My current Ethereum mining rig cost me equivilent of a bit less than 2 Bitcoin (not counting the power supply, that would probably put it a bit OVER) - but it wasn't intended originally for Ethereum so it's nowhere near optimised for mining.

 20/20 hindsight, I'd use a different motherboard and case - I've found a very few motherboards that DON'T put a PCI-E slot right at the edge of the motherboard (limiting that slot to a single-width card in every case I've seen to date), or they have 4 total slots spaced out so that 3 of them are fully usable for dual-width cards (abet a bit tight on cooling clearance) AND that 4'th slot that's only single-width-card usable.


 It's too bad nobody ever made any single-width card more powerfull than some of the AMD HD 7750s (especially the low-profile Sapphire card), or that case makers aiming at "high-end gaming systems" don't pay better attention to making that end slot USEFULL and leave enough space and the slots to mount a dual-width card in that last slot BY DESIGN.


 If anyone knows of any such cases, feel free to post/send me links!

If the sensor is on the chip like the Spondoolies "Rockerbox" based gear (SP20E etc) those temps aren't that bad.

 If it's on the board, DOWNCLOCK NOW OR BURN.


 And no, AFAIK they didn't include a fire extinguisher with original sales of the Neptune - but they should have.

Same reason I'm not selling my A2s yet. They'll be profitable for quite a lot of difficulty increase - and depending on the price the A4 stuff finally gets released at, diff increase to the point of making my A2s unprofitable might take months or even a couple years.

 8-)

My Litecoin mining, back in the last 5-6 months before the GC3355 was introduced, paid outright for my Sapphire 7750, 5x HIS 7750s (of which one has died outright, the other 4 have had fan failures within a year of purchase, the other 4 are now running with a single Delta 60mm screamer twist-tied to the HS 'cause I had several left over from my overclock Athlon Thunderbird/Palomino days), and a couple of higher-end 7xxx series got mostly paid for before Litecoin got unprofitable for GPU mining at my electric cost.

 Up side - I was going to buy most of those cards anyway (AMD cards are GREAT for RC5 crushing, NVidea does very very badly on that particular usage).

 Down side - hmmmm, can't really think of one.


 It mostly worked for me because I already HAD another use for the cards, the Litecoin GPU mining was more about "buy more sooner than I could have afforded to otherwise" than anything else.



 And now history repeats itself - I was going to build up my current rig anyway as a multi-cruncher (RC5 on the A10 GPU, Prime95 on the CPU cores, Folding on the GTX950s) but I'm mining Ethereum for a while on the GTXes instead to help pay for them and perhaps a fair bit of my planned "next multicruncher rig".

 8-)


 For Ethereum, right now the best cards on a hash/power basis are the AMD Fury line, especially the Nano. My GTX950s aren't bad, but at about 10MH each while soaking 90 watts or so they're not directly competative (nor are any of the other Maxell cards ESPECIALLY THE 750Ti) with AMD 78xx/79xx and later rebadge/upgrade cards.

 I'm NOT sure how the 7750 would compare, but I suspect it has the same "not enough RAM" issue the 750Ti has. I might experiment with one of mine at some point, since I've already got a few....



 When a 2 Gig card isn't enough any more, or Ethereum goes POS, a-Folding I will go (unless another good-profit-for-a-short-while type GPU mineable coin shows up).

 Ethereum is well known to not work well on the 750Ti, it doesn't have enough RAM or some such issue. The GTX 950 is about the lowest Maxwell card that works well with it (I get appx. 10Mh out of each of my GTX950s) but Ethereum is a coin that prefers AMD hardware.

 X11-based coins might be profitable on your machines, like DASH, but your machines really aren't optimised for mining (too much CPU that soaks a fair bit of power, very little GPU horsepower).

 Is there a reason other than cost that you are running the lowest-end of the Maxwell cards?

Ethereum is such a small market it's pricing has NO effect on bitcoin pricing.

 Unfortunately.

 Nah, I saw the moonlanded - the Alcheminer chip was, from what I've seen of it, about the SAME efficiency as the A2.
 I just don't like the limits the USB/board type design put on the chip for power and cooling.


 Given Innosilicon hasn't even released the A4 chips yet, how can you say how their miner design is going to look?
 Their *A2* designs has bucks, but they have stated the A4 is designed to work in a string setup - I don't recall that being an A2 option.



 Last announcement on Innosilicon's web site was "June" timeframe for the A4 chips to start getting shipped, "July" for miners, as I recall.

 The BIG question is "are they sticking with A3 release at the same time" that they stated in one of their early mentions of both chips?
 The NEXT big question is "how good will the A3 be when they release it"?

 I might be misremembering, but I'm pretty sure that mine got squirrly and did wierd voltage things if I tried to set the start voltage higher.

 Auto on fan speed I class as worthless.

 Fan speed seems to be a matter of taste thing - YS Tech doesn't have quite the rep for longevity that Delta has, but I've had very good luck with them lasting a long time in previous experience with them. For me it was all about "max the cooling so I can push the hashrate as hard as I can" when my electric was relatively cheap....

 Bitfury has 3 videos showing their new chip, including specs. No miners yet though.

