SATA power connectors are NOT designed to handle the power draw of a GPU - that may be your issue right there if you are using SATA-power risers.

 They're only designed to handle 40 watts or thereabouts, the riser is going to be putting more like 70-80 watts of draw on them.

 I don't know where you are seeing this LIE, but any card with multiple power connectors will flat out refuse to work AT ALL if they aren't both connected.

 Trying to run on one will totally mess up the power circuitry ON THE BOARD to the point the board probably wouldn't work at all and might FRY that power circuitry AND THE GPU AND MEMORY, even if it DIDN'T check for both connected in the BIOS code before the BIOS enables the GPU for actual usage.

 Waste of time.

 The GC3355 used in the Orb is very very outdated on the SHA256 side - it was UNPROFITABLE even during the last few months the Orb was being MADE, which is why Gridseed shifted to making the "blade" and "GBlade" models that were Scrypt specific.

 The SCRYPT side is no longer profitable unless you have FREE electric.

 It took a "blade" 40 chips to ALMOST MATCH the performance of the Futurebit Moonlander II (which uses a SINGLE chip, I believe the same chip the Innosilicon A4 uses but the maker has never confirmed or denied that), and the power usage was ballpark 10 TIMES higher than what the Moonlander II uses.




 The algorithms folks are seeing 400 hash/sec out of on GPUs are NOT SCRYPT.
 Scrypt hashrates would be more like 1-2 MEGAhahsh out of most recent mid-to-high-end GPUs, and a few of the top end ones should be in the 3-4 Mhash/sec ballpark.

 Lots.

 I ran a multi-ASIC + multiple GPU rig farm for about a YEAR on a junk Virgin Mobile 3G connection that saw 200kbps DOWN on good days without issues from the mining.

 Most major mining software writers support Ubuntu.
 In fact, I can't think of one that does NOT (though Claymore was anti-LINUX for a while he changed his stance when he finally realized how much he was LOSING by not doing so).

 Hash rate generally isn't OS specific - the differences are minor at most, and are normally DRIVER dependent issues not OS specific ones.

 The main tradeoff is that LINUX is more stable, but the tools for overclock and undervolt are a lot more primitive and don't work as well as Afterburner for Windows.

 You do NOT need a high-end processor AT ALL for mining, and mining motherboards don't tend to work well as high-end gaming motherboards as mining doesn't NEED the throughput to the GPUs that gaming does.
 I ran my triple R9 290 based ETH rig on a Sempron 145 low-end single-core CPU for many months, and saw ZERO hashrate change when I upgraded that rig to a Athlon X2 240 dual-core (the only difference is that ETH's DAG file generation went quicker, but that normally happens BETWEEN blocks and doesn't affect hashrate).

Hacksaw + miter box.

Circular saw + metal cutting blade.

Table saw + metal cutting blade.

Square tubing is EASY to get square cuts on as compared to round tubing (though the miter box and table saw methods both work well on round tubing, the circular saw method is iffier).

There is no way to limit a coin to CPU only - if it gets popular enough SOMEONE will write a GPU miner for it.

 The best you can do it make it a "hard on memory usage" coin, like Monero, or make it something like BURST where the hashes are PREcomputed then stored on a HD, and are complicated enough it's difficult to impossible to compute them "on the fly" fast enough to matter.

From a power standpoint, those 2 PS should work fine.

 Higher-end EVGA power supplies some with a "power test" adapter that is basically a "turn the thing on" plug that plugs into the ATX power connector, works well in a mining rig setup.
 I'm NOT sure if the GQ line includes that adapter, all of my EVGA are G2 models (I refuse to own any more PS with a "sleeve" type fan bearing, no matter how fancy a name they put on it they ALL die too fast in a high temp environment).

 You won't FIT 10 Moonlanders on that, the ports are too tightly spaced.

 Ballpark $150-200, but won't see them THAT low 'till ETH profitability drops quite a bit more.

 They use the SAME algorithm, that stresses the SAME parts of the card.

 The point to Dual Mining is one algo will be MEMORY heavy (ETH) the other will be CORE heavy thus allowing more full usage of the ENTIRE card.

 And it's "standard" 2 card spacing, NOT wider for good cooling.
 Note that the Pandaminer was designed around PASSIVE COOLED mobile-type GPU cards - one of the reasons it has so many case fans - and thus it can get away with closer spacing while still having good cooling.

 Why avoid Molex - the connectors can EASILY handle the needed 75 watts (the actual connectors are rated for about 150 watts IIRC) as long as the WIRING is beefy enough.

 Avoid SATA definitely, those connectors aren't rated for more than about 45 watts and WON'T handle the load.



 to rslx - if you're willing to wait, your best choice is to buy from the manufacturer(s) direct.

