[OS] nvOC easy-to-use Linux Nvidia Mining

[fullzero]

So I was trying to run 7 Zotac 1070 Minis with Asus Z270F using Z270A image, the most it could detect is 6 Zotacs. Someone here said that Z270F runs 7 GPUs out of the box, but not in my case. I need to try windows to verify that. Do you have Z270F and planning to post an image? Running 6 Zotacs with TB85 for now, but Minis are so tiny that 8 could fit in Rosewill easily.

Thanks.

I just got one of these mobos in yesterday.  I can get 7x cards to be detected by the system, but there is a conflict with x and the driver.  I can manually OC the first 6 cards, but the 7th has no response to OC commands. 

There is also what appears to be a TPM / Secureboot problem preventing subsequent boots into the same install.  Initially changing the bios setting for enabling above 4gb encoding is required to get 7 cards to work.  There are also some TPM / Secureboot bios changes that are needed, but I still need to isolate which.  I started with the MSI z270-A PRO image.

I will work on this more tonight.

[1715081783]

1715081783

[1715081783]

1715081783

[1715081783]

1715081783

[fullzero]

Hi fullzero,

Excavator will support Linux very soon according to zawawa.

I often do rentals at Nicehash and for this case I would want to use the NH's optimized Excavator instead of EWBF or Claymore.

Hope you will add Excavator Linux when its available.

Thanks

I will add Excavator to the next version after it is released.

[fullzero]

Hey fullzero

On ETH only it is not that great. According to web I should get close to 22 - 24 but on my 150cpu oc and 900mem oc my ETH only runs at between 18.8 and 18.1 per card. I also have the same Zotacs as you being the 1060 AMP+ 6GB edition.

I think it is because of the powerlimit/level being stuck on 2 and cant go to 0 so that it can get more 'juice'. I have read other threads regarding the powelevel/limit being stuck on 2 but I don't know enough about linux to go about upgrading nvidia drivers etc to test. I have played with my cpu and mem oc but my limit stays on P2.

Kind Regards
Roes5000

I have the non + version, this is the first Pascal card I have seen with this problem.  Does 22 mh/s with ETH no problem with OC.

What happens if you open the guake terminal ( press f12 ) open a new tab if cpu mining, then enter the command:

sudo nvidia-settings -a [gpu:0]/GPUPowerMizerMode=1

enter password ( miner1 ) when prompted

Also have you tried not using a powerlimit?

[Roes5000]

Hey fullzero

On ETH only it is not that great. According to web I should get close to 22 - 24 but on my 150cpu oc and 900mem oc my ETH only runs at between 18.8 and 18.1 per card. I also have the same Zotacs as you being the 1060 AMP+ 6GB edition.

I think it is because of the powerlimit/level being stuck on 2 and cant go to 0 so that it can get more 'juice'. I have read other threads regarding the powelevel/limit being stuck on 2 but I don't know enough about linux to go about upgrading nvidia drivers etc to test. I have played with my cpu and mem oc but my limit stays on P2.

Kind Regards
Roes5000

I have the non + version, this is the first Pascal card I have seen with this problem.  Does 22 mh/s with ETH no problem with OC.

What happens if you open the guake terminal ( press f12 ) open a new tab if cpu mining, then enter the command:

sudo nvidia-settings -a [gpu:0]/GPUPowerMizerMode=1

enter password ( miner1 ) when prompted

Also have you tried not using a powerlimit?

When I enter that command into the quake terminal it says atttribute set to 1 , but when I go nvidia-smi , it stays on P2. I have now disabled powerlimit setting and I have increased it to 150 , it stays on P2 my side. My ETH does not go over 18.8. Can you confirm yours is on P0 ? I suspect my side might be driver related or the image I flashed isn't 100% compatible. Is there a easy'ish way to upgrade Nvidia drivers ?

Kind Regards

[fullzero]

Hey fullzero

On ETH only it is not that great. According to web I should get close to 22 - 24 but on my 150cpu oc and 900mem oc my ETH only runs at between 18.8 and 18.1 per card. I also have the same Zotacs as you being the 1060 AMP+ 6GB edition.

