Hello, i've searched many forums but didn't quite manage to find the problem i have. I built my first HomeLab recently and have the following main specs:
- AsRock B650D4U-2L2T/BCM
- AMD EPYC 4345P
- Nvidia RTX 5060

- Unraid 7.2.2

Ever since setting up my Server around a week ago i couldn't detect my GPU fully.

The device doesn't show up in Plex when i try to choose a device for transcoding.

lspci shows my GPU:
01:00.0 VGA compatible controller: NVIDIA Corporation GB206 [GeForce RTX 5060]

01:00.1 Audio device: NVIDIA Corporation GB206 High Definition Audio Controller

But nvidia-smi doesn't find the device:
root@Tower:~# nvidia-smi

No devices were found

root@Tower:~#

GPU Statistics shows the GPU:

But in the Dashboard, it only says "PCI Bus Utilization":


Nvidia Driver does have the Open Source version that is required for 50XX cards, but the device is not found:

I've also not bound anything in system devices, no checkmarks for:

IOMMU group 13:				[10de:2d05] 01:00.0 VGA compatible controller: NVIDIA Corporation GB206 [GeForce RTX 5060] (rev a1)
				[10de:22eb] 01:00.1 Audio device: NVIDIA Corporation GB206 High Definition Audio Controller (rev a1)

Another output from the terminal that clearly detects my gpu:
root@Tower:~# lspci -ks 0000:01:00.0

01:00.0 VGA compatible controller: NVIDIA Corporation GB206 [GeForce RTX 5060] (rev a1)

Subsystem: CardExpert Technology Device f331

Kernel driver in use: nvidia

Kernel modules: nouveau, nvidia_drm, nvidia

root@Tower:~#

The only logical reason for me is that the driver doesn't support my gpu, but in the post it clearly says i need the open source, meaning the v580.95.05 version which i have installed. I've run the latest version before, but it gave out the same results as now with the open source version..

Are there any other reasons my gpu gets detected in some ways but doesn't really work for Unraid?

6 hours ago, ElementG8 said:

i've searched many forums but didn't quite manage to find the problem i have.

Please post your Diagnostics, without them I can't say much.

Hi everyone,

Could someone briefly confirm whether I can run full CUDA acceleration with Ollama using an RTX 5060 Ti on Unraid 7.2.2 with the latest open source NVIDIA driver?

Thanks in advance!

2 hours ago, waterfallbay said:

I can run full CUDA acceleration

What is the opposite from full CUDA acceleration?

4 hours ago, ich777 said:

Please post your Diagnostics, without them I can't say much.

Here are my Diagnostics, hope that's correct:
Diagnotics Unraid

46 minutes ago, ElementG8 said:

hope that's correct

You can drag them just into the textbox within the forums.

First of all, your plugin is broken, you can see that on these lines:

Please remove the plugin, reboot, install the plugin again, select the open source driver, click Download & Update and reboot again.

From what I see from your syslog:

Dec 14 02:27:30 Tower kernel: NVRM: nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from memdescAllocInternal(pMemDesc) @ memdesc.c:1359

Dec 14 02:27:30 Tower kernel: NVRM: _kgspBootGspRm: unexpected WPR2 already up, cannot proceed with booting GSP

Dec 14 02:27:30 Tower kernel: NVRM: _kgspBootGspRm: (the GPU is likely in a bad state and may need to be reset)

Dec 14 02:27:30 Tower kernel: NVRM: RmInitAdapter: Cannot initialize GSP firmware RM

Dec 14 02:27:30 Tower kernel: NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x62:0x40:2015)

Dec 14 02:27:30 Tower kernel: NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0

Dec 14 02:27:32 Tower kernel: NVRM: nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from memdescAllocInternal(pMemDesc) @ memdesc.c:1359

You are suffering from the same issues as others do, this is most of the times caused by hardware incompatbility or a wrong set setting in the BIOS.

I would recommend that you do the following:

1. Make sure your BIOS is up to date

2. Load default BIOS settings

3. Add nvidia-persistence at the end of your go file (/boot/config/go)

4. Make sure to enable Above 4G Decoding in the BIOS

5. Make sure to enable Resizabel BAR Support in the BIOS

6. If you have the setting in your BIOS, enable BME DMA Mitigation (see this post)

I did all that, including the nvidia-persistance in the go file, my Bios is up to date and i don't have a BME DMA Mitigation option
The red marked error on the plugin page is gone now, but the gpu still is not detected.

Can't drag it in here, don't know why, sorry:
https://www.swisstransfer.com/d/9ef0ffb6-34a9-4561-92e8-a5538eb2493e

Then i guess my GPU won't work with that MB, ffs... How do i find out or check what GPU works with my MB, there wasn't anything on the page...

