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

[cvitanihc0011]

This looks nice so far .
I have a GTX 1070 Gaming X with Micron memory.  Anybody knows how much I can push these in Linux ?
In windows I can get them between 30-31 but only if Claymore is already running . If i want to start it with mem oc +600 it crashes

On nvOC is see that my Mh/s is lower it I oc mem to +600. I get around 28mh ... lower then for same settings in Win . Am I doing something wrong ?
Thanks !

[fullzero]

nvOS v0015
matherbord: "ASUS H87M-e" AND "asus h61m-k"
video: asus 1050 ti  4gb OC


what can be done?

Are you using 1x GPU?

If so see: 

https://bitcointalk.org/index.php?topic=1854250.msg19449945#msg19449945

Let me know if you are using more than one GPU; or if you are using one GPU and that fix isn't working.

[fullzero]

This looks nice so far .
I have a GTX 1070 Gaming X with Micron memory.  Anybody knows how much I can push these in Linux ?
In windows I can get them between 30-31 but only if Claymore is already running . If i want to start it with mem oc +600 it crashes

On nvOC is see that my Mh/s is lower it I oc mem to +600. I get around 28mh ... lower then for same settings in Win . Am I doing something wrong ?
Thanks !

I don't have a Gaming X 1070; but some 1070s will OC stability up to 1100.  I have noticed that the OC offsets don't have the same results in linux as in windows; you should have to OC a little higher in linux to have the same result.  However you should also be able to OC higher in linux for this same reason.

When using Ethash, setting the mc OC too high will most likely result in a Claymore soft crash.  Setting the cc OC too high will most likely crash the system.  I would keep bumping up the mc OC by 100 until you get a Claymore soft crash.  Then reduce it by 50, and see if it is ok.  If not down another 50. 

[fullzero]

I don't have that specific motherboard; but I have other z97 mobos with similar 6 GPU difficulties.  I will get one out and see if I can find the problem tonight.

To help me find the problem:

press f12 to open the guake terminal

If cpuminer-opt is running rightclick over the guake terminal and select new tab or press ctrl + c to stop it.

then enter:

lspci | grep VGA

please tell me what this outputs

00:02.0 VGA compatible controller: Intel Corporation Xeon E3-1200 v3/4th Gen Core Processor Integrated Graphics Controller (rev 06)
01:00.0 VGA compatible controller: NVIDIA Corporation Device 1c02 (rev a1)
03:00.0 VGA compatible controller: NVIDIA Corporation Device 1c02 (rev a1)
06:00.0 VGA compatible controller: NVIDIA Corporation Device 1c02 (rev a1)
07:00.0 VGA compatible controller: NVIDIA Corporation Device 1c02 (rev a1)
09:00.0 VGA compatible controller: NVIDIA Corporation Device 1c02 (rev a1)

lspci doesn't see the GPU because the riser wasn't plugged in on boot. If it was plugged on power up and boot, would get error 55 and the system wouldn't even boot neither would get any video or interaction, hard reset is the only option there. The problem is not in the riser or GPU or slot because this is happening on all 5 rigs (z97x and 1060), replacing risers, GPU or power cables doesn't help.

I'll try now to leave it past the bios and then plug in the sixth riser on GRUB choose.

What happens if you disable the onboard graphics?

So I just got my 1st z97 mobo to work with 6x GPUs; it is not the same model as yours; but yours might have a similar setting that needs to be changed.




I had to change the setting called: DMI Gen2 Speed from Disabled:

to Auto

in order to get the sixth card to work properly.





I have 1 more z97 mobo; I'll try it as well.

[fullzero]

This looks nice so far .
I have a GTX 1070 Gaming X with Micron memory.  Anybody knows how much I can push these in Linux ?
In windows I can get them between 30-31 but only if Claymore is already running . If i want to start it with mem oc +600 it crashes

On nvOC is see that my Mh/s is lower it I oc mem to +600. I get around 28mh ... lower then for same settings in Win . Am I doing something wrong ?
Thanks !

