Why bother with a pool for a coin that went MERGE MINED years ago?

My experience is that an area that has frequent power cuts like that isn't worth mining in.


 Most UPS that are "inexpensive" are only designed to have a runtime of 5-20 minutes - so you have to add a lot of battery to extend that to the 2+ hours you want, by which point the UPS is getting to be a fairly large percentage of the cost of your mining rig itself - and you aren't going to find an "inexpensive" UPS that can handle more than about a 4 card RX4xx/5xx rig at all.

 Keep in mind that if your outages are FREQUENT, that it takes quite a bit of time to RECHARGE the batteries after an outage = more capacity = more runtime but also = more RECHARGE time.
 If you're losing power for 1-2 hours a DAY a UPS is not going to help for very long, as it won't have the TIME to recharge the batteries fully between outages.

 I'd figure on an APC 1500 (which specific model doesn't matter, they all perform pretty close to the same - and have about an 850 WATT limit on how much you can back up with them, 1500 is their VA rating and ASSUMES your power supply does not have power factor correction) and add 2 car batteries to that (get a PRO to do it if you're not a qualified electronic tech, it's not hard *IF* you know what you're doing but you do need to KNOW what you are doing).

 The reason I specify APC is that I've worked with them for DECADES, and ALL of them have "current-limited" charge circuits so you can safely add more battery capacity to ANY APC model without any chance of overloading the charging circuitry. My current router/firewall appliance (and a "florescent for when power goes out so I can SEE" lamp) currently are connected to an APC 200 that has a car battery on it - and has a runtime measured in DAYS with that light of a load on it - but also would need probably a week for a full recharge if it ever DID have to run for days.

 At that point you're probably looking at $350 rough ballpark, presuming a new APC (don't try to put USED car batteries on such a setup) - might be more like $400-$450 depending on how good of batteries you put on it, and keep in mind I've not priced out UPS units OR batteries for a while so I might be on the LOW side on those numbers.


 Now figure out how long that is going add to your ROI.


 Now figure that you are going to have to replace those batteries every few years - they SHOULD last for their warrenttee period or somewhat longer, as a UPS doesn't load them nearly as hard as starting a car does.


 You can probably get by with leaving the monitor on the system NOT connected to the UPS - this will reduce the load on the UPS a little, which will help battery life AND will help UPS longevity a little.


 This is why running a mining rig on a UPS is generally considered to be a BAD idea and a waste of money.

 Not even close - even CELLULAR provides have higher limits than THAT.

 BURST mining doesn't use a lot of bandwidth - the only thing you are sending are a nonce and a bit of "who is sending this nonce info" when you find one with a low deadline, the nonces aren't all that big.

https://www.newegg.com/Product/Product.aspx?Item=9SIA73M5RB6419 lowest cost for an IN STOCK RX 470/480/570/580 on Newegg right now - but the price gouge is INSANE, they want more for that than for some GTX 1080 cards.

 NVidia gotta be LOVING this....



 The Vega Frontier edition is supposedly going to be out in June or early July - but that's the WORKSTATION version, not a consumer one, and will probably be priced accordingly.

 The consumer Vega cards are due later this year, apparently.

Seasonic X-1250 or EVGA G2 1300 would probably be overkill for a 6x RX 4xx/5xx rig - but I'm not sure the 1000 watt versions of those would be ENOUGH unless you undervolt and drop TDP, and even then it might be a little risky when you first boot up the machine and your undervolt/lower TDP software hasn't kicked in yet (undervolt via BIOS would not have that risk).

At recent BURSTcoin profitability, he should pay off those HDs in 6 months rough ballpark, perhaps less if he got some LOW COST per TB drives like the Seagate Archive 8TB model - and his electric usage is going to be VERY low, BURST is about the only coin that doesn't lose significant profitability when your electric is over 20c/kwh.

About the only way you will be able to consider DIY is to have million(s) to spend with Bitfury to buy a TON of chips at a time, then pay quite a bit more building miners out of those chips - or spend MANY millions on getting your own chip designed and made by a foundry then spend lots more making miners.

 *OR*

 pay the millions Bitfury wants for one of their immersion-cooled BIG mining containers.


 Sidehack is a special case, with a long reputation for building small miners and better connections than you or I have.



 In theory, you could take a current air-cooled miner and convert it to water-cooled, but it would be a nightmare at best.


 

It is VERY rare that a cloud mining site will earn you much if any profit - keep in mind that if they could earn more from just mining, they WOULD be.