From the Terminal - GSP Firmware Errors

root@Tower:~# cat /proc/cmdline BOOT_IMAGE=/bzimage initrd=/bzroot root@Tower:~# nvidia-smi No devices were found root@Tower:~# dmesg -T | egrep -i "gsp|NVRM|Xid" | tail -n 100 [Sun Dec 14 15:33:30 2025] NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64 580.95.05 Release Build (dvs-builder@U22-I3-B17-02-5) Tue Sep 23 09:55:41 UTC 2025 [Sun Dec 14 15:34:55 2025] NVRM: kgspLogXid119: ********************************* GSP Timeout ********************************** [Sun Dec 14 15:34:55 2025] NVRM: kgspLogXid119: Note: Please also check logs above. [Sun Dec 14 15:34:55 2025] NVRM: nvAssertFailedNoLog: Assertion failed: expectedFunc == pHistoryEntry->function @ kernel_gsp.c:2119 [Sun Dec 14 15:34:55 2025] NVRM: GPU at PCI:0000:01:00: GPU-977a01c6-93b1-cd5f-8a7a-9e1d2bb797c3 [Sun Dec 14 15:34:55 2025] NVRM: Xid (PCI:0000:01:00): 119, Timeout after 6s of waiting for RPC response from GPU0 GSP! Expected function 4097 (GSP_INIT_DONE) sequence 0 (0x0 0x0). [Sun Dec 14 15:34:55 2025] NVRM: GPU0 GSP RPC buffer contains function 4124 (GSP_LOCKDOWN_NOTICE) sequence 0 and data 0x0000000000000000 0x0000000000000000. [Sun Dec 14 15:34:55 2025] NVRM: GPU0 RPC history (CPU -> GSP): [Sun Dec 14 15:34:55 2025] NVRM: entry function sequence data0 data1 ts_start ts_end duration actively_polling [Sun Dec 14 15:34:55 2025] NVRM: 0 73 SET_REGISTRY 0 0x0000000000000000 0x0000000000000000 0x000645ea65990143 0x0000000000000000 y [Sun Dec 14 15:34:55 2025] NVRM: -1 72 GSP_SET_SYSTEM_INFO 0 0x0000000000000000 0x0000000000000000 0x000645ea65990140 0x0000000000000000 [Sun Dec 14 15:34:55 2025] NVRM: GPU0 RPC event history (CPU <- GSP): [Sun Dec 14 15:34:55 2025] NVRM: entry function sequence data0 data1 ts_start ts_end duration during_incomplete_rpc [Sun Dec 14 15:34:55 2025] NVRM: 0 4124 GSP_LOCKDOWN_NOTICE 0 0x0000000000000000 0x0000000000000000 0x000645ea65e6a050 0x000645ea65e6a050 y [Sun Dec 14 15:34:55 2025] NVRM: -1 4124 GSP_LOCKDOWN_NOTICE 0 0x0000000000000001 0x0000000000000000 0x000645ea65e69edf 0x000645ea65e69edf y [Sun Dec 14 15:34:55 2025] NVRM: -2 4124 GSP_LOCKDOWN_NOTICE 0 0x0000000000000000 0x0000000000000000 0x000645ea65e568af 0x000645ea65e568af y [Sun Dec 14 15:34:55 2025] NVRM: -3 4124 GSP_LOCKDOWN_NOTICE 0 0x0000000000000001 0x0000000000000000 0x000645ea65e53d53 0x000645ea65e53d54 1us y [Sun Dec 14 15:34:55 2025] NVRM: -4 4108 UCODE_LIBOS_PRINT 0 0x0000000000000000 0x0000000000000000 0x000645ea65a47571 0x000645ea65a47571 y [Sun Dec 14 15:34:55 2025] NVRM: -5 4108 UCODE_LIBOS_PRINT 0 0x0000000000000000 0x0000000000000000 0x000645ea65a47567 0x000645ea65a47568 1us y [Sun Dec 14 15:34:55 2025] NVRM: -6 4124 GSP_LOCKDOWN_NOTICE 0 0x0000000000000000 0x0000000000000000 0x000645ea65a467fe 0x000645ea65a467fe y [Sun Dec 14 15:34:55 2025] NVRM: -7 4124 GSP_LOCKDOWN_NOTICE 0 0x0000000000000001 0x0000000000000000 0x000645ea65a43901 0x000645ea65a43902 1us y [Sun Dec 14 15:34:55 2025] kgspRpcRecvPoll+0x650/0x820 [nvidia] [Sun Dec 14 15:34:55 2025] ? kgspWaitForRmInitDoneIMPL+0x36/0xf8 [nvidia] [Sun Dec 14 15:34:55 2025] ? kgspBootstrap_GH100+0x385/0xaca [nvidia] [Sun Dec 14 15:34:55 2025] ? kgspInitRm_IMPL+0x96a/0x1680 [nvidia] [Sun Dec 14 15:34:55 2025] NVRM: kgspLogXid119: ******************************************************************************** [Sun Dec 14 15:34:55 2025] NVRM: nvCheckOkFailedNoLog: Check failed: Call timed out [NVERR_TIMEOUT] (0x00000065) returned from rpcRecvPoll(pGpu, pRpc, NV_VGPU_MSG_EVENT_GSP_INIT_DONE, 0) @ kernel_gsp.c:4992 [Sun Dec 14 15:34:55 2025] NVRM: nvAssertOkFailedNoLog: Assertion failed: Call timed out [NV_ERR_TIMEOUT] (0x00000065) returned from kgspWaitForRmInitDone(pGpu, pKernelGsp) @ kernel_gsp_gh100.c:980 [Sun Dec 14 15:34:55 2025] NVRM: kgspBootGspRm: unexpected WPR2 already up, cannot proceed with booting GSP [Sun Dec 14 15:34:55 2025] NVRM: kgspBootGspRm: (the GPU is likely in a bad state and may need to be reset) [Sun Dec 14 15:34:55 2025] NVRM: RmInitAdapter: Cannot initialize GSP firmware RM [Sun Dec 14 15:34:55 2025] NVRM: iovaspaceDestruct_IMPL: 1 left-over mappings in IOVAS 0x100 [Sun Dec 14 15:34:55 2025] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x62:0x40:2015) [Sun Dec 14 15:34:55 2025] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0 [Sun Dec 14 15:34:59 2025] NVRM: kgspBootGspRm: unexpected WPR2 already up, cannot proceed with booting GSP [Sun Dec 14 15:34:59 2025] NVRM: kgspBootGspRm: (the GPU is likely in a bad state and may need to be reset) [Sun Dec 14 15:34:59 2025] NVRM: RmInitAdapter: Cannot initialize GSP firmware RM [Sun Dec 14 15:34:59 2025] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x62:0x40:2015) [Sun Dec 14 15:34:59 2025] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0 [Sun Dec 14 15:35:15 2025] NVRM: kgspBootGspRm: unexpected WPR2 already up, cannot proceed with booting GSP [Sun Dec 14 15:35:15 2025] NVRM: kgspBootGspRm: (the GPU is likely in a bad state and may need to be reset) [Sun Dec 14 15:35:15 2025] NVRM: RmInitAdapter: Cannot initialize GSP firmware RM [Sun Dec 14 15:35:15 2025] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x62:0x40:2015) [Sun Dec 14 15:35:15 2025] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0 root@Tower:~#

