3 minutes ago, neoherozzz said:

I selected new feature branch by mistake. The open source shows follows:

Please give me about 20 minutes and don't visit the Nvidia Driver plugin page in the meantime.

I'll compile a new package.

@neoherozzz the new package for your card is now available and you should see it hopefully on the plugin page, if not please delete the file: /tmp/nvidia_driver and refresh the plugin page.

 

However I also want to make you aware that some users reported that a Nvidia driver bricked their card, don't know if this was only a Windows only issue or not but I assume most users where on Windows:

https://wccftech.com/nvidia-geforce-rtx-5090-5090d-gpus-getting-bricked-possibly-driver-bios-pcie-issues/

@ich777 Thanks for your flash update!

Everything works perfect so far. Hope the drivers wont brick my 5080🤣

Hello, I recently replaced a P400 (that had been in my system for years) with a P2000 for hardware transcoding -- all else stayed the same.  With the P2000 Plex hangs when I try to transcode something, but Emby works fine.  I've removed and reinstalled the Nvidia plugin, but no help.  Someone suggested that I do a clean install of the Nvidia drivers, but I can't figure out how to do that.  Can anyone point me in the right direction on that?

 

Edit:  Nevermind.  I happened to see a note on the last page that said that Plex is bugged when using the web browser to change the resolution on the fly.  Tried it with an app and works fine.

Edited February 4, 20251 yr by Adam64

Hey there,

i have this error while updating to 7.0.0
 

diagnostics-20250206-1135.zip

The update itself is stuck here:

Any ideas what to do?

 

Thanks in advance!

Edited February 6, 20251 yr by orhaN_utanG

32 minutes ago, orhaN_utanG said:

i have this error while updating to 7.0.0

Please open up a Unraid Terminal and issue this command:

wget -qO- https://raw.githubusercontent.com/ich777/unraid-plugin_update_helper/master/force_package_update | bash

 

This should download the driver for Unraid 7.0.0 otherwise you can also just reboot since the driver plugin will also try to download the driver on boot but do keep in mind boot will take a bit longer than usual because the driver is about 230MB in size.

 

However if that also fails please check in your firewall if IDS/IPS isn't too strict or your AdBlocker is not blocking the GitHub API.

Hello,

I updated plugin, I restarted my server but I think that I see some strange behavior. In GPU statistics my GPU power states are jumping like crazy from P8 to P5 to P0. GPU is doing nothing... no VM, no Docker just idling. I am using also script from SpaceInvader to optimize power.

Do you have an idea what it could be?

1 minute ago, ELP1 said:

I am using also script from SpaceInvader to optimize power.

Please disable the script and only put this line below emhttp in your go script and restart your server:

nvidia-persistenced

 

This will ensure that Nvidia Persistenced is enabled, you don't need to execute it on a schedule or anything else, just put it in the go file and that's enough.

1 hour ago, ich777 said:

Please disable the script and only put this line below emhttp in your go script and restart your server:

nvidia-persistenced

 

This will ensure that Nvidia Persistenced is enabled, you don't need to execute it on a schedule or anything else, just put it in the go file and that's enough.

Thank you, but I read your older comment here: 

, that it should not be used if it is used also with VM. Is it not true anymore? I do not want any crash of the server :-).

And this script is not needed anymore in general because it is handled by Unraid itself?

 

24 minutes ago, ELP1 said:

Is it not true anymore?

This is true, however I think @alturismo has a script for that.

 

However you can still use the script from SIO, are you really sure that nothing is using the GPU?

 

The power state should be always P8 regardless if you use the script from SIO or nvidia-persistenced or of something used the card after the server is booted for the first time like Plex.

 

I usually can‘t give support if you are using the GPU in a VM and on the host since this can always lead to unexpected behavior.

5 minutes ago, ich777 said:

This is true, however I think @alturismo has a script for that.

 

However you can still use the script from SIO, are you really sure that nothing is using the GPU?

 

The power state should be always P8 regardless if you use the script from SIO or nvidia-persistenced or of something used the card after the server is booted for the first time like Plex.

 

I usually can‘t give support if you are using the GPU in a VM and on the host since this can always lead to unexpected behavior.

I have just checked it again and now it looks that it keeps P8 state. I think that I already found the reason why it was happening. Before I used HDMI dummy and I had no behavior like this. But yesterday I connected my GPU directly to TV via HDMI cable. When the TV is turned ON the GPU state is bouncing like I mentioned despite no activity (no VM running), once TV is off it keeps P8 state.

And today I did NVIDIA plugin update so I thought that it is related to update, but apparently not. 

Unraid and RTX 3080 NVIDIA Driver Issues - "NVIDIA-SMI has failed"

 

Hello everyone,

 

I'm running into a frustrating issue with my Unraid server and my RTX 3080 GPU. I've installed the NVIDIA drivers using the community-provided NVIDIA plugin, but when I try to use nvidia-smi, I get the dreaded 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."

 

I've tried reinstalling the drivers, rebooting the server, and checking the plugin configuration, but nothing seems to work. I'm at my wit's end!