 The Bitmain-based stuff has proven to be reputable, but my first statement still applies.

 Don't confuse "pools" like Eligius and Slush with cloud mining sites like Hashnest.
 NOT the same thing at all.

 If you can get through the whole plot on a HD before the next block is initiated, you're fine - otherwise you lose a little.

 Generally though, a SSD is way overkill for the required speed and THEY COST TOO MUCH to be worth using in BURST mining.

The L3+ is the most efficient Scypt miner at this time.

 The folks that bought KNC's assets out of bankrupcy are *talking* about building a new Scrypt miner that will be more efficient - but I'll believe THAT when I see it, KNC has a rather long record of inflated claims they rarely could live up to once they finally managed to get their gear out the door almost ALWAYS way late if at all.

 Out of production for at least a couple months, try FINDING one at all much less at a reasonable price.

 Nice if you CAN find one though.

 

 To crazyivan - I'm not the one with the MSI 1070 cards, mine are all Gigabyte or EVGA and at this time I don't worry about "optimise for sol/s" as I'm on a "fixed price for electric usage" lease.
 (That should change VERY shortly though).

 More figures for your grist mill - I earned 157 Curecoin yesterday which would equate out to *ABOUT* 5300 FLDC.

 Some folks that do FLDC aren't on Team Curecoin, but some Team Curecoin folks don't do FLDC which makes is a bit more complicated than the basic 7470 coins vs 250k coins per day multiplier.
 This works out to about $68 at RIGHT NOW prices for both on Bittrex on 11 x 1070 and a 1080 (my current loadout for my Folding rigs).

 Those same cards would have made about $72 working DGB-Skein or DGB-Groestl, or would have been about $60 working ETH or ZEC (profitability on those on a 1070 is pretty much dead even right now).

 Had I sold earlier in the day, I think FLDC was high enough to get that total to almost $80 - but I don't HAVE that FLDC for a few more days, no telling where the price will be by the time I DO have the coin available for sale.

 (edit - FLDC did a very rare "right after distribution" spike, so I actually made MORE for once).

 If you can afford to hold Curecoin and Foldingcoin for the spikes, you can probably make MORE on high-end NVidia cards via dual-payment folding then via cryptocoin mining right now - but it's somewhat riskier given FLDC only pays out once a month, and the price usually dumps right before or right at the start of the payout.

 Worst case, though, if the current price ranges hold and another 100M PPD don't join the team, it should be close - and it's close enough I can AFFORD to fold for the time being. 8-)


 BTW - for some of us it's not "Triple Payout", it's Quadruple - "bragging rights" are in there somewhere. 9-)

 In my case, I go with A10 APUs 'cause the GPU side is quite efficient on Distributed.Net work (which barely even touches RAM at all as it all fits into CACHE, data too) - and I've been pushing DNet for almost 2 decades now (28 Febuary 1998 - yes I WAS involved in DNet efforts to brute force crack DES and RC5-64, RC5-72 is proving to be a tough nut though).

 CPU cores do the BURST mining, and the discrete GPUs earn the serious bucks.

 

 Not effectively, I'm pretty sure.

 DGB-SHA, DGB-Qubit and DGB-Scrypt are ASIC mined (Bitcoin miners like the S9, the Baikal, and Litecoin miners like the L3+ and A2/A4 respectively).

 DGB-Skein and DGB-Groestl are GPU mined.

 The neoscrypt coins have GPU miners available, but I've never really looked into those to date.

 As far as I know, the only thing worth mining at this time on a CPU is the cryptonight coins like XMR(Monero), at which your Ryzen is fairly good.


 I didn't notice any pool-side improvement on DGB-groestl using the KlausT fork of ccminer vs trpv, and very little if any better on the miner reported rate.
 
 Best to check for at least a 3 hour period though, normal variation could EASILY exceed that 50% variance you say on a 15-minute test.



 Trying to mine scrypt on a CPU has been worthless for YEARS - GPUs for scrypt got replaced 4-5 YEARS ago by ASIC, and that was a year or two after CPUs got blown out of the water by the first GPU-based miner.

 Do you even get 100khash/sec on your Ryzen for scrypt?

 For comparison - my Innosilicon A2 units (which are GENERATION OLD ASIC and pretty badly outperformed by the more recent Bitmain L3+ and Innosilicon A4) pull 110 MEGAhash (on the miner, about 105 usually at the pool) on 1250 watts draw at the wall.

 I can't speak to the other projects, but I have a LOT of experience with Moo Wrapper and how to optimise my setups for D.Net work in general (after 19+ years I should know SOMETHING about that latter!).

