We're talking VERY SMALL scale here, 17KW on a hydro powerplant isn't even counted as "micro" by industry standards.

 The question is "is that "installed plate capacity" or is that the "average power output", there's a HUGE difference in those numbers for most hydro power.

 For an example, the Grand Coulee dam has an "installed capacity" of just over 6.8 Gigawatts (6809 Megawatts) - but on average over the course of a normal year it only puts out ballpark 2.3 Gigawatt-hours per hour for various reasons - and it's considered to be above average by a bit in percent of capacity actually USED over a year in most years.

A power supply will commonly operate a little more efficiently from 234 VAC than from 117 VAC because there is less resistance loss in the input circuitry due the current being HALF as high at the higher voltage.

 Nothing complicated, just simple Ohm's Law about IIR losses.

 It's also why a lot of "server-type" power supplies have a LOWER rated power capacity at 117 VAC - the input circuitry is only designed to HANDLE a certain amount of current.

 Also, your miner is pulling more than 1200 watts out of the PS - the official spec works out to be 1230 MIN to 1353 MAX when you factor in the "+10% discrepency" part of the spec for watts draw of the miner FROM the power supply on a 13.5 TH batch miner, at 1550 wall draw that would be as high as 87% efficiency (which is in line with the efficiency curve if you are at the max specified power draw).


 The important part is YOU ARE OVERLOADING THE POWER SUPPLY by running it on 117 VAC at your stated conditions.

 Yes, I DO own one - though it doesn't have a lot of cards on it yet.

 There is NO logical reason for it to be limited to 13 "standard" cards - and it's NOT a drivers issue, it's a BIOS limitation issue on the part of ASUS.


 LINUX used to support mixed AMD and Nvidia cards, but recent versions seem to have lost that ability - I'm not sure if it's something in the LINUX kernal that changed, or if the drivers themselves changed that much - but I'd bet on the drivers.
 AMDGPU-PRO in particular has a hard problem working at ALL, much less playing nicely with anything else, and the old fglrx drivers don't support the RX series at all.

 Mixed GPU rigs is one of the VERY few things Windows does better than LINUX at this time.

Yeah, I missed the "xmr" comment when they started talking sol/s - I always automatically think Equihash when someone starts talking sol/s 'cause that's an Equihash-specific terminology.

The "equihash" part was mistype that I corrected - apparently right after you noticed it.

 I have no idea what any GPU I have does on Cryptonight except my 750 ti cards, my single Vega56, and my GTX 950/960 cards (that would rarely switch to Cryptonight when I was running them on Nicehash but USUALLY were on other stuff).

 Not automatically true - some like the ATX power supplies because they intend to use them for other stuff once the ASIC miner dies - or in my case I would use my EXISTING G2-1300 or X-1250 supplies because I have a couple floating around not currently being USED for anything.

 8 PCI-E connectors on 1200 watts works out to 150 watts per connector - the EVGA "dual" cables should be ok as they apparently use 16AWG wiring to the first connector, then feed the second one with 18AWG, on the power and ground leads.
 There IS a 20 AWG wire in there, but that's a sense wire that doesn't carry significant power.



 Given how many S9 units ALONE are already in use, plus the various other models from other companies AND some folks still running older units like the S7, there is no way a single Bitmain S9 batch is going to have a HUGE effect on difficulty - the last few batches have only been pushing it by around 10%, and the "20% jump" before that was partly folks bouncing around into and out of BCH not just the new S9 batch.

Most folks that have posted about working with more than 4 Vega in one system have said it's a nightmare when it DOES work - and I have yet to see ANY posting of more than 6 actually working.

i3 might make a profit if the electric cost is low enough - it's not a high hashrate CPU though, due to the much smaller cache size vs. the i5 or i7 lines limiting how many threads you can run EFFECTIVELY on Cryptonight miners.


 Xeon E5 series by current standards isn't very power efficient - but the older E5 servers CAN often be bought cheap, which might make them worthwhile if you have very cheap power.
 Quantum "Windmill" open-compute servers in particular go for a song any more, and the Winterfell versions aren't much if any more on a per-CPU-socket basis.
 Down side is those particular servers only accept EARLY generation low-to-midrange power E5 CPUs like the 2520L/2550/2560/2570 and such (might be there's a BIOS update that lets them take later-gen CPUs but the later-gen CPUs tend to COST a lot more).

 Nah, that was Claymore a year or two back.

 This year I think he was shopping for a Malibu beach-front home - or was it Santa Fe Springs?

 You can append a worker address, by going walletaddress.workername as usual.

 I can't speak to appending an email address as I've never worked with a pool that wants that.


 You've got something WIERD going on if you have to run Chrome to get max hashrate - perhaps you don't have Afterburner set to run at startup, or something else is interfering?

When I am assembling a rig, it is on a power meter as part of the process - more to "tune" the card settings to hit my 6 amp target power draw at the wall than any other reason, but sometimes it's a handy tool for troubleshooing issues.

 Otherwise I don't bother.

 Turn off hardware acceleration in Chrome.

High traffic when sale started should have been EXPECTED - look at the typical Bitmain site overload before they started FORCING folks to use their BCH crapcoin to pay with.

Limit to 1 per person - on as small of a batch as this one turned out to be, I think that made sense - though it can be worked around, doing so isn't easy or simple when the site gets overloaded - it's STILL a crapshoot to get anything at all.


 The core issue was HOW TINY this particular batch was.