I think it is because of the powerlimit/level being stuck on 2 and cant go to 0 so that it can get more 'juice'. I have read other threads regarding the powelevel/limit being stuck on 2 but I don't know enough about linux to go about upgrading nvidia drivers etc to test. I have played with my cpu and mem oc but my limit stays on P2.

Kind Regards
Roes5000

I have the non + version, this is the first Pascal card I have seen with this problem.  Does 22 mh/s with ETH no problem with OC.

What happens if you open the guake terminal ( press f12 ) open a new tab if cpu mining, then enter the command:

sudo nvidia-settings -a [gpu:0]/GPUPowerMizerMode=1

enter password ( miner1 ) when prompted

Also have you tried not using a powerlimit?

When I enter that command into the quake terminal it says atttribute set to 1 , but when I go nvidia-smi , it stays on P2. I have now disabled powerlimit setting and I have increased it to 150 , it stays on P2 my side. My ETH does not go over 18.8. Can you confirm yours is on P0 ? I suspect my side might be driver related or the image I flashed isn't 100% compatible. Is there a easy'ish way to upgrade Nvidia drivers ?

Kind Regards

This is the newest driver: http://www.geforce.com/drivers/results/118524

Code:

It's not recognized yet (and may not be as it is not a long term support version) by the Ubuntu ppa with which you would:

sudo add-apt-repository ppa:graphics-drivers/ppa ( I have already added )

sudo apt update

then go to the top right and type ad, select additional drivers, and change to the newer driver.

So you have to do it manually:

download the driver, then extract the folder.  There should an install script inside.  Right click the install script and go to properties, ensure the allow execution as a program box is checked.  Now drag the script to the guake terminal, move the cursor to the front of the command and add: sudo before the 'absolute_path_command'.  Make sure there is a space between the sudo and the '.

Before doing this I would recommend making a image of your modified usb with hddrawcopy on your maincomputer.

[fullzero]

ASRock H81 PRO BTC R2.0 LGA 1150 is in Stock at Newegg 1 per customer.   

This motherboard is fully supported by nvOC and will allow for 6x GPUs.

https://www.newegg.com/Product/Product.aspx?Item=N82E16813157724

Get one fast before they are gone.

[fatsailor]

The zip image for the H81 appears corrupt. I can't unzip on a mac. It says the header is corrupted when I try to unzip - I downloaded twice.

Any chance you can post SHA256 hashes for the images?

[fullzero]

The zip image for the H81 appears corrupt. I can't unzip on a mac. It says the header is corrupted when I try to unzip - I downloaded twice.

Any chance you can post SHA256 hashes for the images?

for H81-PRO-BTC_or_Biostar-TB85__nvOC_v0014.zip the SHA256 hash is:

Code:

46e53e99edb2bfa654ebd1f2d1fed53df97e41756e88898da3536590981050d8

I don't have a mac currently, so I can't test on one.  I will get one eventually.

The default OSX disk tools may not like the zip.  I am not familiar with third party OSX disk apps, generally I like rufus, but I don't believe there is a mac version.

If your image produces the same hash, then it is most likely a problem with the OSX disk tools.  If you install bootcamp with windows or linux you should be able to decompress the zip.  I also imagine there is a third party mac app that can unzip the image. 

[xleejohnx]

The zip image for the H81 appears corrupt. I can't unzip on a mac. It says the header is corrupted when I try to unzip - I downloaded twice.

Any chance you can post SHA256 hashes for the images?

for H81-PRO-BTC_or_Biostar-TB85__nvOC_v0014.zip the SHA256 hash is:

Code:

46e53e99edb2bfa654ebd1f2d1fed53df97e41756e88898da3536590981050d8

I don't have a mac currently, so I can't test on one.  I will get one eventually.

The default OSX disk tools may not like the zip.  I am not familiar with third party OSX disk apps, generally I like rufus, but I don't believe there is a mac version.

