[Plugin] Nvidia-Driver

[ich777]

31 minutes ago, cenzar said:

I'm having an issue where the Nvidia Driver isn't detecting/reporting GPUs. I had a 1050ti in my first PCiE slot and everything was working no problems. I recently moved the 1050ti to the 3rd slot and added a 3060 to the 1st slot. This is a configuration I had in the server a few months ago. But now the driver isn't finding the GPUs even though Unraid is and can use them in VMs. I tried deleting the plugin and driver and reinstalling but no luck, any help would be greatly appreciated.

Sorry but without Diagnostics I can't say anything.

[cenzar]

2 hours ago, ich777 said:

Sorry but without Diagnostics I can't say anything.

Apologies, that was my mistake diagnostics attacheddiagnostics-20240903-1638.zip

[alturismo]

7 hours ago, cenzar said:

Apologies, that was my mistake diagnostics attacheddiagnostics-20240903-1638.zip

may take a look at your vfio binds

 

BIND=0000:01:00.0|10de:2504 0000:01:00.1|10de:228e 0000:04:00.0|10de:1c82 0000:04:00.1|10de:0fb9

 

both slots are vfio bound, which makes then unaccessable for unraid host (VM usage only)

 

01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GA106 [GeForce RTX 3060 Lite Hash Rate] [10de:2504] (rev a1)
	Subsystem: Dell GA106 [GeForce RTX 3060 Lite Hash Rate] [1028:c974]
	Kernel driver in use: vfio-pci
	Kernel modules: nvidia_drm, nvidia
01:00.1 Audio device [0403]: NVIDIA Corporation GA106 High Definition Audio Controller [10de:228e] (rev a1)
	Subsystem: Dell Device [1028:c974]
	Kernel driver in use: vfio-pci

04:00.0 VGA compatible controller [0300]: NVIDIA Corporation GP107 [GeForce GTX 1050 Ti] [10de:1c82] (rev a1)
	Subsystem: Micro-Star International Co., Ltd. [MSI] GP107 [GeForce GTX 1050 Ti] [1462:8c96]
	Kernel driver in use: vfio-pci
	Kernel modules: nvidia_drm, nvidia
04:00.1 Audio device [0403]: NVIDIA Corporation GP107GL High Definition Audio Controller [10de:0fb9] (rev a1)
	Subsystem: Micro-Star International Co., Ltd. [MSI] GP107GL High Definition Audio Controller [1462:8c96]
	Kernel driver in use: vfio-pci

 

[SugiUnraid]

Hi there - I'm having issues using my 4060 in an X470D4U board with a 5950x. LSPCI sees the card but keep getting the nvidia-smi failed error. I've also made sure it's not bound to VFIO as well. Attached are the diagnostics as well. Thank you for the help!

njhome-diagnostics-20240904-0745.zip

[ich777]

1 hour ago, SugiUnraid said:

Hi there - I'm having issues using my 4060 in an X470D4U board with a 5950x. LSPCI sees the card but keep getting the nvidia-smi failed error. I've also made sure it's not bound to VFIO as well. Attached are the diagnostics as well. Thank you for the help!

Please enable Above 4G decoding, Resizable BAR support and Support for large address space in your BIOS (you won't find every option in your BIOS but enable if you find anything).

 

This error indicates that something is wrong with allocating address space, please maybe also try another PCIe slot if your motherboard has another slot:

Sep  4 04:36:20 NJHome kernel: NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
Sep  4 04:36:20 NJHome kernel: NVRM: BAR0 is 0M @ 0x0 (PCI:0000:2b:00.0)
Sep  4 04:36:20 NJHome kernel: nvidia: probe of 0000:2b:00.0 failed with error -1
Sep  4 04:36:20 NJHome kernel: NVRM: The NVIDIA probe routine failed for 1 device(s).
Sep  4 04:36:20 NJHome kernel: NVRM: None of the NVIDIA devices were initialized.