I don't have a Gaming X 1070; but some 1070s will OC stability up to 1100.  I have noticed that the OC offsets don't have the same results in linux as in windows; you should have to OC a little higher in linux to have the same result.  However you should also be able to OC higher in linux for this same reason.

When using Ethash, setting the mc OC too high will most likely result in a Claymore soft crash.  Setting the cc OC too high will most likely crash the system.  I would keep bumping up the mc OC by 100 until you get a Claymore soft crash.  Then reduce it by 50, and see if it is ok.  If not down another 50. 

I tried ETH with a 6x zotac 1070 mini rig; for short term at least this was the highest stable setting: cc+200, mc+1100, powerlimit 125 (because more power did not generate more hash).

[citronick]

switched from Teamviewer to ssh...

removed the # from file as per onebash instructions

upon ssh to rig IP

user: root
password: miner1

Error message: Access denied

Anybody can help?

try as the user:

m1

that works thanks!

from where do I find / how edit to the onebash file?

[Nexillus]

This looks nice so far .
I have a GTX 1070 Gaming X with Micron memory.  Anybody knows how much I can push these in Linux ?
In windows I can get them between 30-31 but only if Claymore is already running . If i want to start it with mem oc +600 it crashes

On nvOC is see that my Mh/s is lower it I oc mem to +600. I get around 28mh ... lower then for same settings in Win . Am I doing something wrong ?
Thanks !

I don't have a Gaming X 1070; but some 1070s will OC stability up to 1100.  I have noticed that the OC offsets don't have the same results in linux as in windows; you should have to OC a little higher in linux to have the same result.  However you should also be able to OC higher in linux for this same reason.

When using Ethash, setting the mc OC too high will most likely result in a Claymore soft crash.  Setting the cc OC too high will most likely crash the system.  I would keep bumping up the mc OC by 100 until you get a Claymore soft crash.  Then reduce it by 50, and see if it is ok.  If not down another 50. 

I tried ETH with a 6x zotac 1070 mini rig; for short term at least this was the highest stable setting: cc+200, mc+1100, powerlimit 125 (because more power did not generate more hash).

I noticed too that CC does very little with other hash on Eth, while memory does a lot more.

Also are you mining ZEC right now instead? Do you have some numbers on your 1070's? I was averaging around 440sol/s each.

[fullzero]

switched from Teamviewer to ssh...

removed the # from file as per onebash instructions

upon ssh to rig IP

user: root
password: miner1

Error message: Access denied

Anybody can help?

try as the user:

m1

that works thanks!

from where do I find / how edit to the onebash file?

it is located at:

Code:

'/media/m1/1263-A96E/oneBash'

You probably already know this but I'm sure a lot of members don't :

IMO when navigating by the cmd line the easiest way to is use a few shortcuts

first see what is in the current directory (also called a folder) with:

Code:

ls

this will show you what is there

start typing the desired sub directory or file then press the tab key

as soon as what you have started typing is distinguishable; the tab will finish for you

to move up a directory (to the parent directory) use

Code:

cd ..

the three of these should allow very quick cmd line navigation once you get used to using them.

the last shortcut that is really helpful is to press the up key (single or multiple times) to scroll through previously entered cmds.

[fullzero]

This looks nice so far .
I have a GTX 1070 Gaming X with Micron memory.  Anybody knows how much I can push these in Linux ?
In windows I can get them between 30-31 but only if Claymore is already running . If i want to start it with mem oc +600 it crashes

On nvOC is see that my Mh/s is lower it I oc mem to +600. I get around 28mh ... lower then for same settings in Win . Am I doing something wrong ?
Thanks !

I don't have a Gaming X 1070; but some 1070s will OC stability up to 1100.  I have noticed that the OC offsets don't have the same results in linux as in windows; you should have to OC a little higher in linux to have the same result.  However you should also be able to OC higher in linux for this same reason.