If your image produces the same hash, then it is most likely a problem with the OSX disk tools.  If you install bootcamp with windows or linux you should be able to decompress the zip.  I also imagine there is a third party mac app that can unzip the image. 

I have no issues on mac with winzip for mac

[machiavellious]

to launch oneBash with screen you would edit 2unix to contain the following:

Code:

#!/bin/bash

expect -c 'spawn sudo dos2unix /media/m1/1263-A96E/oneBash
expect "*password*:"
send "miner1\r" '
sleep 6

expect -c 'spawn sudo ldconfig /usr/local/cuda/lib64
expect "*password*:"
send "miner1\r" '
sleep 6

screen -m -t name_of_your_window bash '/media/m1/1263-A96E/oneBash'

Let me know how this goes.

Thank you so much.  This works for starting screen, but the onebash script keeps creating new instances...  
I thought I would be able to fix this, but I couldn't figure out how you're making onebash reinitialize.
How are you starting onebash on terminal startup? I checked ~/.bashrc and /etc/bash.bashrc, I might have missed it, but didn't seem to be in there.  

Thanks again, you're enabling me to learn a lot.

[fullzero]

The zip image for the H81 appears corrupt. I can't unzip on a mac. It says the header is corrupted when I try to unzip - I downloaded twice.

Any chance you can post SHA256 hashes for the images?

for H81-PRO-BTC_or_Biostar-TB85__nvOC_v0014.zip the SHA256 hash is:

Code:

46e53e99edb2bfa654ebd1f2d1fed53df97e41756e88898da3536590981050d8

I don't have a mac currently, so I can't test on one.  I will get one eventually.

The default OSX disk tools may not like the zip.  I am not familiar with third party OSX disk apps, generally I like rufus, but I don't believe there is a mac version.

If your image produces the same hash, then it is most likely a problem with the OSX disk tools.  If you install bootcamp with windows or linux you should be able to decompress the zip.  I also imagine there is a third party mac app that can unzip the image. 

I have no issues on mac with winzip for mac

thanks that should help fatsailor

[Longsnowsm]

Finally got around to getting some USB drives and through some other tasks that were more pressing.  So I think I am ready to try and setup an nvOC miner.  I am using an Asrock H97 Anniversary for this rig.  Which image should I download for the best support of that mobo?  Thanks.

[fullzero]

to launch oneBash with screen you would edit 2unix to contain the following:

Code:

#!/bin/bash

expect -c 'spawn sudo dos2unix /media/m1/1263-A96E/oneBash
expect "*password*:"
send "miner1\r" '
sleep 6

expect -c 'spawn sudo ldconfig /usr/local/cuda/lib64
expect "*password*:"
send "miner1\r" '
sleep 6

screen -m -t name_of_your_window bash '/media/m1/1263-A96E/oneBash'

Let me know how this goes.

Thank you so much.  This works for starting screen, but the onebash script keeps creating new instances...  
I thought I would be able to fix this, but I couldn't figure out how you're making onebash reinitialize.
How are you starting onebash on terminal startup? I checked ~/.bashrc and /etc/bash.bashrc, I might have missed it, but didn't seem to be in there.  

Thanks again, you're enabling me to learn a lot.

You would need to change the gnome terminal profile setting to alter this.  I have it set to relaunch on command exit to ensure that dos2unix will eventually properly convert oneBash, as it will not always do so successfully on the first attempt.

So long as you don't reconnect your usb to a windows computer this should already be done on your rig, so altering it should be ok.

when mining process is open select the gnome terminal (where gpus are mining) then move the cursor to the top right, select preferences, ensure mining (profile) is highlighted, click edit, select the command tab, change "when the command exits" to exit the terminal

This should be what you are looking to do.

[fullzero]

Finally got around to getting some USB drives and through some other tasks that were more pressing.  So I think I am ready to try and setup an nvOC miner.  I am using an Asrock H97 Anniversary for this rig.  Which image should I download for the best support of that mobo?  Thanks.

