I am trying to run some processing jobs with a 3060Ti. I was running the processing jobs then when I want back to run another processing job, I noticed the GPU was not operating and stuck at P8. I ran nvidia-detect which said something like “nvidia corporation device 2489 (rev a1) not supported”. I updated it so it now says RTX 3060 Ti but now it says “Uh oh. Your card is not supported by any driver version up to 545.23.08. A newer driver may add support for your card.”
I updated the driver so it is now 550.67 with cuda 12.4. Nvidia-smi shows the driver and cuda version but nvidia-detect still returns the same error “Your card is not supported by any driver version up to…”. I have no idea what to do at this point. Help would be appreciated.
I have attached the nvidia-bug-report output. Thanks
nvidia-bug-report.log (1.8 MB)
Hi there @cbalding and welcome to the NVIDIA developer forums.
To be honest I do not see an issue with your setup. Do you experience any other problems beside nvidia-detect not working? If not then I would rather recommend updating that package and its dependencies in Debian to a newer version if you want it to detect the GPU correctly.
That package is part of the distribution, not part of the GPU driver package.
Thanks for your help. This is where it is strange and I apologize if I didn’t explain it accurately. This is the output for nvidia-smi:
±----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.67 Driver Version: 550.67 CUDA Version: 12.4 |
|-----------------------------------------±-----------------------±---------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce RTX 3060 Ti Off | 00000000:03:00.0 On | N/A |
| 35% 49C P8 25W / 240W | 339MiB / 8192MiB | 1% Default |
| | | N/A |
±----------------------------------------±-----------------------±---------------------+
±----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 2124 G /usr/lib/xorg/Xorg 153MiB |
| 0 N/A N/A 2241 G /usr/bin/gnome-shell 54MiB |
| 0 N/A N/A 3982 G …seed-version=20240408-050137.449000 122MiB |
This is the output for nvidia-detect:
Detected NVIDIA GPUs:
03:00.0 VGA compatible controller [0300]: NVIDIA Corporation GA104 [GeForce RTX 3060 Ti Lite Hash Rate] [10de:2489] (rev a1)
Checking card: NVIDIA Corporation GA104 [GeForce RTX 3060 Ti Lite Hash Rate] (rev a1)
Uh oh. Your card is not supported by any driver version up to 545.23.08.
A newer driver may add support for your card.
Newer driver releases may be available in backports, unstable or experimental.
So when I try to fire up a processing job, the GPU is stuck in P8 doing nothing. So even though nvidia-smi says everything is fine, nvidia-detect is saying it isn’t there. Nvidia-smi even acknowledges the updated driver 550.67 while nvidia-detect says I’m using an outdated version
I am thinking of doing some things like disabling secure-boot, updating the Linux kernel, and some other things but I cannot find the problem. Any thoughts appreciated. Thanks
nvidia-detect is a third-party tool, known buggy. Please forget about it, its output is irrelevant. What kind of “processing jobs” are you trying to run?
I had previously ran some embedding tests and wanted to run some different embedding tests. The GPU was clearly in heavy use based upon power and other metrics. However, after changing the file paths and trying to rerun the code on a different batch of data, it seems the GPU isn’t being used. Nvidia-smi recognizes the drivers and cuda but it seems stuck on P8 only rarely being used at only at low levels. I am really not sure what the problem is because every thing I check does not seem to be a problem