6 hours ago, ElementG8 said:

Then i guess my GPU won't work with that MB, ffs... How do i find out or check what GPU works with my MB, there wasn't anything on the page...

You'll also see that in the syslog, no dmesg needed, you have a Xid 119 error, please see here.

I'm not sure, that could be also be a driver issue. Have you yet tried to reseat the card, please also check your Power Supply if it's up to the task, don't forget that power supplies also degrade over time.

Maybe also try to set the PCIe Generation to something like Gen 4 or Gen 3 and see if that helps.

I upgraded to Unraid 7.2.2 from 7.2.1 and all my dockers which use the NVIDIA driver (plex, jellyfin and Tdarr) have stopped working.

Checked the nVidia driver config page and it states:

"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'm at the latest version: v590.44.01

Not really sure how to fix this.

tower-diagnostics-20251216-0802.zip

Edited December 15, 2025Dec 15 by bobalot

I just also ran into this issue as well (have had the nvidia plugin driver installed and working fine, plex working, etc). Came to find Plex stopped and refused to start. Even tried a remove and reinstall and discovered nvidia driver was the issue.

In the plugin I was showing the driver installed and on the latest version. However it also stated nvidia-smi failed error. Further research shows that the latest driver has dropped support for several older cards including my Quadro P2000 it seems.

After changing to the production driver v580.119.02, allowing it to download/install and a reboot, it now shows my Quadro P2000 and Plex is running once again.

17 minutes ago, The Alien said:

I just also ran into this issue as well (have had the nvidia plugin driver installed and working fine, plex working, etc). Came to find Plex stopped and refused to start. Even tried a remove and reinstall and discovered nvidia driver was the issue.

In the plugin I was showing the driver installed and on the latest version. However it also stated nvidia-smi failed error. Further research shows that the latest driver has dropped support for several older cards including my Quadro P2000 it seems.

After changing to the production driver v580.119.02, allowing it to download/install and a reboot, it now shows my Quadro P2000 and Plex is running once again.

Any idea on how to downgrade ???

Nvidia Info:

Nvidia Driver Version: 580.119.02

Open Source Kernel Module: No

Installed GPU(s):
Failed to initialize NVML: Driver/library version mismatch
NVML library version: 590.44

I dont know why i keep getting this error when using the last version

42 minutes ago, The Alien said:

I just also ran into this issue as well (have had the nvidia plugin driver installed and working fine, plex working, etc). Came to find Plex stopped and refused to start. Even tried a remove and reinstall and discovered nvidia driver was the issue.