When using Ethash, setting the mc OC too high will most likely result in a Claymore soft crash.  Setting the cc OC too high will most likely crash the system.  I would keep bumping up the mc OC by 100 until you get a Claymore soft crash.  Then reduce it by 50, and see if it is ok.  If not down another 50. 

I tried ETH with a 6x zotac 1070 mini rig; for short term at least this was the highest stable setting: cc+200, mc+1100, powerlimit 125 (because more power did not generate more hash).

I noticed too that CC does very little with other hash on Eth, while memory does a lot more.

Also are you mining ZEC right now instead? Do you have some numbers on your 1070's? I was averaging around 440sol/s each.

ETH uses Ethash which is a memory intensive algorithm.  ZEC uses Equihash which is a core intensive algorithm.

With ZEC using more power will get you more hash.  Going up to powerlimit 165 should give you 465-470 sol/s with cc+200, mc+900 (or higher)

I usually get around 445 sol/s with  cc+200, mc+1100, powerlimit 135

[min3333r]

What happens if you disable the onboard graphics?

That did the trick in combination with gen2 in BIOS for a PCI-E slot.

Managed to get it working on one rig but can't get it on other four. The four rigs after adding sixth card start booting without error but getting no output on any of the GPUs. All rigs are set to start mining on boot, so it must be that the process has stopped somewhere after booting as there is no debug code on the LED display. Going to try with SSH but need to punch some holes in switch first.

I'll keep 'playing' with this and write here if I find a solution.

Now I at least know there is a way. :-)

Thanks man!

Edit:

Thanks for the information and images of the mobo, tried changing to Gen2 earlier before but it doesn't seem to work always like it should ^ ( see text above ). Change to Gen2 only worked on one rig out of five and my mobo doesn't have "DMI Gen Speed" or similar. I'll keep playing with this and write back if I manage something.

Also just to let you know, I have the same problem as post #502 running with 5x 1060.

[Maxximus007]

switched from Teamviewer to ssh...

removed the # from file as per onebash instructions

upon ssh to rig IP

user: root
password: miner1

Error message: Access denied

Anybody can help?

try as the user:

m1

that works thanks!

from where do I find / how edit to the onebash file?

it is located at:

Code:

'/media/m1/1263-A96E/oneBash'

You probably already know this but I'm sure a lot of members don't :

IMO when navigating by the cmd line the easiest way to is use a few shortcuts

first see what is in the current directory (also called a folder) with:

Code:

ls

this will show you what is there

start typing the desired sub directory or file then press the tab key

as soon as what you have started typing is distinguishable; the tab will finish for you

to move up a directory (to the parent directory) use

Code:

cd ..

the three of these should allow very quick cmd line navigation once you get used to using them.

the last shortcut that is really helpful is to press the up key (single or multiple times) to scroll through previously entered cmds.

If you are not used to the command line you can install Midnight Commander. This will give you a more graphical way to move around folders (directories), start the editor etc.

Code:

sudo apt install mc

it will ask for the password.
after that you can start it with:

Code:

mc

[citronick]

switched to ssh because I wanted to a quick way to edit the onebash file and test the rig on various clocks.

I am used to the nano editor so...

root@m1-desktop:/home/m1# nano /media/m1/1263-A96E/oneBash

do the changes

root@m1-desktop:/home/m1# sudo reboot



after reboot, ssh into again

how do I get the mining console?

root@m1-desktop:/home/m1# screen -x m1

doesn't get me anything

[chaimroid]

Hi! First of all, that's so much for making this. It's been super easy to use and (as far as I know) is reporting the highest hashrate for me so far!

I have an ASUS Prime Z270-A with 4 EVGA 1070 SC2 cards. I've had trouble trying to OC them (have seen maybe you can't because they're factory OC'd?) and oneBash spits out errors trying to assign GPUTargetFanSpeed when it starts up. Maybe you know how to get around this?

