It's ENTIRELY practical to set it up as a profitable business on a LOT less than "millions" - just not HUGE scale, and you have to mind the costs a bit closer than the big-scale folks can get away with.

 $100k would be PLENTY to start up a small professional mining farm BUSINESS - I'm pretty sure I've got less than HALF that invested INCLUDING reinvestment of a lot of my mining income for the 4 years or so I mined "as a hobby".


 Cooling costs had better be a LOT less than 50% of your electric use - you can do mechanical air conditioning for less than that, and THAT makes costs quite a bit too high to be truely competative.
 If you're putting much more than 10% of your electric usage into cooling, you need to rethink your cooling setup.

 Don't count on getting a business loan though, Cryptocoin mining is a LOT too speculative for most banks to even think about dealing with.


 For a small farm, you don't NEED staff.

 Water usage should be very small, unless you are using evaporative cooling - then it's STILL pretty small.

 Idaho electric rates aren't all THAT low - for low electric rates in the USA, you're looking at 3 specific counties in Central Washington state.
 Idaho might be a good bit less than 11 cents/KWH in areas it's got hydropower access though - but it's certainly not in the 3 cent ballpark.

 Well gee, Bitmain IS a Chinese company....

 I suspect the largest part of the cost of the first few batches of the S9 was the design and implimentation cost of the chip, since then I'm quite sure that the costs have dropped quite a bit (more so than the sales price) and Bitmain has been raking in lots of money to put towards their own mining farm *AND* eventual design of a "next generation chip".

Dash hasn't been CPU mineable for a profit for YEARS now - it went GPU 3 years or so back, then ASIC last year.

 Monero might be mineable at a profit though, it's proven to be pretty CPU friendly - but you might want to make a point of getting a current-generation or at least "on the current process node" CPU if you want to be able to mine it profitably for more than a very few months.

 Other CPU mineable coins I can't speak to.

 For Global Foundrys, there is also the "by terms of contract" amount of their foundry capacity that IBM and AMD have first call on, dating back to when IBM and AMD spun off their foundry operations into what became Global Foundrys.
IBM and AMD have absolute first call on a certain amount of foundry capacity (they also have a REQUIREMENT to use a certain amount of capacity as a minimum) that GF can NOT avoid without incurring some rather large contractual penalties (I believe there is an "acts of god" provision in those contracts however).

 You are still sucking that 12000 CFM of air in from the outside from SOMEWHERE, or you get no airflow.

 Setting it up as INTAKE ventilation lets you have more control over what the input air is doing, and where it comes in at, and allows for the possibility of filtering it.

 I've never understood the preference for "exhaust" fans over "intake" fans except in certain very narrow circumstances which don't generally apply to mining.



 I suspect the air exhaust out of your miners, even if the intake air was at 80% RH, would be well under 40% and likely so low you can't measure it with a "common" humidity meter.

 I can't remember the last electronic device I've seen that wasn't specified to work at 80% RH - many go to 90%, and while they do tend to specify "non-condensing" that's NOT an issue with a miner that's inside a building.

 I believe AMD is still aiming for late 2'nd / sometime in 3'd quarter release on the first VEGA cards - but no new news on them for a while, so looking more like 3'd quarter sometime.

 It IS interesting that NVidia chose to reset the pricing on their high-end cards very recently though - part response to the release of the 1080 Ti I'm sure, but I also get the strong impression that they are doing a "preemptive strike" on pricing vs. Vega in the range it should land in.

The "case" setup on the Gridseed 80 blades was specifically designed to channel maximum airflow through the heatsink assembly - opening it up won't let them run without a fan unless you're running them at a VERY low clock setting, as they need the airflow over the power supply parts to keep THEM from frying.

 The limit isn't the GC 3355 mining chips on those units, and never was.

 I believe the fan you ordered is the SAME model I used as a Gridseed replacement fan, it's very very close if not and should prove to provide plenty of airflow.
 I'm also a long-standing fan of NMB fans on lower-flow needed setups, they were widely used in servers because they last a very long time with no maintainance.
 NMB doesn't (or at least they didn't used to) make super-high-flow fans like the high end Delta units and the rare Delta competition, so they're a bit less known to miners in general.

24/7 operation or forget about making any significant income.

 12 hours a day, you are almost guarenteed you will never make back enough income to pay the cost of the miner before the miner becomes unprofitable.

 Have you tried any of the newer versions OTHER THAN 16.12 series, which had KNOWN ISSUES on all cards in a mining setup with poor performance?

 There's a reason I specified the 16.10.x series (and the WHQL version in particular).

