1 minute ago, gecl said:

 

Not passing GPU through to VM

When GPU is connected all the HDMIs on the Motherboard and the GPU stop transmitting. Is it a bios setting to make the motherboard GPU still display boot screen?

You set IGD in the bios as primary display

4 minutes ago, SimonF said:

You set IGD in the bios as primary display

Thanks, and what has the GPU display boot?

15 minutes ago, gecl said:

Thanks, and what has the GPU display boot?

 

Hi there.
I am using latest Unraid 7.0.0-rc.1 and have an ASUS PRIME Z790M-PLUS motherboard with a nvidia RTX 4060Ti 16Gb installed.
If i install the latest nvidia driver through this plugin i get the following dmesg error output:

Since ASUS does not offer any support for Linux Distros, i was wondering if this is a driver problem and can be fixed through an upcoming update.

I also have the problem, that in the moment i plug in any nvidia GPU into my motherboard (regardless of i installed the nvidia driver plugin or not), my system is not able to reach higher c-states than C3 anymore. If i remove the nvidia card i reach C10 again.

Does someone else experience this and is there any workaround?

Thanks in advance and kind regards

5 minutes ago, Stogie said:

If i install the latest nvidia driver through this plugin i get the following dmesg error output

Please post your Diagnostics, there is too less information in your post.

 

5 minutes ago, Stogie said:

my system is not able to reach higher c-states than C3 anymore. If i remove the nvidia card i reach C10 again.

This is expected, if you install a PCIe addon card you won't be able to reach anything higher than C3, however this was often discussed with addon cards especially in the German sub thread, even if you are now on C3 you won't be drawing significantly more power, the higher power consumption is most of the times just the addon card and not caused by the power state.

Hi there.
Thanks for your reply.
I share my diagnostics here.

According the C-State topic it is just confusing for me, that my system takes 17 Watt without the nvidia gpu, but 36 Watt in idle with the gpu plugged in although the gpu statistics plugin shows 5-7 Watt in P8/Idle-Mode.
 

diagnostics.zip

 

15 minutes ago, Stogie said:

According the C-State topic it is just confusing for me, that my system takes 17 Watt without the nvidia gpu, but 36 Watt in idle with the gpu plugged in although the gpu statistics plugin shows 5-7 Watt in P8/Idle-Mode.

that is (sadly) normal ... you are using a RTX Card which itself needs some Watt (which you can see) and now ... your Mainboard also needs more Power as there is a card to supply, used pcie lanes, ... and so on ... so yes, normal behaviour and explained many times on all c state, power consumption related topics  a GPU will for sure ...

 

assuming this card if for your Win 11 VM ... thats how it is

 

if you want to use this card for transcoding ... take the card out and use your (very efficent) iGPU ...

1 hour ago, Stogie said:

If i install the latest nvidia driver through this plugin i get the following dmesg error output

This is a firmware bug, so to speak caused by your BIOS, nothing that I can do about.

Watch out for new BIOS updates and update as soon as one is available.

 

15 minutes ago, Stogie said:

According the C-State topic it is just confusing for me, that my system takes 17 Watt without the nvidia gpu, but 36 Watt in idle with the gpu plugged in although the gpu statistics plugin shows 5-7 Watt in P8/Idle-Mode.

I agree with what @alturismo wrote above.

If you just user your dGPU for transcoding use your iGPU instead.

Okay, i see.
One last question:
I am still confused if i should use the latest version or the Open Source Driver.
Currently i use the Open Source Driver, but i am not sure, if it is better or not.
In the description it says that i have to create a nvidia.conf file, but the link redirects to an old Readme file from version 515.76, where it says, that creating that file with the parameter "options nvidia NVreg_OpenRmEnableUnsupportedGpus=1" will be removed in an upcoming update.

If i check the Readme from latest 565.57 driver, it seems like it is not necessary anymore to create that file.
Is that correct? Because then maybe the plugin description needs an update.
https://download.nvidia.com/XFree86/Linux-x86_64/565.57.01/README/kernel_open.html

 

Thanks in advance and kind regards

3 hours ago, Stogie said:

I am still confused if i should use the latest version or the Open Source Driver.
Currently i use the Open Source Driver, but i am not sure, if it is better or not.

