that much more efficient than the 1300 - the numbers you're citing would make it a Platinum challenging for TITANIUM level supply, not a gold. 92-93% MAYBE at 50%.
 It would take a LONG time to make up the difference in electric usage vs the additional BIG COST of the bigger PS in that case.

 I noticed we had a big buyer today, did the math on their near-10k buy (might have been 10k before I got on to check the exchange) and a few $k a week WOULD be noticeable.

 On the other hand, a $Million as a "floor" order would be nice too, especially if it shifted up a bit every day or even every week.