 

[SkepticalOptimist]

Attempted upgraded to 6.12.13 today from 6.12.11. 

The upgrade itself failed and lost both network support and found corrupted BZ files which I manually restored to USB.

Was able to reboot and find build still at 6.12.11.

Ran upgrade again and rebooted. This time on 6.12.13. Things seem to be working. HOWEVER..

No NVIDIA plug-in showing.  IOMMU does have the card showing up:

IOMMU group 28:                [8086:6f04] 00:02.0 PCI bridge: Intel Corporation Xeon E7 v4/Xeon E5 v4/Xeon E3 v4/Xeon D PCI Express Root Port 2 (rev 01)
     [10de:2484] 03:00.0 VGA compatible controller: NVIDIA Corporation GA104 [GeForce RTX 3070] (rev a1)
     [10de:228b] 03:00.1 Audio device: NVIDIA Corporation GA104 High Definition Audio Controller (rev a1)

 

is the upgrade failure/loss of NVIDIA related or a red herring? Should I attempt to reinstall the plugin?

Edited September 4 by SkepticalOptimist

[ich777]

25 minutes ago, SkepticalOptimist said:

is the upgrade failure/loss of NVIDIA related or a red herring? Should I attempt to reinstall the plugin?

Yes it might be, the plugin and the Plugin Update Helper which is part of the Nvidia Plugin try to download the version that is necessary for the new Unraid version and you usually get a notification that the download is in progress/finished/save to reboot.

 

If you don't wait and directly reboot then it will most certainly fail to download and leave the plugin in bricked state.

However I can't say much since you didn't provide Diagnostics if that's really the case ore something else is going on.

 

Please remove the plugin (also check the Plugin Error tab if it's in there) and reinstall it from the CA App.

[coltonc18]

Hello, 

I went to install the new OS version, and got an error on Nvidia driver download. 

[ich777]

4 hours ago, coltonc18 said:

I went to install the new OS version, and got an error on Nvidia driver download. 

Can you please post your Diagnostics?

 

How much free space do you have on your USB Boot device?

Is your server able to communicate with GitHub and the GitHub API?

Do you have any AdBlockers on your network?

Can you post a screenshot from the Nvidia Driver plugin page, are there any errors on there displayed?

 

You could try to remove the plugin, reboot and then install the plugin again after the reboot.

[coltonc18]

11 hours ago, ich777 said:

Can you please post your Diagnostics?

 

How much free space do you have on your USB Boot device?

Is your server able to communicate with GitHub and the GitHub API?

Do you have any AdBlockers on your network?

Can you post a screenshot from the Nvidia Driver plugin page, are there any errors on there displayed?

 

You could try to remove the plugin, reboot and then install the plugin again after the reboot.

looks like removing the plugin and reinstalling did it.....

The upgrade got hung up trying to upgrade third party drivers after that, but I cancelled the upgrade, and redid it, and that seems to have done it.  It didn't recognize any issues at that point.  

Thanks!

[lacariuz]

Hi I just installed unraid for the first time. so I'm completely green in the area I want to start transcoding on jellyfin but it's not going so well. Have an i5 6600k cpu and a GTX 1050. want to use the graphics card for transcoding. Have tried to follow a lot of different guaids but nothing wants to work.

 

The last one I tried is from the jellyfin forum.

 

Extra Parameters: -e NVIDIA_DRIVER_CAPABILITIES=all  -e NVIDIA_VISIBLE_DEVICES=all  --gpus all
That's the exact line I used, I copy and pasted exactly that, obviously without the Extra Parameters part.
Than I created 2 variables in the container editor.

Variable 1:
Name: NVIDIA_VISIBLE_DEVICES
Key: NVIDIA_VISIBLE_DEVICES
value: all

Variable 2:
Name: NVIDIA_DRIVER_CAPABILITIES
Key: NVIDIA_DRIVER_CAPABILITIES
Value: all

 

Is there something obvious I'm missing?

kraken-diagnostics-20240910-2125.zip

[ConnerVT]

How about this, from the first page of this thread:

 

[lacariuz]

First thing i tried.