13 regular and 6 mining-specific if you mix AMD and Nvidia mining cards, per the manual - though I have no clue why you can't do 8 of each.

I am uncertain if any of the BIOS upgrades changed those limits, and have no CLUE why there would be any difference in the interface of a "mining-specific card" that would allow them to be used in higher quantities than "regular" cards.
VERY odd that.

 Mine to an exchange address counts as an "external wallet" to Nicehash - and you're not paying DOUBLE transaction fees if you do it that way.

 Only Ethhash algorithm coins (ETC, ETH, and such) use a DAG file.
 It is NOT AN ISSUE on anything else in cryptocoin mining.

 R9 290 are 4GB on the reference models, I think the only options were 4GB and perhaps 8GB on some aftermarket versions.
 Same on the R9 390.

 They're also still some of the highest hashrate ETH mining cards, though efficiency is not good due to their high power draw (even with an undervolt modded BIOS).

 7850 and 7870 do NOT do 300 sols/s or even 200. More like 150 on the 7870 and a bit less on the 7850.
 R9 280x as I recall was around 280-300 sol ballpark - DEFINITELY not 400 or more.
 Never tested my R9 290 cards on ZEC, but they should be quite a bit faster than the R9 280x - 400 sol ballpark would seem likely on those - but they make more on ETH mining than on ZEC mining.

 I doubt ANY of them can do 900 sol/s except PERHAPS the 7990 with it's DUAL GPUS ON ONE CARD (and those being the same GPU as on the 7970 or R9 290 or R9 390) - and that would be for the card, NOT for each GPU and do note that it counts as *2 GPUs* against driver limits.

1800x doesn't hash much if any better than the 1700 (when OC) or the 1700x (when OC) - they all tend to OC to the same clock speed or very close.

 If you want the 1800x for gaming, or already have it, then it makes sense - but for JUST mining on go with the lower-cost CPUs.

 For that matter, if you're building the rig from scratch for mining, and can find a good AM3/AM3+ motherboard at a reasonable price, the FX 8320e pulls 70-80% of the hashrate of those Ryzens while costing about ONE THIRD of the price of a 1700 (and even less compared to the 1700x or 1800x which have some added costs due to NO COOLER INCLUDED over and above their higher cost for the CPU itself - 1700 normally includes a Wraith cooler).

 GDDR5X has latency issues that severely hurt Ethhash mining on the 1080 and 1080 ti.
 This appears to be WHY the GTX 1070 had HIGHER ETH hashrate than the 1080, and fairly close to the 1080 ti even though the 1080 ti is a MUCH higher end card.
 No, the 1080 will NOT achieve 30+ ETH Mhash, unless you get INSANELY lucky on the silicon sweepstakes and can OC it a TON higher than most manage - and likely not even then.

 1080 is a better card on algorithms that are NOT severely memory-limited though, like equihash used by ZEN/ZCL/ZEC.

 At least Nicehash doesn't kill 50% or more of your hashrate on many algorithms due to shitty implimentation of VARDIF that screws you straight up front by ramping the diff up to IMPOSSIBLE levels fast then taking forever to ramp down, then repeat once it finally DOES ramp down to a viable level.

 Don't get me started on how many "engineer mistakes" I've found and had to help fix in the years I worked as an engineering technician.
 It's bloody AMAZING the issues they don't think about on how stuff works in the real world.

 The point on "riser and MB need to be on the same PS" is usually valid on the old ribbon-type risers where there often IS a +12VDC connection between the MB and the riser, but it DOES NOT APPLY to USB-type risers where there is ZERO power connection between MB and riser.

 Also, the BIGGEST issue with using SATA connectors to power a riser is THEY ARE NOT RATED FOR THE CURRENT DRAW.
 They are only RATED for 54 watts of +12VDC - the PCI-E bus is rated for 75 watts draw, and THE RISER ITSELF ALSO HAS CONVERSION LOSSES.
 Even if the WIRING is heavy duty enough to support the draw, the CONNECTOR is not.

 USB-cable type powered risers have ZERO power connection to the MB, though they do have a ground connection in common probably for signal shielding reasons.
 The only time it matters on having the MB and the Risers on the same PS is if you are using the old-type RIBBON risers, where there MIGHT be a +12VDC link involved.

 On the other hand, while the GPU itself PROBABLY doesn't "merge" the +12VDC input from the PCI-E bus and the PCI-E power connectors, it MIGHT do so - keeping the card and the riser on the same PS is less likely to cause issues than worrying about the riser and the MB being on the same PS.

 There will be no "supply catch up with demand" as long as the 14/16nm PRODUCTION LINES are totally swamped by BIG chip volume folks (like AMD, Nvidia, Apple, Samsung, and such) and the available capacity to the SMALL FRY like Bitmain, Caanan, BW.Com, Innosilicon, and such are having to do what they can with the available "SMALL PRODUCTION" slots.

 There is no "conspiracy" involved, it's lack of supply vs massive demand on TSMC and GF and Samsung (Intel isn't sharing, they need all of their OWN production at this point, and Samsung is pretty much booked out between their OWN demand and NVidia).

 Also, the manufacturers make more FASTER by selling the units then taking the proceedes to make MORE units (as much as they CAN do so) as opposed to taking the MONTHS to mine with them just to get back the sunk cost.
 They're NOT deliberately limiting sales - they just can't get enough chips TO sell as many miners as they want to sell.


 I don't see this changing for the forseeable future, unless Bitcoin pricing COLLAPSES to the low 4-digit range and causes demand to mostly dry up.