Ok,
So first off, my philosophy is a bit different from others here, maybe most.
SOE is AMD CPUs, Nvidia GPUs and Linux OS - if something doesn't fit into that box I generally don't do it.
Also, I'm lazy (and old) so I usually just get my code from the repos or as debs.
I run 12 cards in 6 rigs (standard cases actually, but I've been the whole crate route in the past.) Currently I'm hitting about 205Mhs and I don't pay for electricity.
5 rigs are in my (small) living room so there's a 10K BTU window air conditioner in there as well.
So I was looking forward to the 1060s and 3 weeks ago I got a pair of MSI 1060 6GB cards and I'm really happy with them.
Reporting 36 Mhs for the pair and averaging 35Mhs as actual.
Stock settings, running ethminer, fans set to 45% and 40%, temps are 55c for each card.
And yesterday I took delivery of 2 Gigabyte 1060 6GB cards to finish my final rig.
Hashrate should be about the same - eventually. Less happy this time.
So I posted a couple of reviews on Newegg.ca which is where all the cards came from and I thought I'd reproduce them here.
*****************
1.
5 out of 5 eggs
Great mining card
08/23/2016
This review is from: MSI GeForce GTX 1060 DirectX 12 GeForce GTX 1060 GAMING 6G 6GB 192-Bit GDDR5 PCI Express 3.0 x16 ATX Video Card
Pros:
Fast , cool
Bang for the buck
Rock solid 18Mhs at stock with Linux and ethminer from repos.
Cons:
Bang for the buck?
Bought 2 of these 21 days ago for $370 CDN each.
Now they are $481 CDN. Up $111 in three weeks??!
Because the gamers love them so much? A card you cannot SLI?
Is it Newegg or MSI? Way to gouge buddy.
2.
1 out of 5 eggs
Worst mining card ever!
08/23/2016
This review is from: GIGABYTE GeForce GTX 1060 Windforce OC GV-N1060WF2OC-6GD Video Card
Pros:
Short
Cons:
Hot.
So the two I ordered came yesterday - stuck them in a rig, fired it up, fans at 40% otherwise stock settings running ethminer.
5 minutes later the top card is showing 82c at about 105w.
Cranked the fans up to 75% - over 2000rpm - temp drops to about 75c
Turn off fan control and undervolted to 80 watts - temps are now in the 70s, no loss of hashrate, fans are at about 50%.
Undervolt to 70 watts - yes! 70WATTS- now the temps are at 68C - which is about the same as the power draw, hashrate drops 3Mhs.
So I have a card which actually produces 1 degree c of heat for every watt of power it uses. WOW!
Of the 12 Nvidia cards that I run the next hottest is a 980 classy at 59c, and that sits right on top of a 980ti.
Other Thoughts:
Hot.
Really hot.
Really, really hot.
Like a pizza oven.
Buy one of these for Xmas - give it to someone you hate.
My only air-cooled 980ti runs at 53c and it's pulling 150w+.
This is NOT a qc issue, this is by design - shoddy design and a fu attitude.
I have 2 Gigabyte 970 triple fan cards - they have been running 24/7 for over 9 months. Zero problems.
Now this.
No refunds, no exchanges, RMA for two more of the same. Cost of shipping, loss of use.
Just have to wait until the fans burn out - which they will. Then RMA and sell the new cards to a gamer.
And add Gigabyte to my shitlist.
*********
So, those are the reviews
Here's some proof -

OK, so I had to sit and wait a bit for the wattage to fluctuate to get that exact match.
Comments
So, I tried compiling for cuda mining and I can't get it right. In the end had to install Windows before I could get the 1060 going.
My query. Can you point out what exactly you downloaded to get it working in Linux?
You need to add the repos to Synaptic.
sudo add-apt-repository ppa:ethereum/ethereum - also I add
sudo add-apt-repository ppa:ethereum/ethereum-qt
sudo add-apt-repository ppa:ethereum/ethereum-dev
Update the repos and then search for "ethereum" and "ethminer"
Those will pull in all the other bits and pieces.
You may not actually need ethereum if you are just mining but it's a habit I have now and I sometimes use geth.