[ich777]

1 hour ago, lacariuz said:

Is there something obvious I'm missing?

Please read the second post again:

Quote

1. Add '--runtime=nvidia' in your Docker template in 'Extra Parameters' (you have to enable 'Advanced view' in the template to see this option)
2. Add a variable to your Docker template with the Key: 'NVIDIA_VISIBLE_DEVICES' and as Value: 'YOURGPUUUID' (like 'GPU-9cfdd18c-2b41-b158-f67b-720279bc77fd')
3. Add a variable to your Docker template with the Key: 'NVIDIA_DRIVER_CAPABILITIES' and as Value: 'all'
4. Make sure to enable hardware transcoding in the application/container itself

 

You never mentioned --runtime=nvidia and this is also a really important part.

Please post your docker run command if possible (just click the link to read how to do that).

[lacariuz]

I tried that but did not get it to work.

 

will try to fix ” Docker run ” after work today.

[ich777]

1 hour ago, lacariuz said:

I tried that but did not get it to work.

What container are you using? Do you use the official one? I would recommend you go with the official one first.

 

From what I saw this can only be a configuration error since your GPU seems to work fine.

[heavyghost]

On 8/21/2024 at 2:02 PM, ich777 said:

Other than that you can just step down to a Pascal card eg: Nvidia Quadro P400 however I would recommend that you invest a bit more and spend the extra on the T400

Haha perfect! Thanks for the tip, I grabbed a P400 for $35 USD, should arrive next week. I realise a T400 would be better but I'm budget-maxxing this build.... 🤪 

[lacariuz]

I have tried the binhex-jellyfin container and now i use the official one.

 

I'll be extra careful and really check so I enter everything correctly when I enter all the parameters. When I enter the variables, is it important to use uppercase and lowercase letters?

[ich777]

15 minutes ago, lacariuz said:

I'll be extra careful and really check so I enter everything correctly when I enter all the parameters. When I enter the variables, is it important to use uppercase and lowercase letters?

Yes, did you follow the instructions step by step from the second post in this thread?

Please also double check for any spaces in the GPU UUID at the beginning or the end.

It should be pretty easy to set up with the second post since I included a tutorial for Emby/Jellyfin and Plex.

 

Please also don't forget to post your docker run command.

[lacariuz]

I think I got everything in. but it won't work. Can you see anything??

[lacariuz]

And I restarted jellyfin and the server.

 

Why is jellyfin active app inte the gpu and do nothing ?

Edited Wednesday at 03:15 PM by lacariuz

[lacariuz]

or is it decoding now? Not sure what to watch.

[ich777]

58 minutes ago, lacariuz said:

or is it decoding now? Not sure what to watch.

It is indeed decoding, these indicators show that Jellyfin is using the GPU and a the bars on top you can see that it has some work to do:

 

You just have to make sure that you force a transcode that it will actually start transcoding, please also don't forget that the load on the CPU is probably the audio that needs to be transcoded too.

[lacariuz]

Thanks for the help

[UnraidNewbie2211]

Hey. I've installed unraid on my brand-new server:

 

- ASUSTeK COMPUTER INC. ROG STRIX B650E-I GAMING WIFI

- AMD Ryzen 9 7950X

- RTX 4070 TI SUPER

 

NVIDIA drivers installed correctly and nvidia-smi command works both inside containers and on host machine, but nvcc command is not found and /usr/local directory does not contain anything related to cuda.

Is it possible to install nvcc with drivers & make it system-wide and accessible inside containers?
I noticed, that i can't enable HW accel. for ffmpeg inside homebridge, because it looks at nvcc command existence at build time: https://github.com/homebridge/ffmpeg-for-homebridge/blob/latest/build-ffmpeg

 

 

My output:

 

And the last question is why GPU panel is not appearing at the dashboard?
Thanks!

homie-diagnostics-20240912-1442.zip