Hello NVIDIA team,
I am trying to install and use NVIDIA drivers on a system with an NVIDIA H100 (GH100 PCIe) GPU, but I am unable to get nvidia-smi working.
System details
-
GPU: NVIDIA H100 (GH100 PCIe)
-
OS: Ubuntu 25.x (plucky)
-
Kernel: 6.14.0-37-generic
-
Use case: Basic GPU validation (
nvidia-smi), no containers involved at this stage
What I have tried
-
Installed multiple NVIDIA drivers via APT, including:
-
nvidia-driver-535-server -
nvidia-driver-550 -
nvidia-driver-570-server -
nvidia-driver-580
-
-
All drivers install successfully via
apt
Observed behavior
After installing any of the above drivers, nvidia-smi consistently fails with:
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver.
Additionally:
lsmod | grep nvidia
shows no NVIDIA kernel modules loaded.