This is totally up to your preference, in my opinion there is no black and white since nobody knows what's in the propietary driver, it just works.

Use whatever you like and works for you.

If you are just using it for transcoding it really doesn't matter.

 

3 hours ago, Stogie said:

In the description it says that i have to create a nvidia.conf file, but the link redirects to an old Readme file from version 515.76, where it says, that creating that file with the parameter "options nvidia NVreg_OpenRmEnableUnsupportedGpus=1" will be removed in an upcoming update.

Yes, but this driver version still requires it.

 

3 hours ago, Stogie said:

Because then maybe the plugin description needs an update.

This is already taken into account and this message is displayed only when it's needed, when it's not needed anymore then it is simply not there. For example when Unraid 7.0.0 stable is released (and possible the next RC version) this message wont be there because the driver package doesn't needs it anymore.

Hello,
from a nubby 
I suddenly have a problem finding my GPU. 

I recently added a PCI expansion card for two extra HDDs, and everything was running fine until I added the two new drives to my pool. Now, suddenly, my GPU has disappeared.

Any advice 

Thanks in advance and kind regards

tower-diagnostics-20241223-1739.zip

40 minutes ago, alex-dev said:

I suddenly have a problem finding my GPU. 

Sorry but I also can't find the GPU in your Diagnostics, it's not even listed in your pci devices list.

 

Please make sure that the card is seated correctly (maybe reseat it) and external power is plugged in (if the card needs it).

Anyone else having an issue with this plugin in Plex?  I have tried the official Plex docker and the linuxserver.io docker and they have the same behavior.  It will work on automatic, but if I manually change the resolution/transcode it will kill the encode.  I installed the GPU monitor and you can see the encode/decode go to zero and never recover.  It's like the card "gives up" if you change it manually.  I know this isn't the plex docker thread but its the same nvidia driver doing the same thing on two different dockers.  I don't think it is a configuration issue as I can see the card in Plex and can see it transcoding via the plex dash app on my phone (before I manually change the resolution/bitrate).

23 minutes ago, rxnelson said:

Anyone else having an issue with this plugin in Plex?

Please post your Diagnostics.

 

23 minutes ago, rxnelson said:

It will work on automatic, but if I manually change the resolution/transcode it will kill the encode.

Do you try this from their web player,so to speak from a web browser, or a native app (Android/iOS/...)?

On 12/26/2024 at 1:24 AM, ich777 said:

Please post your Diagnostics.

 

Do you try this from their web player,so to speak from a web browser, or a native app (Android/iOS/...)?

It was from Chrome.  I tested it on the Apple TV and it seems to work.  Since you specifically mentioned it, known issue?  Diagnostics still needed?

53 minutes ago, rxnelson said:

Since you specifically mentioned it, known issue?

Yes, this is a known issue with their web player (for years now).

 

To circumvent that you have to set a fixed quality in the options and then start the playback in the web player, if you switch the quality in the web player it will simply stop.

 

54 minutes ago, rxnelson said:

Diagnostics still needed?

If this answeres your question, no.

7 hours ago, ich777 said:

 

If this answeres your question, no.

It does.  Thank you very much for your help and quick responses.

I just upgraded to Unraid 6.12.14 and then upgraded the nvidia driver from 535.216.03 to 565.57.01 and now my dockers will not use my GPU. I have since tried several other versions of the drivers available, rebooting server after each installation and dockers still do not see the GPU. The dockers worked fine utilizing the GPU before. The original driver I was using does not show as an option in the list of available drivers in the nvidia addon. Motherboard is ASUS Proart Z790 Creator WIFI, GPU is GTX 1660 TI, and Intel 19-13900K. Using ter GPU with Plex docker as well as Code Project AI Docker. First image is a error on startup after first trying to upgrade driver, 3rd image is a error on plex docker startup.

 

server-diagnostics-20250106-2136.zip

2 hours ago, Itsaulgoodman said:

and now my dockers will not use my GPU

Please let this split up into two sections Code AI and Plex.

 

2 hours ago, Itsaulgoodman said:

Code Project AI Docker

I assume you are talking about this container:

Can you please share a bit more detail on how do you run the container?