Then open a terminal and -
ethminer --farm-recheck 200 -G -F http://10.42.0.1:8080/rig id
or similar.
I used to be a big fan of AMD cards particularly because of the DP floating point.
But increasing driver problems and the introduction of the Maxwell cores sent me to team blue.
Genoil's Cuda only works on Ubuntu 14.04 and 14.10 (I think) which is the main reason I don't use it.
But there are no AMD drivers available for 16.04 so 14.04 / 14.10 is probably the way to go.
But there is one other problems with Linux / AMD from my perspective.
The utility AMDOverdriveCtrl is no longer supported.
This is just a GTK front end for the AMD command line utility but it is / was really excellent.
Enabled you to over/under clock and to set temperature/fan gradients.
The image I posted is from the Nvidia-smi utility which is excellent but you have to set fixed values for the fan speeds
Hope that helps.
And definitely go with 14.04 as it is the LTS release. For adjusting the parms I use a combination of aticonfig/atitweak/atiflash.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814125903&cm_re=1060_gigabyte_6gb-_-14-125-903-_-Product
I compared it against a PNY 1060 2 fan, long card. They both ran at 66c with stock settings, so maybe you got a bad Gigabyte card? I don't know why a two fan version would run so hot. Also, the watt per hash ratio is aweseome
On another note, I was able to get claymore on win 7 , 64 to finally get this card to mine @ around 19MH/s.
Unfortunately when I put it in a Win 10 Pro machine, with the same drivers, and cuda toolkit, it will not mine correctly.. any help on 1060 mining is appreciated (on win 10)
I'm currently running ~23.5Mh/s with 65% PL +200/1000.
You can tip me eth at: 0x6Fd004d638031aaCB6409AE6dA9Ad5eA915678f7
@ethfan If you're having trouble installing genoils, here's a quick guide on installing it on Ubuntu 16.04 (or derivatives of it).
For the Nvidia 10xx series you want to download the Cuda 8.0RC and install that instead of 7.5.
wget http://developer.download.nvidia.com/compute/cuda/repos/
ubuntu1404/x86_64/cuda-repo-ubuntu1404_7.5-18_amd64.debsudo dpkg -i cuda-repo-ubuntu
1404_7.5-18_amd64.debsudo apt-get -y install software-properties-common
sudo add-apt-repository -y ppa:ethereum/ethereum
sudo apt-get update
sudo apt-get install git cmake libcryptopp-dev libleveldb-dev libjsoncpp-dev libjsonrpccpp-dev libboost-all-dev libgmp-dev libreadline-dev libcurl4-gnutls-dev ocl-icd-libopencl1 opencl-headers mesa-common-dev libmicrohttpd-dev build-essential cuda -y
git clone -b 110 https://github.com/Genoil/cpp-ethereum.git
cd cpp-ethereum/
mkdir build
cd build
cmake -DBUNDLE=cudaminer ..
make -j8
cd ethminer
./ethminer -v
For 14.04, just change
libjsonrpccpp-dev
tolibjson-rpc-cpp-dev
Maybe the MSI has better memory
msi z170a gaming m5 motherboard
6 6gb msi gtx 1060 gaming x 6g gpus
All cards on risers
Pentium g4400 3.3ghz
8gb ram
60gb 500mb/s solid state
1500w psu
operating claymore in Windows 10 pro. I have the HDMI plugged into the motherboard.
I have tried different bios for motherboard (the may release and 1.D) no difference. I have gone through different bios setting guides and the common problem seems to be getting all six cards to work. I have got them all running, they are just hashing at a fraction of what they should be. I have run ddu uninstaller and reinstalled the drivers a number of times and installed cuda 8 and the windows c++ redistributable packages for visual studio. No matter the many different things I've tried, if the cards are hashing they are doing it at this speed. I dont know what I've missed.
has anyone had a similar problem or does this sound at all familiar to anybody? anyone have ideas of what i might have done wrong or what part of my setup is causing this? any help is greatly appreciated. this was quite an investment for me and is performing quite disappointingly. thank you for any help you offer
This GPU has 9Gbps GDDR5 memory.. Am not expert but could it be the case? Can anyone else confirm?