In the plugin I was showing the driver installed and on the latest version. However it also stated nvidia-smi failed error. Further research shows that the latest driver has dropped support for several older cards including my Quadro P2000 it seems.

After changing to the production driver v580.119.02, allowing it to download/install and a reboot, it now shows my Quadro P2000 and Plex is running once again.

That worked! I also had a Quadro P2000.

@shremi, goto settings->NVIDIA driver->select production branch

My issue is that i cannot seem to downgrade .

Whenever the plugin is installed it gets the 590 version , and if i manually select the 580 production version i get the :

Installed GPU(s):
Failed to initialize NVML: Driver/library version mismatch
NVML library version: 590.44

So i am in a loop cannot do anything

7 hours ago, bobalot said:

Not really sure how to fix this.

Please read this post:

1 hour ago, The Alien said:

Further research shows that the latest driver has dropped support for several older cards including my Quadro P2000 it seems.

Please also see this post:

1 hour ago, shremi said:

I dont know why i keep getting this error when using the last version

Please always post your Diagnostics when posting hardware related questions, please also see this post:

21 minutes ago, shremi said:

Whenever the plugin is installed it gets the 590 version , and if i manually select the 580 production version i get the

Did you also reboot after selecting driver version 580?

Again, Diagnostics would be very helpful.

1 hour ago, ich777 said:

Uploading Attachment...Did you also reboot after selecting driver version 580?

Again, Diagnostocs would be very helpful.

Hi there sure , it seems that i am stuck in some weird loop where my system does not want to install another driver , yes i have tried multiple times rebooting after installing the 580 version , so far i managed to get it to work using the opensource version , that one it did wanted to take however once with that one installed it simply does not want to go to the 580 version.

In the meantime i have managed to instal an RTX 2XX card i had as my pc backup/spare but will be super usefull to know why is my system behaving like this .

So far if i can revert to the open source package i think i can make it work until i find a solution

Uploading Attachment...

tower-diagnostics-20251216-0004.zip

Edited December 16, 2025Dec 16 by shremi

24 minutes ago, shremi said:

it seems that i am stuck in some weird loop where my system does not want to install another driver

Have you seen this recommended post yet:
https://forums.unraid.net/topic/98978-plugin-nvidia-driver/page/178/#findComment-1487747

Please also look at the post above, if you have the same two PHP errors like in the screenshot follow the steps from the linked post.

On 11/18/2024 at 9:44 PM, ich777 said:

The plugin is broken, you‘ll see that on the two php warnings on the page.

 

Please do the following:

1. Uninstall the Nvidia driver plugin,

2. Reboot

3. Pull a fresh copy from the CA App

4. Go to the Nvidida Plugins website and slelct the driver that you want to use

5. Reboot again

Thanks , yes this is the process i am doing and it is not working

4 minutes ago, shremi said:

Thanks , yes this is the process i am doing and it is not working

And what is the popup saying when you change the driver version?

Any output? I can't see your screenshot.

What about Nvidia Driver version: 580.119.02 or have you yet tried the version below that version?
I'll trigger a recompilation from that driver.

42 minutes ago, ich777 said:

And what is the popup saying when you change the driver version?

Any output? I can't see your screenshot.

What about Nvidia Driver version: 580.119.02 or have you yet tried the version below that version?
I'll trigger a recompilation from that driver.

So far it does not work with :

v580.119.02

v580.105.08

I am going to reboot and check if the open source driver is working for me other than the 590.44.01 those 2 are the ones who stick

52 minutes ago, shremi said:

I am going to reboot and check if the open source driver is working for me other than the 590.44.01 those 2 are the ones who stick

Do you have additional diagnostics from a 580 driver?

5 hours ago, The Alien said:

I just also ran into this issue as well (have had the nvidia plugin driver installed and working fine, plex working, etc). Came to find Plex stopped and refused to start. Even tried a remove and reinstall and discovered nvidia driver was the issue.

In the plugin I was showing the driver installed and on the latest version. However it also stated nvidia-smi failed error. Further research shows that the latest driver has dropped support for several older cards including my Quadro P2000 it seems.

After changing to the production driver v580.119.02, allowing it to download/install and a reboot, it now shows my Quadro P2000 and Plex is running once again.

Thank you for this! This has fixed my issue as well, shame that nvidia have dropped support.

2 hours ago, showstopper said:

Thank you for this! This has fixed my issue as well, shame that nvidia have dropped support.

Please read this post:

22 hours ago, ich777 said:

Do you have additional diagnostics from a 580 driver?

tower-diagnostics-20251216-0004.zip
I captured this yesterday does this work ?