Which repository, do you also use a tag for the container?

Maybe share a screenshot from the Docker template with Advanced View enabled.

 

2 hours ago, Itsaulgoodman said:

with Plex docker

What is not working exactly?

The log shows not really an error that is related to the Nvidia GPU.

What container are you using (repository)?

Do you force a transcode in the webplayer from Plex or how did you do to test that it's not working any more?

 

 

From what I see in your Diagnostics the driver is installed and working fine.

This exact driver page which you are using has about 15k downloads on it and I assume that everything is working for most people.

7 minutes ago, ich777 said:

Please let this split up into two sections Code AI and Plex.

 

I assume you are talking about this container:

Can you please share a bit more detail on how do you run the container?

 

Not sure what you mean, I use it as AI for Blue Iris security Cameras.

 

7 minutes ago, ich777 said:

Which repository, do you also use a tag for the container?

Maybe share a screenshot from the Docker template with Advanced View enabled.

 

Added below

 

7 minutes ago, ich777 said:

 

What is not working exactly?

I am not getting hardware transcoding on Plex, and the Code Project AI is also not utilizing the GPU. Code Project AI docker should be using the GPU constant 24 hours a day as the cameras process everything. All of this worked prior to updating drivers. When looking at nvidia-smi, it shows nothing using the gpu, again, code project AI is in constant use of GPU before upgrade.

 

7 minutes ago, ich777 said:

The log shows not really an error that is related to the Nvidia GPU.

What container are you using (repository)?

Plex-Media_Server , not the binhex container

 

7 minutes ago, ich777 said:

Do you force a transcode in the webplayer from Plex or how did you do to test that it's not working any more?

 

Yes, force transcode on plex, does not utilize hardware transcoding. It worked before upgrading drivers

 

7 minutes ago, ich777 said:

 

 

From what I see in your Diagnostics the driver is installed and working fine.

This exact driver page which you are using has about 15k downloads on it and I assume that everything is working for most people.

 

18 minutes ago, Itsaulgoodman said:

I assume most certainly the tag is wrong.

Please try to remove the tag from the Repository and see if that fixes the issue.

 

19 minutes ago, Itsaulgoodman said:

Yes, force transcode on plex, does not utilize hardware transcoding. It worked before upgrading drivers

The webplayer from Plex is known to have issues and when forcing a transcode there will most certainly result in a black screen and playback stopping, please try it with a native client iOS/Android/...

6 minutes ago, ich777 said:

I assume most certainly the tag is wrong.

Please try to remove the tag from the Repository and see if that fixes the issue.

 

Not sure what you mean by remove tag?

 

6 minutes ago, ich777 said:

 

The webplayer from Plex is known to have issues and when forcing a transcode there will most certainly result in a black screen and playback stopping, please try it with a native client iOS/Android/...

Forcing transcode on my android phone is showing HW transcode  

 

Just now, Itsaulgoodman said:

Not sure what you mean by remove tag?

This here is the "tag":

I can't imagine that the maintainer is relaying on such an old version from the driver and is not supporting newer versions from the driver.

 

2 minutes ago, Itsaulgoodman said:

Forcing transcode on my android phone is showing HW transcode  

So Plex is working correct?

Just now, ich777 said:

This here is the "tag":

I can't imagine that the maintainer is relaying on such an old version from the driver and is not supporting newer versions from the driver.

I see, I had to add that tag in order to get a newer version to download. I cant remember what it said there before, but I found a list, and that version was the most recent one that used GPU with CUDA.

 

Just now, ich777 said:

 

So Plex is working correct?

Yes, it is showing plex as hardware transcoding. 

 

Just now, Itsaulgoodman said:

but I found a list

Can you maybe share that list so that I can also take a look at that?

 

However even the documentation says that you should install the latest Nvidia driver, which you have installed:
https://www.codeproject.com/ai/docs/install/install_on_linux.html#nvidia-gpu-support

 

However you could join their Discord and see if they can help you there:
https://discord.com/invite/adxDt9YTX2

 

I'm not sure if you really need that old of a driver for the container to run.

 

BTW, I see that some python process is running on your GPU which is most likely this container: