The massive recent "new folders" fueled PPD increase is harder on income generation than the Curecoin drop from it's spikes.

 On the other hand, without the spikes we probably wouldn't HAVE the massive PPD increase.

 Haven't tried it on any AMD rigs yet, as all of my AMD rigs are LINUX.
 The "nicehash" miner they posted works fine on 3 NVidia rigs I've tried it on though (dual 950 rig, dual 960 + 1 950, and a short period to do some testing on a single 1070 in a rig I was building up tonight).
 All Win 7, Nvidia driver version 472.90 IIRC (NOT the most recent one, that one has MAJOR issues with some stuff I've tried it on so I ended up rolling back to a KNOWN working version).


 950 and 960s are both pulling very close to 10 sol/s, 1070 was pulling ballpark 30, all with right about the same to a hair lower power consumption vs. Ethereum mining.


 ethmine.org went down for about 10 hours - POOL stats showed that, wasn't just you.

Known issue that "genoil's" miner is crash prone.
Unfortunately, most of the code isn't his, and he's not had time to dig into it and fix the issues (or preferably write a NEW miner from scratch that meets his usual efficiency standards AND is reliable).


 The nicehash miner version posted on flypool has been running stable for me for hours (edit - 2 days on the stable machine now) (on one machine, the other machine crashes fairly often running ETH for reason(s) I've not had the time to troubleshoot while moving, so I don't blame the nicehash software for the MACHINE crashyness issue).

 The Nicehash miner (or at least a patched version of it) works fine on zcash.flypoool.org - which is run by the *REPUTABLE* folks behind ethermine.org / etherpool.org using industry-standard STRATUM connect protocal.
 Genoil works when it's not crashing, but the crashyness is a known issue with the "genoil Zcash miner" - hopefully he'll drop that thing and code his own and blow everything else out of the water.


 Nicehash themselves I don't have any use for, their "reported" payouts on their own website on everything I've checked are quite a bit lower than they SHOULD be.

 SO far, I am VERY underwhelmed by everything I've been reading about this software and the associated pool.

flypool working fine on Windows.

 IMO just drop the cloudmining garbage, that's NOT OS -specific and it's all crazy overpriced as usual.

 ZEC is active on Bitttrex (and I recall ethereum wisdom showing at least 4 other activer exchanges for it).

12 c / kwh is a "kill profit" excessively high electric rate for almost any miner.

The KEY to actually making money at mining is to get somewhere with a LOW electric rate - your "right around US average rate" by cryptocoin mining standards is HIGH.


 7c / kwh you might be able to eek out ROI but even that's gonna be marginal.


 Hint - I moved away from Iowa 'cause the 7.5-8c / kwh LOWER RATE IN NON-SUMMER MONTHS was hammering my mining profits, despite using every available trick to lower my rate.
 The 14c+ "3 months of the summer" rate forced me to shut pretty much EVERYTHING down as being unprofitable.

The GPU on an A10 is in the same ballpark of performance (in the current A10 - 78xx line) as the old AMD HD7750 (same number of cores, near-identical clock, loses a little due to the slower RAM involved).

 While it IS useable for mining coins that don't need in excess of 2GB RAM access (forget ETH/ETC and varients), it's not exactly a high-end mining GPU nor it is highly energy efficient.

 Older and lower-end A10s have less cores, and often clock lower, so even WORSE performance.


 No, I don't bother trying to mine on my A10 GPUs for a profit - they do earn some Gridcoin but that's a side-effect of my long-time participation in the distributed.net project, more recently via BOINC's Moo Wrapper project. I suspect the gridcoin they earn MIGHT pay for the electric usage, but it's close if they do and forget every trying to ROI on this.

It goes in waves.

 Intel lost their crown to Zilog for a while, back in the 8080/8085 vs Z80 days - the 8086 and 8088 got it back especially after IBM adopted the 8088 for the original PC.

 Intel held that crown for a quite long time before AMD briefly challenged it in the K5 vs Pentium days.

 Intel then LOST the crown outright for a while in the Athlon vs. Pentium III and early P4 days, but AMD hit scaling issues on the Athlon and the P4 eventually passed up the Athlon.

 Intel next lost the crown when AMD introduced the AMD 64 and Opteron 64 series, but eventually the Core line won it back.

 Dunno what's going to happen with Zen - seems like a lot of the competition currently is more in the "integrated with GPU" lines where AMD kills Intel on graphics performance but Intel wins on the CPU side.