 The GC 3355 doesn't use a lot of power mining Scrypt, so it doesn't need a lot of cooling. 1 watt or a bit less per chip - the heat sinks on the "80 blade" units are actually severe overkill for the actual heat dissipation they have to handle.

 It *might* be possible to refit the heatsinks with Sidehack "pod" boards, but since he seems to be putting heatsinks on both sides of his "pod" board designs it seems unlikely a retrofit would work - might be doable if you tap new screwholes in the right place of the Gridseed heatsink assemblies, but kind of a waste.

 There is no such thing any more as a "cool and quiet" miner, it takes way too much computing ability to be able to compete which means quite a bit of power soaked and quite a bit of heat. I think the closest to a "quiet" miner that is currently competative is the Bitmain R4, but even that thing isn't particularly COOL.

Central Washington - and the evap works quite well, most months we only see 2-3 days it precipitates at all and I rarely see 40% humidity 5 feet from the OUTPUT of the evap.
 I'm thinking I might need to get a second one at some point once I move into the new place I should be relocating to in a month or two (better ventilation than where I'm at now AND twice the power availability).

 Semi-arid biome, bloody near a desert - I think we average 10-12 inches of rain/equivilent a year or thereabouts.
 For an evap to work noticeably better, you'd have to move to an actual desert like the Mohave or Death Valley.

 There's a reason the Grand Coulee Dam was built primarily to irrigate the region I'm in (the power generation was a LESSER design function in it's initial buildout, a lot of the power generation was added in later "upgrades").

 Overkill but it'll work - and if you get a couple more cubes you already have power for them.

Vega won't be cheap - it's aimed to be AMD's new "HIGH END" card line, slotting in ABOVE the existing RX 580 at the Vega LOW end, and aiming for the 1080 Ti and the Titan X Pascal on the top end.

I'd guess you won't see Vega cards under $300 at all and likely $400+ range for all of them for the first year or so they exist.


 Keep in mind that VEGA is specifically AMD's "next generation upgrade to the Fury" line HBM memory and all.

Run the "precise" test sometimes works where the shorter tests flat out refuse to work, especially on algorythms like ETH that have a long setup time - but sometime you still have to run the test 2 or 3 times before it will work.

 I have not used a 1080 TI.

 With that said, I'm quite happy with my Gigabyte 1070s, the Gigabyte 1080 I bought last week, and am reasonably content with my EVGA 1070s.

 Also, I don't buy the numbers on that table - I was commonly seeing more than 600k on my single-card 1070 rigs (thought they were all Gigabyte cards with a fairly high factory OC) and am NOT seeing 800k but more like 750k out of my Gigabyte 1080 (which replaced a Gigabyte 1070 in one of those single-card rigs).

 In terms of raw card stats it makes sense - both cards are probably maxing out the "fast return bonus" and the 1080 has 1/3d more CUDA cores than the 1070 it replaced but the memory is only 20% faster on raw transfer rate and has higher latency.

 1080Ti is a LOT more cores though, and I think they managed to clock it a bit higher yet.

 Win 10 is in no way shape or form an "upgrade" for a mining rig.

 Even fan-type cards exhaust a LITTLE air out the back of the card, just not as much as a blower-style, so you want to work with that flow and bring air in from the front and exhaust it out out the back.

 Dehumidifier = mechanical A/C = LOTS of power consumption, and NOT NEEDED if you take the air comming out of the miners (which will ALREADY be pretty low humidity) and evap cool that then send it back to their intake.
 As long as you have more airflow THROUGH the area than you have going through your evap setup humidity buildup should not be an issue.

 The temperature of the water being evaporated is an INSIGNIFICANT factor in the cooling - the amount of heat to evaporate one gram of water is appx. 22 TIMES the amount of heat needed to raise that gram of water from freezing to boiling.

 Best rate in Washington state on an "all up" basis is closer to 3 cents than 2 - though it IS under 3.

 I suspect the only areas that see a LOT less in China are way out-of-the-way near an underutilised hydro dam complex - but getting net into those areas likely isn't all that hard, given a significant probability of existing mining farms already in those areas.

 I don't think Iceland gets that low - but they DO have a very low cost of cooling given their climate.



 To GMPoison - don't sweat humidity, the miners don't really care about it as long as water isn't condensing on them.
 Since they generate a LOT of heat, the air comming out of them is going to be pretty dry.
 Think of them as a low-density hair drier.
 Volume of airflow should be plenty, but keep in mind derating for temperature on your electric wiring and breakers when you plan THAT part out.