Its like premium GTX1060 model... Costs 350+ EUR now..
I should also add that I followed cryptobadger's guide. this is the third page here, where i also commented, looking for help. http://www.cryptobadger.com/2017/04/build-ethereum-mining-rig-windows/#comments
two other people have posted there since myself with the same issue.
are you also using Windows?
Guys I'm having another issue, got 6x1060 rig with 5xMSI GeForce GTX 1060 3GT OC 3GB and 1x EVGA GeForce GTX 1060 SC GAMING 3GB. The thing is all cards are running the exact same clocks (according to MSI afterburner) and EVGA does ~23MH/s while the others do 19MH/s. These are overcloced settings, on stock they do 19MH/s and ~15MH/s respectively.
As of GPU-z these are the exact same cards with exceptions of memory - Samsung vs Hynix. When mining the usage on EVGA is ~96%GPU 100% memory controller whereas on MSI it's 100%GPU and ~95% memory controller. Again clocks are the same.
Have any of you notices such a big differences between two 1060's?
First I put everything I would need on a flash drive so I wasn't downloading anything to the rig.
I started with none of the GPUs plugged into the PCIE slots, as in I left all the risers unplugged from the motherboard. Still had them plugged into the PSU. I did a clean reinstall of Windows 10, keep nothing. Next I installed the Windows anniversary update (Windows 10 1607). I realized at this point, I maybe could have just done 1607, because I went through the same process as when I reinstalled Windows the first time.
I did NOT update Windows after at any time. I did use Badger's link to prevent Windows from auto-updating.
Next, I installed the 1.D BIOS for the motherboard.
Then I installed all of the drivers. There were two I could not install (The serial IO driver, which I found out was for tablets and touch pad devices. The other one was the Intel SGX, which I did not find a full understanding of what that is.
I did not install any of the mobo utilities.
Next I plugged the risers from the GPU into the motherboard. It seemed like it was installing those drivers on it's own. But, next I installed the GPU driver from MSI's site. After that, I installed all the GPU utilities from MSI's site. I even let GForce experience install, and I switched off the update notifications.
Next I installed the visual c++ 2017 redistributable.
I attempted to install CUDA after this but it said it wanted visual studio. So I went back to my PC and put visual studio community the flash drive and then went back to the rig.
I installed visual studio community on the rig and tried CUDA again. It said still wasn't getting what it wanted from visual studio, so I did the install for visual studio enterprise (It's all available to install from that first visual studio file I downloaded). After this CUDA still said it wasn't getting the visual studio stuff it needed, so I just proceeded anyway with the installation.
I also had the Claymore folder unzipped to that flash drive with my batch file already in it, so at this point, I just copied it to the desktop. BTW, I let all the apps put their shortcuts on the desktop too. I'm pretty sure I didn't do any custom settings during install on anything, I just made sure I disabled updating on everything.
I did a reset before I fired up the miner, and there it was! mining like it's supposed to.
Both times I've gone through the Windows setup, the option to not turn the screen off after a certain time has been there, but the option to not shut the rig down hasn't been available, until it's already shut itself down one time, so I've had to wait before I can set that to "never"
I hope this helps other people. Probably only if you have an identical or very similar setup. I got this to work for me with a lot of help from a few different forums.
I've been my whole life in red team, today received two of EVGA 1060 6gb ITX cards.
I've seen some tutorials, I'm getting ~21,5 per card, but what I want to know regarding nvidia cards - should I play with undervolting in afterburner to save some electricity? I know thats a go with AMD cards.
Thanks!
I would advise you to try different drivers. And... here is a real kick in the head... for ME, the drivers that are recommended in the Claymore startup give me what YOU are getting - about 2.4 mH/s. Try 376.33. I get ~21mH/s out of 1060's and a tad over 30 from 1070's.
Not sure what the scoop is, but after running the 376.33 drivers without any issues (and good performance), I figured that maybe the recommended drive might get me a little more rate. It literally cut my hashrate to 12.5% of what is had been.
Try 376.33.
As far I've seen here, there are complaints about it. Should I keep 'em? Also, is there anyone here which has this one working that could provide me it's standard and OC hashrate/consumption? Thanks!