Hello unRAID Community,

 

I'm experiencing some issues with my GPU (nVidia) on my unRAID server (6.11.5) and could really use your collective wisdom to help me solve this puzzle. I'm trying to start nvidia/cuda:12.1.0-base-ubuntu18.04 and I've tried a few earlier tags too with the same issue.

 

Here's a brief overview of my setup and the issue I'm facing:

• GPU: NVIDIA GTX 780
• I have installed the NVIDIA Unraid plugin and the drivers for my GPU
• nvidia-smi is showing that the GPU is recognized but no running processes are found
• I have tried running a GPU-enabled Docker container, but it fails to start with no useful logs
• I'm getting a repeating error message in my unRAID log stating:

 

Apr 17 21:37:12 Tower kernel: resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000a0000-0x000dffff window]
Apr 17 21:37:12 Tower kernel: caller _nv000720rm+0x1ad/0x200 [nvidia] mapping multiple BARs

 

I've tried the following steps to troubleshoot the issue:

1. Updating my unRAID system and NVIDIA drivers
2. Reseating the GPU
3. Verifying hardware compatibility
4. Checking BIOS settings related to PCI and GPU functionality
5. Searching for similar issues in forums

 

Despite trying these steps, I'm still unable to resolve the issue. I would greatly appreciate any advice, guidance, or GPU whispering techniques you could provide to help me fix this problem.

 

Let me know if there are any other logs or such that would help?

 

Thanks for helping me get my GPU back in the game.

 

 

Edited April 17, 20233 yr by kanzie
updated version after testing new rc

You would probably be better off posting in the [Plugin] Nvidia-Driver support thread. 

 

The GTX780 is a pretty old card, and not supported by the current Production Branch of the Nvidia driver.  You would likely need use the v470 legacy driver. 

Yeah I’m already running the 470-version off the driver and confirm it working using Nvidia-smi.