Between those two choices with free electric, I'd go for the Antminer - much more likely to last a long time. The Thunder ran VERY hot and wasn't efficient, though by current standards the S3 isn't efficient either.

 Ignore the moron talking about using a GPU to mine Scrypt with. You'll never make enough to pay for the card before it dies.
 Figure the best GPU setup generates LESS than 1 MH/sec while eating 200+ watts (most were closer to 700kH/sec at appx. 250 watts for the most EFFICIENT setups back before the Gridseed GC3355 ASIC showed up).
 At current difficulty, it takes a little over 100 MH/sec to genrate one Litecoin/day. 1MH/sec generates about 3 CENTS worth of Litecoin per day with current difficulty and price, YOU figure out how many DECADES it will take to pay back the cost of a GPU that can achieve 1MH/sec even with free electric.


 It's likely the current high-end AMD cards will manage more than 1MH/sec - but they're also going to cost quite a bit more than the old "efficient" setup using 7870s would today. Nvidea cards were never good at Scrypt - it wasn't complex enough for them to take advantage of their higher-compexity compute cores, similar to the reason AMD cards CRUSH Nvidea on RC5 work (but not by quite as wide a margin on Scrypt).


 Innosilicon has ALREADY announced a "next-gen" scrypt miner, due in June/July, with better efficiency than anything else on the market (some estimates have their A2-based units generating more Scrypt hashrate than everything else COMBINED right now). I anticipate a significant climb in Scrypt difficulty starting in late June, though current-gen ASIC gear like the A2/Alcheminer/Titan should still be profitable for a while with cheap-enough electric.

 Car battery AH are figured at 20 hours also, unless that got changed in the last 5 years or so.
 Reserve minutes on a car battery (and most car-sized deep-cycle batteries) is figured at a 25 amp draw.

 Deep cycle batteries are expected to drain down close to zero and designed for that, normal car "starting" batteries are not.

 You need more than just a buss bar to run batteries in parallel - they very enough in manufacturing that you can AND WILL drain one down while another is still nearly full charged. Current equalization is THE WAY - one way or another - for paralleling batteries.

 It would be more viable to run one hashboard off one battery, the other + controller off a different battery, which would keep the draw closer to 25 amps that most car and deep-cycle batteries are designed to easily handle. Then use a proper diode isolation circuit to keep one battery from feeding off the other while feeding them all from your solar setup, something similar to what some high-end "boom car" competitors or some RVs use to have 2 batteries, one to start the vehicle (and possibly run a FEW low-power accessories) the other running the (high-power) accessories.

 Low end "cheap" invertors generally put out square waves, NOT suitable for running most ATX power supplies on.
 There are exceptions, many of them designed specifically for use with higher-power Solar and Wind power installations.
 *ALL* "Grid-tie" inverters are designed to put out sine waves (or very close), those would be fine to power ATX power supply(ies) with.

 #2 to the Neptune actually. Which is STILL a bad place to be.

 Almost NOBODY would farm if the average production cost was the same as the price. No incentive TO farm in that case.


 The COST to mine a Bitcoin right now with current gear is quite a bit less than the current price, unless you have fairly high electric cost.

 The whole theory of "the production cost = the price" makes ZERO sense, most folks AND ALL MAJOR FARMS are in this to make money, and if the cost and the price on the same THEY CAN'T MAKE MONEY.


 The floor at around $400 seems to be more about the major influx of cash in Nov/Dec timeframe when the Chinese markets started accepting fiat currency for purchaces. It has ZERO to do with any "production cost", which is well under $300 and possibly under $200 (if they've replaced most or all of their gear with current gear) for most or all major farms and a lot of home miners. I suspect most of the major farms cost is closer to $200, but I don't have hard figures on the total overhead for any of them (though I have a VERY good estimate on electric cost for most or all of the large farms in Washington state).

 The issue for home miners is achieving ROI on the cost of the MINER, which cost is quite a bit higher per hash than for large farms (some of which buy the chips and build their own miners to save on equipment cost, the rest get large bulk discounts, earlier access, and in the case of the farms run by the miner manufacturers like Bitmain they run at least some of the gear for a while THEN sell it for a large markup making their effective equipment cost ZERO or a negative number).

Given that Innosilicon has finally committed to dates on their A4 based gear, I'd say probably "June/July" since they were apparently working to the same timetable on the A3 and the A4.
Bitfury could still have something hit market by April, though.

 Bitmain - no clue, but they're not selling "used" S7s yet so at least a month, probably at least 2 months for them. and more likely longer.

 Avalon - VERY long time, they said when they released the A6 they weren't even working on the next gen yet. MANY months, more likely a year if they survive long enough.

 SFARDS - *ROFLMAO* doubt they're even going to bother trying, given the huge fail of their SF100 thing.

 BW.com - B11 in theory could be sold anytime, but I suspect it's too little too late given the Bitfury chip stuff and BW's own announcment of their next-gen chip (with an announced timeframe IIRC of ballpark mid-summer, but keep in mind the delays from their announced B11 original announced original delivery timeframe - I figure LATE summer more likely).


 Who else is left?

I don't think you understand the DC/DCless part.

It's JUST like any other string design, like the Antminer S5, from the sound of it.

 In theory, you COULD use a single buck and run a string from the buck to get voltage control on the string, but I dount anyone will impliment that sort of a setup.

As of right now, a WORKING KnC Titan is the most efficient Scrypt (Litecoin, Doge, etc) miner in existance - but they had a bad tendancy to have cores die a lot and they're still a bit pricy most of the time.
 The Innosilicon A2 and Alcheminer were a tossup for next most efficient, with a TON of A2 units still hashing away with very good reliability - the Alcheminer folks didn't sell nearly as many units, but it seems to have good reliability as well. There were some issues with some pools not getting along with the early Alcheminer software, but those issues seem to have been fixed over time, and some pools never had an issue with them (the early software needed a quite HIGH difficulty to work well that most pools couldn't provide).

 Everything else was quite a bit less efficient, to the point that you need VERY VERY cheap electric to make any profit at all with other Scrypt ASIC miners, and the early Gridseed GC3355 based gear can't make a profit at ALL unless you have free or essentially-free electric.


 Trying to mine Litecoin with a GPU became unprofitable over 2 years back, even with VERY VERY cheap electric, and it's probably never going to achieve RoI on anything GPU based today before your GPU dies even on FREE electric.

 Any coin that is based on an algorythm OTHER THAN SHA256 (Bitcoin, Namecoin, etc) or Scrypt (Litecoin, Doge, etc) and POSSIBLY X11 (Dash, etc) should be mineable for a profit with a GPU, if it can be mined at all with a GPU.

 X11 is a question mark right now, there IS an announced "available for sale" X11 ASIC-based miner, but it's not been reviewed by a reputable site yet (it's had ONE review that I could find, but the reviewer was a newbie with no reputation), so the actual existance of the thing is in some doubt.
 It IS being sold by the DualMiner folks, who DO have a reasonably good reputation for delivery on what they claim to be selling, but the claimed manufacturer "consortium" has ZERO previous track record before they announced their miner.

https://bitcointalk.org/index.php?topic=1417712.0

 Your announcement is redundant, though 'till they post in this forum rather than the massively overloaded "announcements" forum that you can't find ANYTHING in, your post does serve a purpose.
 Says a lot that I found this thread 4 days before I found any link to their direct announcement thread (via an accidental find on a Google search, forums search here STILL can't find their direct announcement).



 As far as the no DC/DC converter part goes - that means the ONLY way to control voltage to the chips is to vary the input supply voltage, which basically means ZERO ability to control voltages in software. Same as the Antminer S5, which is part of the reason the SP20E was capable of MUCH better efficiency if set that way in software even though the basic efficiency of the chips used in each was about the same when those miner models were hashing at the same rate.

There is only one viable reply to such an assertation.

 *ROFLMAOSC*


 Etherium has ZERO inherent worth 'till it can convince vendors to accept it (the same applies to any other cryptocoin).
 Bitcoin is so far ahead of ANY other cryptocoin on that front that it's lead is probably untouchable (and the only coins even in the same PLANET are Litecoin, and perhaps Doge and Dash - NO other coin is in the same ballpark).

Nice specs, but only about half the efficiency mentioned in your early announcement back around November (850 MH at under 1000 watts IIRC in TURBO mode, 600 at 720 or some such in non-turbo).
Might be a good thing through for those of us with existing A2 units, should give us more time to achieve RoI before the hashrate goes through the roof.

 Is one miner going to be able to achieve all of the modes listed in your post on your website, or it that going to be determined by how many chips are in a string on the board / voltage on each chip?


 BTW - this probably would have been better posted in the Altcoin Mining forum here, the only way I can FIND this posting is via a Google Search - or it needs a cross-post over there.

When Litecoin had it's halfing last year, the price didn't start increasing 'till about a month and a half (more or less) before the halfing was due.

 If Bitcoin follows that same pattern, the increase won't be due 'till early June.

You can try mining Litecoin with them, but unless your electric is VERY VERY cheap you won't make anything.

 LiteGuardian pool should still work fine with the Gridseeds (I ran my "80 blade" units on that pool for a long time, using the fixed diff server option, I believe that one still can go as low as 64 on diff setting).


 The Scrypt coins with lower difficulties also have the issue they aren't worth much if anything.
 Definitely ignore Doge by itself, it's pretty much all mined via Litecoin miners doing merge mining since it went to being merge-mineable.


 I never had the Orb, just some of the "80 blades" - your Orb can be pushed harder for long-term as it was limited by the mining chip itself, not by a poorly-designed easy to kill voltage conversion section that was EASY to overheat. I believe the Orbs were typically eating 5-6 watts on Scrypt-only mode (my blades had 40 chips per side and ate a bit over 40 watts per side where the Orbs were 5 chips per Orb IIRC).

 The 5-chip orbs got totally unprofitable for mining SHA256 YEARS ago - the blade units were designed for Scrypt-only 'cause that was the only thing the chip could mine even CLOSE to profitably by the point the blade units were being designed.


 Right now, any Innosilicon A2 based unit, any WORKING KnC Titan, or any working Alcheminer can mine Scrypt profitably even at 15 cents/kwh - but it's iffy if you'll manage ROI on any of those before the A4 based gear starts showing up unless you can pick the older gear up CHEAP. On the other hand, if your electric is cheap enough, the A2/Titan/Alcheminer gear should continue to be profitable for a while, as it will take a while for the A4 gear to sell enough to kick the difficulty up a lot (I figure my A2s will still be profitable at almost 3X the current network difficulty in the 5-6c/KWH rate range).

Why are 2 of those links "mail to" links?

Just the wallet you synched - but you COULD copy that updated copy of the blockchain to a wallet on a different machine and bring IT pretty much up to date, if it's the same client for the same coin (and as I understand it the client doesn't have to be the same-OS version, just the same CLIENT version, AKA you could update a Bitcoin 11 wallet on a Windows machine then copy over the blockchain files to a LINUX machine running Bitcoin 11 to update that client).

 Innosilicon is still around, though they've been mighty quiet about status on the A3 and A4 the last couple months. I suspect that one or both of those could be "any day now" given how little in advance they've tended to announce when they're releasing their chips/gear in the past. They also have a substantial existing chip design business outside the cryptocoin world after all, so they can afford to not "crowdadvancesalefund" their development, plus whatever they made from the A1 and A2 (LOTS).

 BW.com (LKetc) is still around, though status of when (if) they're going to start selling the B11 is unknown, they DID announce the next-gen stuff that should be able to compete pretty closely against the Bitfury stuff and it's pretty obvious that the B11 is actually in production.

 Avalon ... meh, their announced stance on the 14/16nm generation might very well manage to kill them as they're WAY far behind the development curve at this point.

 I don't see Bitmain getting most of the pie for much longer.


 Do remember that hope was the last thing out of Pandora's Box.

A4 has not been released yet.

 In theory, it COULD be released literally any day now - but no word out of Innosilicon recently.

People have finally wised up to the fact that the S7 is not going to achieve RoI unless you have VERY VERY cheap electric - and that has always been the case, and that fact just keeps getting worse and worse (unless the current lull in hashrate climb manages to last about 3 months LONGER than I currently expect).


 I suspect Bitmain is having their current fire sale in the hope of getting their money back out of the BM1385 chips they've already bought before they become almost worthless due to next-gen miners blowing them out of the water - and I have some doubt that Bitmain was expecting Bitfury to go public on THEIR next-gen chip much less start pushing it to anyone OTHER than huge farms.
 I get the strong impression Bitmain's answer to the new bitFury chip hasn't even taped out yet, or taped out recently enough they aren't going to have a new miner available before the halfing - leaving them rather behind the curve on what should prove to be THE most important ASIC generation in Bitcoin mining history (other than perhaps the original generation gear).


 There's also the BW.com/LKetc second-gen chip for them to worry about, the question of WHEN Innosilicon will release the A3 (which in theory could be literally ANY DAY NOW)....

SP20E will hit the same hashrate as a S5 using a hair less power (in general, units vary enough that SOME SP20E might need a hair more power than the best S5s out there), but are A TON more flexable on balancing efficiency vs. hashrate, and will pull a LOT more hashrate if you're willing to pay the price of lower efficiency.

 Both seem to have good long-term reliability if the temps are kept reasonable.

 SP20E is a bit quieter, but that's not by a lot. You don't want to park either right by your bed.

Haven't worked with the Corsair RM series, no opinion.

I'm running some MSI motherboards - the ones with the heatsinks on the PS circuitry seem to be decent, the others are too cheap to even look at for a 24/7 machine. I don't have the model numbers I'm using handy, but they're microATX so you wouldn't want the specific ones I have anyway.

EVGA G2 series is good, their other series are marginal at best.

 Silverstone is pricy, but tend to be good.

 Antec and Corsair .... vary. Some are good, some are overpriced junk.

 Seasonic, if you can find them, are good.