Nvidia-smi command shows no device found after cuda toolkit installation (w/ nvidia driver)

Graphics / Linux Linux
#1

Hello.

I installed cudatoolkit 11.3.1 with 465.19.01 driver on a system with “CentOS Linux release 8.4.2105” using “cuda_11.3.1_465.19.01_linux.run” file.

GeForce 3080 ti is mounted to my system.
However, nvidi-smi shows “No devices were found” after installation of the cuda toolkit and included nvidia driver.

Before installation, nvidia-smi worked fine showing driver version 470.
I don’t know how the previous driver were installed (by run file, package or else.)

I ran the nvidia-bug-report.sh file but I don’t know I to attach files…
So, instead, here is the result of the command ‘dmesg | grep NVRM’.
“------------------------------------------------------------------------------------------------”
NVRM: loading NVIDIA UNIX x86_64 Kernel Module 465.19.01 Fri Mar 19 07:44:41 UTC 2021
NVRM: GPU 0000:af:00.0: RmInitAdapter failed! (0x26:0xffff:1242)
NVRM: GPU 0000:af:00.0: rm_init_adapter failed, device minor number 0
NVRM: GPU 0000:af:00.0: RmInitAdapter failed! (0x26:0x65:1242)
NVRM: GPU 0000:af:00.0: rm_init_adapter failed, device minor number 0
NVRM: GPU 0000:af:00.0: GPU has fallen off the bus.
NVRM: GPU 0000:af:00.0: RmInitAdapter failed! (0x26:0xb:1242)
NVRM: GPU 0000:af:00.0: rm_init_adapter failed, device minor number 0
NVRM: GPU 0000:af:00.0: RmInitAdapter failed! (0x23:0x56:643)
NVRM: GPU 0000:af:00.0: rm_init_adapter failed, device minor number 0
“------------------------------------------------------------------------------------------------”
(The last two line repeated for numerous times)

I found some suggestions for Ubuntu, but I couldn’t find a working solution for Centos.

Thank you in advance!

#2

the 465 driver is too old (03/21), support for the 3080ti was only added in 06/21.
The runfile installer asks you if you want to install the driver as well, if you already have one installed, skip it.
Please download a recent driver runfile from https://http.download.nvidia.com/XFree86/Linux-x86_64/
and run it with --uninstall option. Then try reconstruction the original driver, possibly from the rh nvidia repo.

1 Like
#3

Hello there, have same problem, my linux distro doesnt have apt-get, nor yum or anything else, can you please explain for dummies this part
“Index of /XFree86/Linux-x86_64
and run it with --uninstall option. Then try reconstruction the original driver, possibly from the rh nvidia repo.”

I downloaded new drivers for linux, uploaded them via SSH, what should I do next? Thank you in advance.

#4

Please open a new thread describing your issue.
Please run nvidia-bug-report.sh as root and attach the resulting nvidia-bug-report.log.gz file to your post.