Thanks so much!

[Maxximus007]

switched to ssh because I wanted to a quick way to edit the onebash file and test the rig on various clocks.

I am used to the nano editor so...

root@m1-desktop:/home/m1# nano /media/m1/1263-A96E/oneBash

do the changes

root@m1-desktop:/home/m1# sudo reboot



after reboot, ssh into again

how do I get the mining console?

root@m1-desktop:/home/m1# screen -x m1

doesn't get me anything

Have had the same problem, couldn't get it to work by default.
The 2unix is made autostart, and can't be reached with screen or tmux because of it probably.

Have had to change things to oneBash, but because of that an auto restart of oneBash is not possible anymore. At least I wasn't able to get it done..

[Maxximus007]

Hi! First of all, that's so much for making this. It's been super easy to use and (as far as I know) is reporting the highest hashrate for me so far!

I have an ASUS Prime Z270-A with 4 EVGA 1070 SC2 cards. I've had trouble trying to OC them (have seen maybe you can't because they're factory OC'd?) and oneBash spits out errors trying to assign GPUTargetFanSpeed when it starts up. Maybe you know how to get around this?

Thanks so much!

Do you use a fast USB stick, as recommended? Setting power limit can cause some problems on slow ones. What is the trouble you're getting with OC? What are the values used?

[citronick]

I have two identical 6 x Zotac 1070 Mini rigs -- exactly the same config.

Upgraded now to nvoc v15 and works nicely on MSI Z270-A-PRO motherboard, i3-7100 CPU and 4GB RAM, 32GB USB3 Sandisk Cruzer Glide

Mining ZEC at Nicehash at 440-ish sols per card.

+200/+1100/165 PL

1100watts from the wall

The issue is not an issue but a strange observation....

The network port of the mobo on the 1st rig is amber/green... which is normal for a gigabit connection.

However the 2nd rig, is amber/amber.... which I think is 100base instead on gigabit/1000base

Both rigs do hash well, ranges from 2400-2800sols, nothing strange I think.

Anyone encountered a similar case?

Anybody tried to convert a similar rig like this for ETH mining? Any good?

[fullzero]

switched to ssh because I wanted to a quick way to edit the onebash file and test the rig on various clocks.

I am used to the nano editor so...

root@m1-desktop:/home/m1# nano /media/m1/1263-A96E/oneBash

do the changes

root@m1-desktop:/home/m1# sudo reboot



after reboot, ssh into again

how do I get the mining console?

root@m1-desktop:/home/m1# screen -x m1

doesn't get me anything

Have had the same problem, couldn't get it to work by default.
The 2unix is made autostart, and can't be reached with screen or tmux because of it probably.

Have had to change things to oneBash, but because of that an auto restart of oneBash is not possible anymore. At least I wasn't able to get it done..

So if I got the situation right; currently you have:

SSHed into rig

opened oneBash on that rig with nano:

Code:

nano '/media/m1/1263-A96E/oneBash'

made changes / edits; then ctrl + shift + X to quit selecting Y to save changes

the next part is this:

Find the currently running GPU mining process. 

We will do this by finding the gnome-terminal process.

Code:

ps aux | grep gnome-terminal

this should list 2 processes: we are interested in the first one which ends in /gnome-terminal-server

note its pid ( should be a 4 digit number at the beginning of the process listing; but may be more than 4 digits )

type:

Code:

kill pid

where pid is the 4 or higher digit number; so if the pid was 2037  I would use:

Code:

kill 2037

now we have stopped the GPU mining process

as we have already changed oneBash we will now restart the GPU mining process

to do this we enter:

Code:

gnome-terminal

and now we have remote modified the nvOC rig 

[fullzero]

What happens if you disable the onboard graphics?

That did the trick in combination with gen2 in BIOS for a PCI-E slot.

