Nvidia-smi No devices were found

Accelerated Computing CUDA CUDA Setup and Installation
1

Hi
I have a problem with nvidia-smi, it tells me No devices were found.
I have a openSUSE leap 15.6

imagen
imagen1281×693 121 KB

Here is the result of the command “lsmod | grep nvidia”

nvidia_drm            135168  0
nvidia_modeset       1703936  1 nvidia_drm
nvidia_uvm           3997696  0
nvidia              11628544  2 nvidia_uvm,nvidia_modeset
video                  77824  3 asus_wmi,amdgpu,nvidia_modeset

So, instead, here is the result of the command ‘dmesg | grep NVRM’.

[    7.613603] NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64  570.169  Release Build  (abuild@OBS)  Sun Jun 15 12:00:00 UTC 2025
[    8.018172] NVRM: RmFetchGspRmImages: No firmware image found
[    8.018177] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[    8.018202] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[    8.974259] NVRM: RmFetchGspRmImages: No firmware image found
[    8.974265] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[    8.974291] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[    8.974513] NVRM: RmFetchGspRmImages: No firmware image found
[    8.974516] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[    8.974533] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[    8.976824] NVRM: RmFetchGspRmImages: No firmware image found
[    8.976831] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[    8.976854] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[    8.977126] NVRM: RmFetchGspRmImages: No firmware image found
[    8.977131] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[    8.977151] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[103314.988281] NVRM: RmFetchGspRmImages: No firmware image found
[103314.988293] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[103314.988370] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[110196.634044] NVRM: RmFetchGspRmImages: No firmware image found
[110196.634053] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[110196.634084] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[110367.277669] NVRM: RmFetchGspRmImages: No firmware image found
[110367.277674] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[110367.277699] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[110990.237950] NVRM: RmFetchGspRmImages: No firmware image found
[110990.237964] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[110990.238039] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[111553.537799] NVRM: RmFetchGspRmImages: No firmware image found
[111553.537805] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:1770)
[111553.537830] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0

nvidia-bug-report.log.gz (509.1 KB)

2

Just a thought, but I didnt think a 1650 would be supported by the open driver that I think you are using:

loading NVIDIA UNIX Open Kernel Module for x86_64  570.169

I think you will need the closed driver

zypper -v install cuda-drivers

have a look at:

3

It just squeaks in, being a Turing device.

4

I followed the steps of 1. Introduction — NVIDIA Driver Installation Guide r575 documentation, but something happened, because now nvidia-smi says the following

NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.
5

Sorry. as rs277 says you could use open or closed driver on Turing.
I believe what you are seeing now is that the driver is not loading. In some ways that can be simpler to deal with than loading and then misbehaving.
what is in:
nvidia-installer.log
(on my system thats in /var/log and should be dated for when you ran the install)

6

Here is the log file
nvidia-installer.log (49.2 KB)
I think this is where the error is since it says that:

creation time: Thu Jan 16 23:50:53 2025
installer version: 550.144.03

What’s the best thing I can do now?

7

Oh dear, Nvidia have certainly made this hard work. Sorry for the slow turnaround.
What the log file shows is that on January 16 the “runfile method” was used and the driver installed successfully.
Your most recent efforts we with zypper and that creates a different logfile and on rereading the driver install doc the runfile method is not suggested any more.
I find the simplest way to get cuda working is to:
install the driver
get it working
install cuda-toolkit.see

So I think the simplest way forward is to uninstall your existing driver with the method appropriate to how you installed it(see the doc I pointed you to). Its not ideal (if you have installed cuda it will want to remove it, for example) but lets get the driver working.
Currently you seem to have a 570 driver installed.

Its possible that there are remnants of the 550 driver left. So if you have the runfile you used then

sudo ./NVIDIA-Linux-x86-550.144.03.run --uninstall

or whatever its called. Dont worry if it errors or you have not got it anymore, for now.
Once that is done follow the suse instructions in the document and copy what it says on the screen as it trys to install.