 Probability ZERO as long as it is the current "top gaming card".

 Be happy it's back in the $700 or a bit under range again, where it was in January.

 Still wasting a ton of screen space.
 The third most irritating thing with the current wallet is the HUGE amount of wasted "white space" in it that makes it a PAIN to be able to see what's going on without completely overloading the entire screen.
 #1 isn't actually the wallet itself - it's the "have to keep tons of coin around to stake in a reasonable timeframe" issue.
 #2 is the frequent wallet CRASHES that have returned with the current version.
 #4 is the "I want to pop up FULL SCREEN IN THE MIDDLE OF THE SCREEN" when started, instead of following STANDARD WINDOWS PRACTICE of start up in the location and size the window was when it was LAST CLOSED.

 I get TIRED of having to REPOSITION AND RESIZE the bloody wallet every day or two due to the combination of the last 2 issues, AND having to re-UNLOCK FOR STAKING every time.

 Why doesn't the wallet AUTOMATICALLY stake instead of having to do that stupid "unlock for staking" every bloody time the wallet is started up?
 It gets OLD having to "unlock for staking" every bloody time the wallet is started up.

 Chips VARY - it's inherent to the semiconductor manufacturing process, and has gotten worse with each generation of process as the size of the "features" on the chip gets closer to the size of atoms the chip is made out of.

 ZEC isn't memory intensive, unlike ETH - try dropping the overclock on your memory ENTIRELY.

 Late Febuary/early March timeframe for AMD RX 470/480/570/580 cards as I recall, about a month later the price on those got so high that the NVidia GTX 1070 and then 1060 cards started getting affected as well.
 Look for when the big ETH price spike started, add about a week for the initial GPU price climb to start, 2 weeks later the shortage was in full swing and the prices were going through the roof - then ballpark 5-6 weeks later is when NVidia started getting noticeably hit on the 1070, add a couple more weeks and the 1060 started getting hit noticeably.

 Never had a noticeable direct effect on the GTX 1080 or GTX 1080ti (By the time the good NVidia ZEC miners started showing up, prices had flattened out and the overall demand was starting to drop) but there may have been some indirect effect from gamers that WANTED the 1070 and decided to "move up" when the 1070 price got into 1080 levels, or due to massive widespread 1070 shortages.

Used Cisco, Netgear Blue Box, and 3Com gear tend to be good - I've got an old 3Com 100BaseT switch that I bought used at least 15 years back, still working reliably.

 I'm distinctly NOT fond of TP-Link or D-Link stuff, had too many issues with both over the years, though TP-Link seems to have done some work on reliability in recent years.

 The large bulk of miners that Bitmain makes DO in fact get sold fairly quickly - days or weeks at most NO MONTHS.
 Most of the rest are part of their Hashnest operation.
 If they mined with them for 2 months then sold them, we would have seen a HUGH hashrate jump 2 months ago on X11 (for one recent example).

 If they WAIT to sell it, they have to DROP the price quite a bit BECAUSE the diff increased in the meantime - and the capitol investment in those miners is TIED UP SO THEY CAN'T USE THAT MONEY TO BUY MORE CHIPS FOR THE NEXT BATCH TO MAKE MORE MONEY WITH.


 YOU are the one that doesn't understand the economics of the situation for Bitmain.
 They don't have infinite cash to work with  - and foundries don't do "credit" with small operations IF they do "credit" at all (I doubt even AMD or Nvidia or Apple get "credit" terms with the foundies - Intel and Samsung don't need to worry about it as foundry owners).

 With one single exception, that's MY tactic.

 Gotta add the HDs to new rigs anyway, might as well make them big enough to be useful on BURST (at a SMALL increase in cost).


 84 drives in one machine is incredible - even Backblaze only manages 60 on their current Storage Pod design (though they've moved to at least 4 and might be moving to 6 or 8 TB per drive so their total pod CAPACITY is likely a bit higher than yours).

 32 TB of drive is slightly MORE expensive (at NewEgg pricing on the Seagate Externals) than most GTX 1080 ti cards - which pull in close to $3/day each at this point.
 Tradeoff is the ELECTRIC usage of that 1080ti will be in excess of 160 watts and more likely close to 200 watts to achieve that income level, while the 4 hard drives are eating around 50 watts AT PEAK and quite a bit LESS most of the time.


 Up side of those Hitachi (HGST) 2TB is they're likely to last a long time, I'm NOT fond of WD in comparison and I'm VERY not fond of most Seagate models for longevity (I don't have a large data set on Toshiba but the set I DO have has been quite negative, close to Seagate low levels).