Use the  ASRock H81 PRO BTC or BIOSTAR TB85 image; it will recognize all six cards and OC the first five without modification.  To get the sixth to OC you can either follow this process:

on first boot close mining process if open

Log out

Log in (yes you need to do this to make sure this process works)

Attach a second monitor or dummy plug to the GPU in the sixth slot.

open Ubuntu start button equivalent on top right. 

Type: n 

now you should see Nvidia X Server Settings

open Nvidia X Server Settings

select X server Display Configuration

look in the display area click each box that has a black boarder  (should only be the last screen) and change "configuration" to: New x screen)

click Save to X Configuration File

save xorg  ( ensure Merge with existing file is checked and the field contains: /etc/X11/xorg.conf )

Note: you may need to browse an existing xorg file; in order to enable selecting Merge with existing file

password is: miner1

quit Nvidia X Server Settings (you must actually quit)

close mining process if open

logout

login

now all 6x cards should OC

or I can show you how to edit the xorg.conf file directly with the terminal commands, gksu, and gedit.

[jcovercash]

First off thanks for doing this. I am setting up my first 4 card system with 1060s and a GA-B250M board so this is perfect for me.

I downloaded the image and verified the checksum. I am using the HDD Raw Copy Tool to install it on an Adata 16GB 3.0 USB that I had but it gives me an error at around 97%, Operation Terminated at offset 15,535,636,480 LBA 30,343,040

The image shows 16GB in HDD Raw and the flash drive shows 15.53 could this be the problem even though the image file is below this size when I look at it in windows.

I did get it to work on a 32Gb 2.0 drive just to test but it was unbearable slow but did work and i was able to mine with one card.



When it started up I got an error about the fan control.

ERROR: Error assigning value 75 to attribute 'GPUTargetFanSpeed'
(m1-desktop:0[fan:0]) as specified in the assignment
'[fan:0]/GPUTargetFanSpeed=75' (Unknown Error).

This is on a 1060 EVGA SC 3GB.

I was only running one card directly in the main PCI slot during this error.


Thanks for your time and assistance.

[fullzero]

First off thanks for doing this. I am setting up my first 4 card system with 1060s and a GA-B250M board so this is perfect for me.

I downloaded the image and verified the checksum. I am using the HDD Raw Copy Tool to install it on an Adata 16GB 3.0 USB that I had but it gives me an error at around 97%, Operation Terminated at offset 15,535,636,480 LBA 30,343,040

The image shows 16GB in HDD Raw and the flash drive shows 15.53 could this be the problem even though the image file is below this size when I look at it in windows.

I did get it to work on a 32Gb 2.0 drive just to test but it was unbearable slow but did work and i was able to mine with one card.

Not all USB keys listed as 16gb are actually 16gb; some of them are only 15.5gb or somewhere inbetween.  I made all the nvOC images with this in mind and ended the primary partition before the 15.5gb point.  So, as in your case, when using a USB key that is not actually 16gb it will give you an error: but the key should still be properly imaged. 

When it started up I got an error about the fan control.

ERROR: Error assigning value 75 to attribute 'GPUTargetFanSpeed'
(m1-desktop:0[fan:0]) as specified in the assignment
'[fan:0]/GPUTargetFanSpeed=75' (Unknown Error).

This is on a 1060 EVGA SC 3GB.

I was only running one card directly in the main PCI slot during this error.

Thanks for your time and assistance.

that error means oneBash tried to implement fan speed on a card that either doesn't exist in the system or is identified differently than the logic implemented in oneBash is programmed to handle.

If you scroll up to the beginning of oneBash do you see gpu0 core overclock and gpu0 memory overclock messages?

[Roes5000]

Hi Guys

I also got that error before I edited my xorg.file.