 

Has anyone else encountered this problem with an RTX 3080 or similar card on Unraid? I'm particularly interested in hearing from those who have successfully resolved it.

 

Here are some details about my setup:

Unraid Version: 7.0.0

NVIDIA Driver Plugin Version: 570.86.16

RTX 3080: GIGABYTE GeForce RTX™ 3080 GAMING OC 10G

Other relevant hardware: Dual Intel® Xeon® CPU E5-2680 v3 @ 2.50GHz 

Memory: 176 GiB DDR4 Multi-bit ECC

Dell Inc., Version 2.17.0
BIOS dated: Tue 11 Apr 2023 12:00 AM

 

Here are some things I've already tried:

1. Reinstalling the NVIDIA driver plugin.

2. Rebooting the Unraid server.

3. Checking the plugin settings for any obvious errors.

 

Any help would be greatly appreciated! Thanks in advance for your time and expertise.

 

Thanks 

Avnit Bambah

Edited February 10, 20251 yr by abambah2
missing the version

1 hour ago, abambah2 said:

Any help would be greatly appreciated! Thanks in advance for your time and expertise.

 

may upload a diagnostics <<click<< so someone can rather help.

 

most common issues are already described many time in this /yes ... long ...) Thread

Attached diagnostics file 

tower-diagnostics-20250209-2037.zip

1 hour ago, abambah2 said:

Attached diagnostics file 

 

there is no Nvidia GPU found on your system ...

 

1/ check BIOS if its there (above 4g enabled, rbar activated)

2/ check pcie seating (may try a different slot)

3/ check power supply, properly seated, enough, ...

4/ may check card in another system if its functional at all

1 hour ago, abambah2 said:

Attached diagnostics file 

Try the things that @alturismo recommended and please also take one set of Diagnostics when the GPU is visible and one set of Diagnostics after it dropped from your system (without rebooting).

 

If the card was never detected please make sure to plug in all the necessary auxilary power cables and check if the PCIe slot even supports up to 75W of Power delivery.

 

Please note since you've got a Dell system that Dell doesn't allow all addon cards to be recognized in such a system.

大佬，出现这个问题时啥原因，装不上插件。

16 minutes ago, baoxinglove said:

大佬，出现这个问题时啥原因，装不上插件。

In this part of the forums only English please.

 

Do you have a proxy installed? I think it is necessary to communicate with GitHub in your region of the world.

 

The driver plugin uses GitHub API calls to get the latest driver version, maybe there are no free API calls are left on your Proxy, you can check how many GitHub API calls you have left by opening a Terminal and issue this command:

curl -L https://api.github.com/rate_limit | jq -r '.rate'

you should see a response like this:

{
  "limit": 60,
  "remaining": 56,
  "reset": 1739174710,
  "used": 4,
  "resource": "core"
}

in the above example you see that I have 56 calls left.

Here is what I am getting when I try to update the plugin:

 

plugin: updating: nvidia-driver.plg Executing hook script: pre_plugin_checks plugin: run failed: 'upgradepkg --install-new' returned 127 Executing hook script: post_plugin_checks

5 minutes ago, YsarKain said:

Here is what I am getting when I try to update the plugin:

 

plugin: updating: nvidia-driver.plg Executing hook script: pre_plugin_checks plugin: run failed: 'upgradepkg --install-new' returned 127 Executing hook script: post_plugin_checks

Please post the full log output including your Diagnostics.

Feb 10 08:49:27 Thor root: plugin: checking: /boot/config/plugins/nvidia-driver/nvidia-driver-2025.02.10.txz - MD5
Feb 10 08:49:27 Thor root: plugin: skipping: /boot/config/plugins/nvidia-driver/nvidia-driver-2025.02.10.txz already exists
Feb 10 08:49:27 Thor root: plugin: running: upgradepkg --install-new /boot/config/plugins/nvidia-driver/nvidia-driver-2025.02.10.txz

 

That's all that shows up in syslog, the previous output was from the upgrade window.

 

 

thor-diagnostics-20250210-0851.zip

5 minutes ago, YsarKain said:

That's all that shows up in syslog, the previous output was from the upgrade window.

I really can't help, maybe try to upgrade to Unraid 7.0.0 and see if that fixes the issue.

 

Over here a reinstallation from the plugin works fine.

OK in the middle of parity check now but I planned on updating once it finished.

For some reason it does not see my Quadro 2000. I get:
 

Quote

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

 

The trouble shooting info at the beginning of this thread says;

 

Quote

If you bound accidentally all cards to VFIO unbind the card you want to use for the Docker container(s) and reboot the server (TOOLS -> System devices -> unselect the card -> BIND SELECTED TO VFIO AT BOOT -> restart your server).

 

 

Do I need the Legacy driver (470xx)? If so, how do you install those .run files?

 

2 hours ago, expo said:

Do I need the Legacy driver (470xx)? If so, how do you install those .run files?

Please post your Diagnostics.

 

For what do you plan to use the card?