 Say a lot that Intel's current line is ALL "GPU integrated" even though the GPUs involved have poor performance and make the overall cost of the CPUs quite a bit higher.

Might depend on the card, my 1070s seem to need a bit of both for best results but more CPU than MEM.

 Should be almost identical, same fans and near-identical cooling setup (except for the S7 batches that tried to get by with 1 fan, those should have been a HAIR quieter).

The final cure to the "can't get the right spacing" issue turned out to be simple.

 Put a SHORT or a LOW PROFILE card in the 3'd slot, leaving quite a bit of open space for the next card over to draw air in.

 Didn't work on the 950 based machine as such, as all 3 of those cards were short, but I eventually got a pair of 960s and put one of the 950s with them, then the other pair of 950s ended up in an older "only has 2 slots" motherboard-based system.

Can't trust that spreadsheet.

 The GTX 1070 IS NOT a consistent 675k PPD performer, more like 600k based on my FARM of them (mix of Gigabyte and EVGA models at this point).
 150 watts is ALSO not right - more like 170 with the overclocking needed to get *TO* 600k PPD fairly consistantly, per Nvidia-SMI (and backed up reasonably closely by watttmeter readings).
 They CAN be found in a few cases for $400 (usually higher for most but not by a lot, and NewEgg USUALLY has at least one mid-to-high end version on sale for $399).

 Makes me wonder if ANY of the numbers in that spreadsheet for PPD or watts can be trusted, but I can't speak to THOSE from significant personal experience.

The HD 7770 is ANCIENT and was low-level even by the standards of when it was new.
I'd be SHOCKED if it could manage 5 Mh/s on Ethereum, and would be a bit supprised if a system with a HD 7770 could manage to make enough to pay for the SYSTEM electrical usage.

 I douldn't even look at ANYTHING less than a HD7850 for mining Ethereum on, and even THAT card isn't a good performer.

0.42 Bitcoin per TH is going to make this unit a non-starter. That would be WAY too high compared to the S9, and even a bit on the high side compared to the INSANELY OVERPRICED R4.

 Hopefully that price is outdated and will be adjusted.

 As opposed to the original Titan pricing at $9995 (IIRC), and USED ones going for $2000 ballpark with NO warrenttee?

 DO keep in mind that a good power supply will last for a VERY long time, and can almost definitely be recycled to another usage when the miner becomes unprofitable or dies.




 For reference - I paid $800 (appx) for my 110Mh/s A2 units back in November - and they've ALREADY paid for themselves (despite semi-high electric rate usage before I moved in June/July).
 My A2 units are STILL PROFITABLE as well - I just don't have enough power available where I am at now to run them all.

 I don't see the A4 taking forever to achieve ROI, though I do figure the increase in network hashrate might kick their ROI timeframe into the "ballpark a year" range.



 In theory, you could run each "blade" of an A4 from a seperate power supply - just don't try to use 2 different power supplies on the same blade - and it should work fine.
 Optimally, though, a big enough PS to run the whole unit form one supply would be best.

 Given the STATED "at the wall" consumption mentioned earlier in the thread by the guy that already recieved his unit, my favorite Seasonic X1250 or the widely-used EVGA 1300 G2 power supplies should run an A4 unit VERY comfortably - my 110 Mh/s A2 units consume 1240-1260 watts "at the wall" for the ones that have X1250 transplants and have been running rock-solid for months now.


 Keep in mind that the rating on a power supply is it's rating for OUTPUT, not how much it is limited to on input from your AC source. Given a typical 90% or so conversion efficiency on a gold-rated PS, you have at least an extra 10% margin to work with when looking at "at the wall" figures vs the rating of the PS itself.





 The days of "2-3 months to achieve ROI" may have been gone in cryptomining for years, but it can STILL be a very good and high-return investment if you do it wisely - and have low electric cost.

 1 Year for ROI is NOT unreasonable now, since the ASIC "state of the art" has now caught up with overall semiconductor "state of the art" and we can anticipate product cycles moving to MULTI YEARS apart instead of well UNDER a year per generation (for Bitcoin which has more competition, Scrypt cycles have been a lot slower due to the much smaller market for Scrypt miners).

 Which is why I specified the LOW hash S9 batches for an X1250 or 1300 G2. They're WAY too marginal for the high-hash batches.



 IMO trying to run an S9 in a DORM is a bad idea - going to be pushing the circuits way too hard and way too loud.

The EVGA 1300 G2 or Seasonic X1250 would be enough to run one of the LOWER HASH (11.x TH) S9s, I wouldn't even think about anything less as you will likely fry the PS or miner, and the setup flat out won't WORK at best.

 Don't even try those on one of the 12TH+ S9 units.

 It's a matter of the MINER needs that much power, not that you can try to run it on less.

 If it wasn't for the CRAZY high pricing, I'd recommend you look at the R4 instead which probably WOULD work on a good 1000 watt PS.



 Other option is to use more than one PS per miner, but that gets complicated on the "feed them power" side.

With proper cooling and good power supplies, not an issue.

I've got a few rigs running AMD A10 78xx series APUs with multiple NVidia GPUs, or with dual NVidia CPUs and an AMD HD7750.

 The most of the CPU cores are working Gridcoin (BOINC / Moo Wrapper / Distributed Net), as does the APU on the AMD A10 (and the HD 7750 if present, Moo Wrapper doesn't need a lot of RAM, and AMD GPUs of ANY size that can do OpenCL at all are VERY good at Moo Wrapper / Distributed Net work).
 The NVidia GPUs are mining something else, depending on which machine and which GPU model is involved.

 At one time I had similar setups (or the same in the case of 2 of the older such rigs), with the CPU cores running the Mersenne Prime (Prime95) program, the GPU on the APU running Distributed.net, and NVidia GPUs doing something else, for *3* seperate activities on one machine. Needless to say I didn't try to do ANYTHING ELSE on those machines as they were very loaded down with the crunching activities.

I was ready to commit to a pre-order buy of 3 of these a couple months back.

 Sadly, the delays caused me to put most of that money in another direction as I couldn't afford to wait.


 Long-term I figure I'll end up buying some at some point.


 Anyone in Central Washington want to buy a couple A2 units, drop me a message. I won't ship them but I will deliver if it's reasonably close.

 It depends on the GPU.

 Given my mixed farm, I refigure the numbers fairly frequently, most recently last night.

 For many GPUs, especially ALL AMD GPUs, even with "merge folding" it is still more profitable to mine Ethereum.

 For pre-9xx series NVidia GPUs specifically including the 750 Ti, merge folding is not very good at all but most of those cards aren't good Ethereum miners either. The 750 TI seems to be best off right now as an XMR (Monero) miner, pre-Maxwell NVidia cards are probably best to sell off to help pay for an upgrade to something much more recent and efficient in the same or lower power-consumption range (or a bit more if you have excess PS capasity).

 For 9xx series NVidia GPUs, it depends - lower end ones like the 960 and down (and probably the 970 but I don't have any of those so I didn't figure the details) are ALSO more profitable on mining Ethereum - but it's getting fairly close, 'till the Foldcoin having in a bit less than 2 months looks like it will possibly put Ethereum firmly in the lead again for a while.

 For the 980/980ti and Maxwell Titan, merge folding wins, but mostly because those cards aren't very good at Ethereum mining compared to their FAH PPD production.
 The one up side is that the 980 and 980ti have dropped in price on used cards to the same ballpark or sometimes less than a 1070 new costs, while they are competative on PPD abet at rather higher power consumption (980 seems to be a bit lower than the 1070 on long-term PPD, 980ti is a little better).

 The 1060 3GB and 6GB are a question mark, but overall they're not all that good of folders while they're also not all that good at Ethereum and they should probably be avoided entirely in favor of either the 1070 or the AMD RX series based on the numbers I've seen reported for them so far. Note that I don't personally own any of these cards.

 The 1070, 1080, and Titan X Pascal are better off merge folding. The Titan X Pascal in particular is a VERY impressive folder, but the cost makes it a fair bit less efficient overall than the 1070 and 1080 on a PPD/$ basis.



 This all can easily change in a day if there is a major price movement on any of the respective coins, or in month as more GPU power gets directed at the various coins.

 Merge folding profitability has dropped by almost HALF in the last month alone as the huge surge of Curecoin/merge folders have moved into it after the price jump, that surge has now dropped profitability to about a tossup with where the profitability was at in the week before the price kicked up over 3 cents per CureCoin. It is set to drop by about a quarter instantly at the Foldcoin halfing point in appx. 55 days from now.

 In that same timeframe, Ethereum hashrate has kicked up enough to drop Ethereum earned by about a third for a given card while Ethereum price has been pretty much flat thus dropping the profitability by over a third.


 All of this is also very depending on your electric cost - I have VERY VERY cheap electric where I live now, if your electric is more expensive *ALL* of the profitability numbers get worse than what I see out of a given card.