Managed to get it working on one rig but can't get it on other four. The four rigs after adding sixth card start booting without error but getting no output on any of the GPUs. All rigs are set to start mining on boot, so it must be that the process has stopped somewhere after booting as there is no debug code on the LED display. Going to try with SSH but need to punch some holes in switch first.

I'll keep 'playing' with this and write here if I find a solution.

Now I at least know there is a way. :-)

Thanks man!

Edit:

Thanks for the information and images of the mobo, tried changing to Gen2 earlier before but it doesn't seem to work always like it should ^ ( see text above ). Change to Gen2 only worked on one rig out of five and my mobo doesn't have "DMI Gen Speed" or similar. I'll keep playing with this and write back if I manage something.

Also just to let you know, I have the same problem as post #502 running with 5x 1060.

Last night I couldn't get my second z97 mobo to work with 6x GPUs. 

Updating the mobo bios is probably the last thing to try.  Unfortunately it didn't help in my case.

[fullzero]

Hi! First of all, that's so much for making this. It's been super easy to use and (as far as I know) is reporting the highest hashrate for me so far!

I have an ASUS Prime Z270-A with 4 EVGA 1070 SC2 cards. I've had trouble trying to OC them (have seen maybe you can't because they're factory OC'd?) and oneBash spits out errors trying to assign GPUTargetFanSpeed when it starts up. Maybe you know how to get around this?

Thanks so much!

Do you use a fast USB stick, as recommended? Setting power limit can cause some problems on slow ones. What is the trouble you're getting with OC? What are the values used?

A GPU being factory OCed means it has a higher base clock; shouldn't have any other impact than using lower offset OC values.

When the GPU mining process starts; before the fancontrol error messages do you see card core clock and memory clock OC messages for cards 0,1,2, and 3 ?

[WarwickNZ]

Hi,

Firstly I'd just like to say, great job I really appreciate the time you put into this and thanks for sharing.

I'm having an issue where I can't get the cards to overclock, smi reports that the powerlimit is set as wanted at 165 through OneBash but the draw wattage just doesn't move just stays stuck around 150w. Clock and memory is set as per your guidance +100 core, +600 mem.

Board is ASUS H270 Prime Plus with 6 x EVGA GTX1070 SC, dual mining ETH+SC

Any help you could offer would be greatly appreciated, loving the plug and play aspect of it, will really help with getting some friends going.

Thanks!

Sometimes, especially with initial boot; a powerlimit will not set as intended.  This may be due to using a slow usb, dos2unix taking longer than expected to close, or Ubuntu installing security updates.  It is also possible to set a powerlimit that is outside of the cards acceptable range for a powerlimit; in which case it will not implement any powerlimit dispite displaying the command.

To manually implement a powerlimit:

Press f12 to open the guake terminal; if cpuminer is running open another tab or press ctrl + c to stop the process and enter:

sudo nvidia-smi -pl 165

you will then be prompted for the password: miner1

this should attempt to implement the command for each card.  You will see a message for each card listing its previous powerlimit and its new one.  If you have selected a powerlimit that is outside of the acceptable range smi will indicate this.

Generally if you are having trouble with oneBash implementing a powerlimit: I recommend changing:

INDIVIDUAL_POWERLIMIT="YES"

in addition to:

POWERLIMIT="YES"

and then ensuring you enter the desired wattage for the global powerlimit and each of your cards individual powerlimits.

Even with a slow usb this should ensure the proper setting of a powerlimit.

Thanks. I sorted the issue just need to be a bit more aggressive with the memory overclock, stable on all rigs @ 1100 and some rigs at 1200. Interestingly the power levels still aren't moving, staying at stock for Dual Mining ETH+SIA, power limit does appear to have applied itself though.

Any scope of updating the image to Claymore 9.5? Just out of interest, how many rigs are you operating?

Again, great project, be happy to help if I can at all.