[Plugin] Nvidia-Driver


ich777

Recommended Posts

On 9/20/2021 at 9:18 AM, ich777 said:

I think you are talking about 460.73.01

 

If you are talking about 460.73.01, you can but you have to be on unRAID version 6.9.2

 

The reason for this is that I only compile the latest current available Production and the New Feature Branch from the driver at the time a new unRAID version is released and all following drivers for this unRAID version.

After a new unRAID version is released this process starts from the beginning.

 

Also don't forget that older versions from the Nvidia driver may fail to compiles on newer Kernel versions.

You are correct. I mean the 460.73.01.

I am running 6.9.2, but that version is not currently listed in the available driver downloads. What can I do to download and install it anyway?

Link to comment
54 minutes ago, Rolucious said:

I am running 6.9.2, but that version is not currently listed in the available driver downloads.

That's because I'm listing only the 8 last drivers since this can be a really big mess...

 

55 minutes ago, Rolucious said:

What can I do to download and install it anyway?

Edit the file "/boot/config/plugins/nvidia-driver/settings.cfg" and change the line where it says "driver_version=..." to:

driver_version=460.73.01

 

After that open up the plugin page and you should see that nothing is selected, just ignore that and click the Download button, then you should see that it downloads the driver 460.73.01, after it finished reboot your server to install it.

  • Thanks 2
Link to comment
On 9/21/2021 at 5:58 AM, ich777 said:

That's because I'm listing only the 8 last drivers since this can be a really big mess...

 

Edit the file "/boot/config/plugins/nvidia-driver/settings.cfg" and change the line where it says "driver_version=..." to:

driver_version=460.73.01

 

After that open up the plugin page and you should see that nothing is selected, just ignore that and click the Download button, then you should see that it downloads the driver 460.73.01, after it finished reboot your server to install it.

 Will this work with 390.144?

Link to comment
11 minutes ago, Wicked_Chicken said:

I have a old Geforce 520 I want to dedicate to Deepstack processing. 

The GeForce GT 520 has 48 Cuda cores I think this would be a waste of electricity even if it draws around 30 Watts...

A Quadro P400 has 256 Cuda cores and also draws around 30Watts...

 

Anyways, no the drivers or better speaking the container toolkit, runtime,... only supports drivers >= 418.81.07 so you couldn't get it to work in a container anyways, besides that I only compile the latest Nvidia driver that is available when a new unRAID version is released and also the following Nvidia drivers for this release cycle and start over again when a new unRAID version is released.

Link to comment
5 minutes ago, ich777 said:

The GeForce GT 520 has 48 Cuda cores I think this would be a waste of electricity even if it draws around 30 Watts...

A Quadro P400 has 256 Cuda cores and also draws around 30Watts...

 

Anyways, no the drivers or better speaking the container toolkit, runtime,... only supports drivers >= 418.81.07 so you couldn't get it to work in a container anyways, besides that I only compile the latest Nvidia driver that is available when a new unRAID version is released and also the following Nvidia drivers for this release cycle and start over again when a new unRAID version is released.

That makes sense, Lol. I appreciate the response and your work as a dev!!! I'll see if I can't snag something more recent for this project.

  • Like 1
Link to comment

Hello guys!

 

I am trying to run my old Quadro FX 1800 on my server but I can not seem to get it to work.

Maybe my HW is just too old?

 

I know that unraid sees the card, because of the following lscpi -v output:

28:00.0 VGA compatible controller: NVIDIA Corporation G94GL [Quadro FX 1800] (rev a1) (prog-if 00 [VGA controller])
        Subsystem: NVIDIA Corporation G94GL [Quadro FX 1800]
        Flags: bus master, fast devsel, latency 0, IRQ 5, IOMMU group 16
        Memory at f6000000 (32-bit, non-prefetchable) [size=16M]
        Memory at e0000000 (64-bit, prefetchable) [size=256M]
        Memory at f4000000 (64-bit, non-prefetchable) [size=32M]
        I/O ports at e000 [size=128]
        Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
        Capabilities: [60] Power Management version 3
        Capabilities: [68] MSI: Enable- Count=1/1 Maskable- 64bit+
        Capabilities: [78] Express Endpoint, MSI 00
        Capabilities: [b4] Vendor Specific Information: Len=14 <?>
        Capabilities: [100] Virtual Channel
        Capabilities: [128] Power Budgeting <?>
        Capabilities: [600] Vendor Specific Information: ID=0001 Rev=1 Len=024 <?>
        Kernel modules: nvidia_drm, nvidia

 

Unforunately i still get the following error:

"NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running."

 

Hope anyone can help!

Thanks in advance.

Link to comment
6 minutes ago, Andru said:

Unforunately i still get the following error:

"NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running."

Like the first post in the troubleshooting section tells you, you can get a list of supported cards here: Click

 

Your card seems too old to me...

  • Like 1
Link to comment
I don't want to include nvtop in the Nvidia Driver plugin itself because it's not a part from the official driver but I can look into it if it's possible to create a plugin.
 