 Innosilicon has released specs for their A4 Scrypt miners and chips, but not yet on the A3.

 BW.com has released specs on their B11 (apparently in production but not for public sale yet) miner and it's chips, and estimated specs on their next-gen chips, but the current miner isn't any better on specs than the S7.


 Date and price is definitely speculation, but it's looking pretty solid that SOME new miners will be getting released before the next Bitcoin halfing in July.
 I would be SHOCKED if nothing "new" showed up in Bitcoin miners before fall - technically the B11 would count as it's not being sold to the public yet, but I would still be shocked if it was the ONLY thing that showed up for sale before fall.



 Oh yeah - the X11 miner showed up for real (one of the big review sites had a review on the one they got), but only a small quantity in the first batch - no idea when their next batch will go on sale.



 S7 won't be dead after the halfing, nor will the B11 or Avalon 6 - but their profit will be totally reliant on VERY VERY cheap electric, and small even then, unless a LOT of earlier-gen miners get turned off for being unprofitable.


 Up side on the 14/16nm full custom generation - it CAN'T get outdated in less than a few years, as it will have reached the current semiconductor "state of the art" - this is THE BIG CHANGE from anything that came before. There might still be room to tweek designs for smaller gains, but those gains if any will be much much smaller than a "new generation" has generally demonstrated.

You have a 4'th option.

 Wait

 The new "next-gen" miners should start hitting the market in the next month or two.

 In the number, no.

 In the SCALE, there's at least one and I'm pretty sure multiple - MegaBigPower and Zoomhash both run some serious-size farms in Washington.
 They're just pretty quiet about that in recent years.


 I DON'T know of MBP is still anywhere near 5% of the TOTAL Bitcoin hashrate any more, but that was their target and they were mightly close to it for a long time.

 Zoomhash seems to be more diversified - the A2 gear they were selling off 'till a week or two ago seems to have come from their farm, as the ones I bought some months back were shipped out of the Wenatchee area. No clue if their farm is anywhere near as big as MBP.

Wrong way around, start has to be LOWER than the other voltages.

 I always ran my fan at 100, but noise wasn't an issue - max cooling was.

 Even at 100 my SP20E was quite a bit quieter than my S5s were - the "focused flow" fans that the S5 uses are capable of better CFM against resistance than the AFB-type in the SP20E but the price is that they're louder.

16 Gig total ram is pretty cheap any more, can find it under $70 on Newegg even pushing faster ram speeds.

 It's probably overkill for current mining, but can be nice to have it for later repurposing of the machine(s).

 Unfortunately, most of my "old" motherboards are microATX with one PCI-E slot, otherwise they'd probably work well.

 Isn't that the area code for San Diego area?
 You should move....


 8-P

I wish Sidehack wasn't anti-Scrypt, I'd love to see what he could come up with for an EFFICIENT Scrypt miner using the A4 (I'm betting Innosilicon and their integrators go for the short-string higher-hash less efficient options).

On the other hand, if Innosilicon actually puts a buck on each string for their A4 based miner(s), it becomes a mostly moot point.

I turned my furnace on twice last winter - got a bit nipper than I thought, so fired it up for about 3 minutes each time to get my place back to comfortably warm. Never did manage to shut all my windows at any point, had to keep SOME air flowing to avoid overheating the place (if we'd had near-record cold temps in Jan or Feb I MIGHT have had to shut all the windows. Maybe. Not sure.).

Getting all 3 cards to cool will be tricky - it would be nice if someone would put out a MB with a 16/1/1/16/1/1/16 or similar configuration to FIX that issue, but too many current motherboards want to put a PCI-E 1-bit slot closest to the CPU thus forcing 2 of your GPUs to be right on top of each other.

 I am seriously considering water-cooling for at least one GPU when I have the money to start building serious numbers of machines.



 Overclocking ANY laptop is a really bad idea - they are NOT designed with good cooling to start with, you're almost guarentteed to have a FAILURE well before the warrenttee runs out - and overclocking probably WILL void that warrenttee.

Yes, transaction fees might (are likely to IMO) increase over time - but RIGHT NOW they are insignificant compared to 25 full Bitcoins (or even the 12.5 after the next halfing) for the block reward.


 They might add up over time - but it would take a LOT of blocks to get ton ONE bitcoin on transaction fees alone at this point.


 To get back to the OP, the cost of mining has ZERO significant effect on transaction fees right now, but that will eventually change as the block rewards keeps getting halved - most likely won't be significant for a decade or more though.




 Move to an area with VERY VERY cheap electric, then wait for the 14/16nm Full Custom machines to start getting produced - or look seriously into various Altcoins for the short term. Possibly both.

Innosilicon has already announced miners due in June/July timeframe on the A4 (the one they wanted partners on 'cause they weren't sure it was worth doing, NOT the A3 they apparently didn't want or need partners on) - and given they stated they expected to have BOTH out at the same time, I'd be watching them VERY closely indeed.

 On the down side, the last announcement that included date estimates on the A4 also showed it to be a fair bit less efficient than their original estimate - implies that the A3 might also have ended up in the S7/B11 efficiency ballpark, which is NOT a good place to be right before the halfing on a new unit....