Again, whattomine doesn't take in to account other variables that can and do affect your profitability. Pool round variance, rejected shares, pool/exchange fees, Tx fees, etc. all affect your bottom line by as much as 20-30%. By default whattomine also uses average 24H difficulty and profitability in it's calculations instead of real time. The best way to figure out what your best profitability for your rig is, is to try different pools if one isn't working for you.

If by normal, you mean before the artificial ICE AGE effects that began in May to prepare for the switch to POS in the Casper release that has been postponed, then yes. The difference from before May is after the September fork, the block reward will be ~3 ETH instead of 5. There is also a second fork planned for November this year, but that should not have many consequences for miners. The switch to POS is also more than a year away.

Profitability calculators like whattomine.com can only show you what the current profitability is based on if you were to sell coins you already mined at that very moment. They don't and can't take in to account other variables like price fluctuations, difficulty fluctuations, pool round variance and different pool/exchange fees. Unless you have a farm where you can mine enough coins to cash out in time to exchange them for the coin you want quickly, the profitability you see will not be accurate.

The block times also went up. That should be the last ICE AGE difficulty adjustment before they push the next fork on September 17th, when the block times will go back to 15 seconds and the difficulty will go back to being predicated on the network hash rate. At the same time the block reward will also be reduced to ~3 ETH.

As I mentioned, the PCI-E slots are labeled in pairs, 1,1_1; 2,2_1; 3,3_1, etc. I use the same card type (AMD or Nvidia) on each of the PCI-E slot pairs.

The black PCI-E connector slots are the primary PCI-E ports (PCI-E 1,2,3,4,5 and 6) and the white PCI-E slots are the secondary PCI-E slots (1_1,2_1,3_1,4_1,5_1,6_1 and 7_1).

Afterburner can only manage up to 8 cards. You can use EVGA Precision to manage the Nvidia card.

You will still need to use 8-pin to 6+2 splitters and sata/4-pin molex to 6-pin PCI-E adapters regardless. The EVGA 850 G2 or P2 come with same connectors. Each EVGA 850 G2/P2 PSU has four 8-pin PCI-E cables, two of which also have a 6-pin pigtail. So two 850 G2/P2 PSU's would give you a total of eight 8-pin PCI-E connectors + four 6-pin PCI-E pigtail connectors. You need a total of six 8-pin and six 6-pin PCI-E connectors for the GPU power inputs. Using the four PSU pigtail cables is enough for four of the cards, for the other two, you will need to use a 8-pin to 6+2 splitter cable for each of the cards.

For the six 6-pin powered risers, you have two PSU PCI-E 8-pin cables left that can power two risers each using a 8-pin to 6+2 splitter cable. For the other two risers, you would need to use a dual sata/4-pin molex to 6-pin PCI-E adapter cable, connecting only one riser per PSU strand. I like to use the sata connectors to power the risers since there are more sata ports on the PSU than 4-pin peripheral ports and I use the molex connectors to power the motherboard PCI-E supplemental power.

You would need to forward the miner api port in your router to same port on your rig local IP. Then use your rigs external IP instead of the LAN IP to connect to api from outside your LAN. You also may need to allow the port in the firewall on your rig to external addresses.

Windows has better GUI overclocking tools and miner support, but smOS, which is based on Ubuntu, is getting better Nvidia support. At only $2 per rig a month, it would be a good option to try. You only need a 16GB flash drive to run it and the only change you make is to add your email addreess to the config file in the root of the flash drive. Then you just boot off the flash drive and it automatically detects all your cards and starts mining. Every other configuration like miners, pools and overclocking is made through the web UI. Very easy and straightforward to use. BBT did a video tonight on a 6 card 1060 build running on smOS that can give you an idea on getting it going. There are also lots of video walkthroughs on setting it up from start to finish on Youtube from BBT and others.

https://www.youtube.com/watch?v=vbaL6UFvK7g

If it's only going to be used as a mining rig, I would skip the case and build my own BBT open air mining frame below for about $40 out of two 8ft pieces of 1/8" x 3/4" angled aluminum, one 8 ft 1" x 3" piece of pine, one 4ft 1" x 2" piece of pine and #8 1/2" self-tapping screws or out of all wood if you want to save some money. The cards will run cooler and it will easier to add more cards later if you want to.

For a motherboard I would also go with a dedicated motherboard like the Biostar TB85 for the same reason. You can put one card in the 16X PCI-E slot and use a USB riser for the other card.

https://www.newegg.com/Product/Product.aspx?Item=N82E16813138436

6 GPU Frame (Dimension 21" W x 15 3/4"" D x 10 3/4" H):

Build Cuts:

For the Angled Aluminum:

-First 8ft (96") Angled Aluminum; .75 inch of waste

4x 15 3/4" cuts (63" use) (Case Depth)
3x 10 3/4" cuts (32.25" use) (Case Height)

-Second 8ft (96") Angled Aluminum; 1.25 inch of waste

4x 21" cuts (84" use) (Case Width)
1x 10 3/4" (10 3/4" use) (Case Height)

And for the wood:

-8ft 1x3 Pine (96"); 2 inch of waste:

6 cuts of 15 3/4" (94" use) (Case Bottom)

-4 ft 1x2 Pine (24" left over):

1 cut of 21" (21" use) (Card Bracket Support)



https://www.youtube.com/watch?v=D5xNrlxsCVs

https://www.youtube.com/watch?v=MA_sH4q2toU&feature=youtu.be&t=283

Did you try installing the OpenCL runtime libraries from the AMD APP-SDK?

http://developer.amd.com/amd-accelerated-parallel-processing-app-sdk

I dual mine UBQ+DCR with my 10 RX 4XX/5XX cards. Not because Ubiq is the most profitable coin to mine, but because I believe it will be a profitable coin to hold. I have a single 1080 that is mining ZEC.