May I ask why or for what do you need it?
Well I'd hate to have a whole plugin just for one command. I can check the with the nerdpack author and see if they are interested?
Link to comment
6 minutes ago, SeeGee said:

Well I'd hate to have a whole plugin just for one command. I can check the with the nerdpack author and see if they are interested?

As said above, since this is a third party plugin and not part of the driver itself I really don't want to include it in the driver package itself, hope this makes sense to you.

 

I think it wouldn't make much of a difference if you install it through the nerdpack or through a plugin the benefit of having a plugin is that when a update from nvtop drops it easier to install it or at least to update it.

Link to comment

Fantastic! Installed and working. However I do get these errors when I run it, but it does not seem to affect functionality.

nvtop: /lib64/libtinfo.so.6: no version information available (required by nvtop)
nvtop: /lib64/libncursesw.so.6: no version information available (required by nvtop)

 

Thank you so much! 

  • Like 1
Link to comment

Sorry if this is a stupid question.  But I am wondering is there a way to use a single  Nvidia GPU on unRAID with this plugin installed. Then be able to ensure that no services are using it and have a VM grab that same card for use.  ?

 

Then when I want to use the card again in say a docker shut down the VM and start up those dockers?

 

 

Link to comment
5 minutes ago, Aceriz said:

is there a way to use a single  Nvidia GPU on unRAID with this plugin installed.

I don't recommend it doing it that way but it is possible but can result in lock ups from the entire server when done wrong.

 

7 minutes ago, Aceriz said:

Then when I want to use the card again in say a docker shut down the VM and start up those dockers?

You have create a custom script or something similar which you can trigger from unRAID itself and you should be good to go.

 

You can start a VM from a script or the CLI with:

virsh start VMNAME

 

You also can do that with a container:

docker start CONTAINERNAME

 

Replace 'start' with 'stop' to stop the VM/containers.

 

Hope that helps.

Link to comment
11 minutes ago, ich777 said:

I don't recommend it doing it that way but it is possible but can result in lock ups from the entire server when done wrong.

 

You have create a custom script or something similar which you can trigger from unRAID itself and you should be good to go.

 

You can start a VM from a script or the CLI with:

virsh start VMNAME

 

You also can do that with a container:

docker start CONTAINERNAME

 

Replace 'start' with 'stop' to stop the VM/containers.

 

Hope that helps.

Thanks for the information.

 

 

 

Do you know if there is a way to stop the unRAID GUI mode from grabbing the Nvidia driver when booted.  In my setup I have it configured where I use an onboard graphics card for unRAID. In the bios I directly set this to be the primary display.   But have noticed when I tried to install the Nvidia plugin (without vfio binding on card so that it was recognized without the Nvidia smi error ) that the gui grabs the Nvidia card.(I know this is happening as my unRAID GUI goes to the display port output which is not the one for the onboard (vga). As the GUI is grabbing Nvidia driver and card.  one process is showing in gpu stats even when no dockers or anything are connected. Which stopped my I think from loading the VM hence the first post.

 

 

 

I am using GUI mode mainly for ability to change display input on my PC to vga  access unRAID terminal directly incase my VM crashes or has issues...

 

 

 

Thanks for thoughts

Link to comment
16 minutes ago, Aceriz said:

Do you know if there is a way to stop the unRAID GUI mode from grabbing the Nvidia driver when booted.

This is a setting that you have to in your BIOS when you got integrated or onboard graphics.

You have to set the primary graphics adapter to integrated or onboard.

 

17 minutes ago, Aceriz said:

But have noticed when I tried to install the Nvidia plugin (without vfio binding on card so that it was recognized without the Nvidia smi error ) that the gui grabs the Nvidia card.

On what unRAID version are you?

 

Open up a Terminal from unRAID and try this command and then reboot:

sed -i '/disable_xconfig=/c\disable_xconfig=true' /boot/config/plugins/nvidia-driver/settings.cfg

(no guarantee that it works but you can try it)

 

 

If you have no GUI output after you've rebooted issue this command from a unRAID terminal and reboot again:

sed -i '/disable_xconfig=/c\disable_xconfig=false' /boot/config/plugins/nvidia-driver/settings.cfg

(this will revert the changes from above)

 

21 minutes ago, Aceriz said:

which is not the one for the onboard (vga).

Is something plugged into the VGA port, some onboard controllers need actually a monitor or dummy plug connected to work even when you've set the primary GPU to the internal/onboard one.

Link to comment

Hello 

 

I have recently put an old NVIDIA Quadro 4000 (2010) in my unraid and tried (and hoped) the plugin would pick it up so I can use it with docker but after installation and 2 reboots I still can't get the UUID from the plugin.

 

I only get:

"NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running."

 

Maybe someone else have seen this and have a solution for it? (if possible).

 

 

Thanks in advance!

 

 

BR

Link to comment
10 minutes ago, Digiboy said:

I have recently put an old NVIDIA Quadro 4000 (2010) in my unraid and tried (and hoped) the plugin would pick it up so I can use it with docker but after installation and 2 reboots I still can't get the UUID from the plugin.

Please read this reply:

 

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.