Please do a lspci | grep VGA and post your output. I had 4 GPU's and my hardware ID's was 01:00 , 02:00 then 04:00 and 05:00:

m1@m1-desktop:~$ lspci | grep VGA
01:00.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)
02:00.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)
04:00.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)
05:00.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)


When you then go into your /etc/X11/xorg.conf and scroll down a bit you will see you VGA entries:

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce GTX 1060 6GB"
    BusID          "PCI:1:0:0"
EndSection

I just made that the first 4 entries correlate to my lspci | grep VGA output . Save and reboot. After that all the settings applied.

Hope that helps

[fullzero]

Hi Guys

I also got that error before I edited my xorg.file.

Please do a lspci | grep VGA and post your output. I had 4 GPU's and my hardware ID's was 01:00 , 02:00 then 04:00 and 05:00:

m1@m1-desktop:~$ lspci | grep VGA
01:00.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)
02:00.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)
04:00.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)
05:00.0 VGA compatible controller: NVIDIA Corporation Device 1c03 (rev a1)


When you then go into your /etc/X11/xorg.conf and scroll down a bit you will see you VGA entries:

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce GTX 1060 6GB"
    BusID          "PCI:1:0:0"
EndSection

I just made that the first 4 entries correlate to my lspci | grep VGA output . Save and reboot. After that all the settings applied.

Hope that helps

This is correct; but for members who are not familiar with linux: you must do this as root.

so press f12 to open the guake terminal

then enter:

lspci | grep VGA

this will list your gpus and their pcie addresses

then enter:

gksu gedit '/etc/X11/xorg.conf'

and enter the password:

miner1

when prompted

gedit should now be open with root access to the xorg.conf file

Find the Device section and alter the

BusID          "PCI:1:0:0"

on each Device to match the addressing from lspci

save

close all mining processes if open

logout

login, and all cards should now OC if you did this correctly.

I will automate this process in a later version, to make everything easier.

[jcovercash]

Thanks im not sure if it helped but I disabled the onboard VGA and setup that 15.5gb 3.0 drive. It worked for the fans and no error for OC settings.

Now a new problems GPU 0/1 are using 91/97 watts and GPU 2/3 are using 111/115 respectively and hash rates on 0/1 are down too because of this. 257/240 vs 296/300

I had to leave for work so I couldn't do more testing

Only thing I changed in one bash was NO to could and my address/pool for zec

I am using a cosair 750w cs750 power supply so it should be more than enough for this system?

Something I'm missing in settings somewhere?

Thanks

[fullzero]

Thanks im not sure if it helped but I disabled the onboard VGA and setup that 15.5gb 3.0 drive. It worked for the fans and no error for OC settings.

Now a new problems GPU 0/1 are using 91/97 watts and GPU 2/3 are using 111/115 respectively and hash rates on 0/1 are down too because of this. 257/240 vs 296/300

I had to leave for work so I couldn't do more testing

Only thing I changed in one bash was NO to could and my address/pool for zec

I am using a cosair 750w cs750 power supply so it should be more than enough for this system?

Something I'm missing in settings somewhere?

Thanks

Do you have 3gb or 6gb 1060s?

If 1060 3gb should be able to handle:  cc 100  mc 350 OC settings, you are going to want to set the oneBash part:

Code:

__CORE_OVERCLOCK=100
MEMORY_OVERCLOCK=350

to implement those clocks. 



If 1060 6gb should be able to handle:  cc 100  mc 600 OC settings, you are going to want to set the oneBash part:

Code:

__CORE_OVERCLOCK=100
MEMORY_OVERCLOCK=600

to implement those clocks. 


Depending on what model you have, you may be able to OC more.

I'm not sure what you said no to:

Only thing I changed in one bash was NO to could and my address/pool for zec

If you set:

Code:

POWERLIMIT="YES"

and then specify the wattage for the powerlimit here (note can be above or below TDP):

#change powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -pl 125

for example if I was setting a 115 watt powerlimit, that part of one bash would look like:

Code:

if [ $POWERLIMIT == "YES" ]
then
sleep 6
#change powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -pl 115
expect "*password*:"
send "miner1\r" '
sleep 6
fi

Also when oneBash starts do you see cc and mc for each card like this? (with different values of course):