The 1600 G2 or P2 has nine PCI-E connectors. You would still need to use 8-pin to 6+2 splitter cables to connect all the GPU power inputs and 6-pin risers.

If you want to use dual 750W or 850W PSU's, they only have four PCI-E connectors each. You are going to need 8-pin to 6+2 pin PCI-E splitter cables and dual sata to PCI-E 6-pin converters.

https://www.aliexpress.com/item/20CM-Black-Sleeved-8-Pin-PCI-E-GPU-to-Dual-8-6-2-Pin-Splitter-PC/32688175733.html

https://www.aliexpress.com/item/Deconn-SATA-15-Pin-Male-to-PCI-Express-6-Pin-Dual-Video-Card-Adapter-Cable/32584991290.html

First PSU powers the motherboard, CPU, accessories + use two of the four PSU PCI-E connectors to power four 6-pin risers using a 8-pin PCI-E to dual 6+2 splitter cable, connecting two risers per PSU PCI-E cable and then use two dual sata to 6-pin PCI-E converter cables to power the other two 6-pin risers, one riser to each PSU sata cable. Use the remaining two PCI-E connectors on the first PSU to power two of the GPU's 8-pin + 6-pin power inputs using the cable 6-pin pigtails or a 8-pin to 6+2 splitter cable.

Second PSU only powers the remaining four GPU 8-pin + 6-pin powers inputs using the cable 6-pin pigtails or a 8-pin to 6+2 splitter cable. Connect only one GPU per PSU PCI-E cable.

I would go with dual EVGA 850W G2 PSU's since two of the PCI-E cables have a 8-pin + 6-pin pigtail that you need. If you go up to a EVGA 1000W G2, it has 6 PCI-E connectors that can power the six GPU  8-pin + 6-pin power inputs and then use a 750W G2 to power the motherboard, all the risers and everything else.

You got it backwards. The ICE AGE effects have been in place since May of this year as preparation for the switch to the POS Casper release, initially planned for later this year. Recently it was announced the switch to POS is not ready to made for what looks like at least 12 months. There was some contention as to whether the ETH developers would delay the programmed difficulty rise and block time increases from the ICE AGE, now that the switch to POS has officially been delayed.

What was confirmed at the last two developer meetings, is the diffusement of the ICE AGE difficulty bomb and block time increases will not happen until they release the next planned fork, in late September or early October this year. At which point the difficulty will go back to being based on the network hash rate, the block times will go back to 15 seconds along with a reduction in the block reward to ~3 ETH at the same time the fork is implemented.

In the meantime the continued planned difficulty and block time increases from the ICE AGE implementation will go on as scheduled. The next being August 25 which is expected to increase the difficulty ~30% from where it is now and then another difficulty ramp up on September 24th. The block times will also go from the current 20 seconds to 25 seconds on August 25th and then to 32 seconds on September 24th.

All of which means starting tomorrow, mining ETH is not going to be very profitable until the next hard fork unless the price goes up substantially from where it is now.

https://etherchain.org/charts/difficultyBomb

Any RX 570 4GB card is better and more profitable than ANY 1060. I also suggest you don't listen to people that tell you mining isn't profitable. It is, in fact more profitable than it's been historically including the beginning of this year. Do your own research and determine what's best for your own situation, just as you should when considering ANY investment, rather than listining to random people with their own agenda or bias.

If you can get two RX 570 4 GB cards for $250 or less, you will get ~315-320 H/s on Equihash from each. Not necessarily cheaper, but ~25%+ more hash for the same cost, so a quicker ROI than a single 1070.

The brand doesn't matter. Theoretically you can run up to 8 AMD + 8 Nvidia GPU's under Windows 10 currently. Supposedly later this year, a driver update will remove that 8 GPU limit. Under Linux you can run 13+ cards of the same type. There are lots of videos on Youtube showing 13 AMD or Nvidia cards running under smOS or ethOS.

One issue I had with this ASRock H110 Pro BTC+ board in Windows 10 when mixing 8 AMD + 1 Nvidia card is that there are 6 PCI-E groups that are apparently shared, meaning:

PCI-E 1,1_1
PCI-E 2,2_1
PCI-E 3,3_1
PCI-E 4,4_1
PCI-E 5,5_1
PCI-E 6,6_1

For a total of 12 shared PCI-E slots and the 13th is PCI-E 7. When running 8 AMD RX 4XX/5XX cards + 1 Nvidia GTX 1080 plugged in to one of the shared PCI-E slots, I had random reboots after 10-20 minutes from driver conflicts. When I moved the Nvidia card to the PCI-E 7 slot which isn't shared, it fixed the problem. If you have a mixed AMD and Nvidia rig, I would suggest to group cards of the same type according to the PCI-E slot pairs to avoid conflicts.

Another issue is you can only manage up to 8 cards in Afterburner, so for overclocking and undervolting I use Afterburner for the 8 AMD cards and EVGA Precision for the Nvidia card.

According to NiceHash, using a GEN 3 PCI-E lane can make a 1-2% difference in hash rate from the higher bandwidth, especially for some data transfer intensive algorithms.


https://www.nicehash.com/?p=news&id=155

For most algorithms I doubt it's actually a 1 to 2% difference, but it's probably not zero either.

That must be your Hynix VRAM. All my RX 570/580 Hynix memory cards, 7 in total from Sapphire and Powercolor have a simple 1500 MHz strap mod and all get at least 28.5 MH/s.