 As of right now, after the big Gridcoin runup of the last couple months, my "best" rig (pair of R9 280x and an AMD A10-5700 running the GPU only) is good for ballpark 800k RAC (once it maxes out, I swapped it to LINUX from Windows about a week back due to Windows borking something during a power outage) - generating rough ballpark 50 GRC/day on an energy cost of about 70 cents/day.
 Swapping the R9s out for a pair of RX 470s would bump the RAC up to probably around 950k RAC generating more like 60-65 GRC a day at an energy cost closer to 40 cents/day.

 In other words, it's NOT particularly profitable even with current GPUs on that project (the same rig with JUST the pair of 280x generates 600 h/s on ZEC for about $7/day right now and uses a hair less power) - but it DOES pay the electric bill pretty comfortably, and was covering electric cost more-or-less even before the recent price runup despite the older power-hungry cards.

 Other projects might stack up better, they might stack up worse, the small amount of testing I did last year though indicated that the projects that looked like they WOULD pay out better had "not enough work" issues most of the time.

zpool.ca has a skein option, shows 4 coins (DGB-skein, Myriad-skein, I forget the other 2 offhand but one of them apparently is merge-mined).
theblockspool has a DGB-skein pool.
suprnova has DGB pools but I'm not sure if skein is one of the options (I know they have a DGB-scrypt pool that works fairly well and a DGB-groetsl pool that didn't work very well).

 Cheapest dual-core Celeron with low power usage you can find is plenty for a mining rig, with VERY RARE exceptions.
 I'd even say "single-core" but I don't remember the last time I saw a new single-core CPU for sale, don't think anyone makes them any more.
 I've got a 3 card R9 290 rig mining ETH on a machine with a Sempron 145 and the only time I notice that the CPU is slow is when it's doing DAG files (LINUX, but I could never get genoil to compile on that machine so it's still running qtminer and rock solid about it - later testing of genoil vs qtminer on a different machine didn't show a significant speed difference anyway, just faster DAG file generation but HAVE to redo the DAG file on every boot).

 More than around 300 h/s on a R9 280x seems to require a BIOS mod, as 300 is the best I've managed out of my pair with 1100/1250 (my cards won't do 1500 memory clock at all, and anything other than 1250 gave lower hashrate).
 Gigabyte Windforce cards (3 fan versions) running 12.4

 I mine ETH on ethminer.org, ZEC on flypool - pool-side monitoring works pretty well there, and my rigs are stable enough for the most part that once or twice a day checking is plenty.

 The heat wave this weekend/now and the other one about a week and a half back though have been murder on my R9 290 hashrates - AMD drivers for LINUX won't let me underclock them so they end up in thermal-limiting due to the limited ventilation on the place I'm in right now in the worst heat of the day (it's supposed to cool off again Wendsday though, and I'm supposed to be able to start moving into my new place not later than Thursday, so hopefully the NEXT heat wave will see those rigs in a place I can keep cooler).

 I'm not sure that Zotac is short enough to work (it's only about an inch to 1.5 inches shorter than most 1070/1080/1080ti standard-length boards), but the Gigabyte ITX does work as it's almost 2 inches shorter than the Zotac.

 $350 IS a sale price on the Zotac, but it does seem to spend a lot of time there - the Gigabyte https://www.newegg.com/Product/Product.aspx?Item=N82E16814125893 is normally at 380.


 https://www.newegg.com/Product/Product.aspx?Item=N82E16813128976  is what's going into my next rig (should have the parts mid-week) - I think it's a slightly lower-cost varient on the board you linked for a 3-card build.


 I went with a pair of 1080 instead of 1080 ti, and a Seasonic X-850 (overkill a bit for the cards I'm using, it would be enough for that pair of 1080 ti and a 1070, but I'm also running a higher-end CPU for my BOINC work which balances it out some).

 I *ALMOST* went for the EVGA G2-850, as it was the same price, but I have a longer history with the X-series - the EVGA would probably work as well though, for those that prefer EVGA or if it was on sale for less than the Seasonic.


 I believe the Skein miner mentioned was the "Alex78" ccminer version, which does have a compiled Windows binary out there somewhere if you do some digging to find it.
 I saw over 600 Mh with a non-optimised 1080 in some brief testing I did last week with it - noticeably better than the trpu (?) ccminer 2.0 version.

I suspect the AMD cards aren't very good at Skein, but it takes some digging to find an AMD-based Skein miner at all and I've not tried it myself yet.