Help Center/ Elastic Cloud Server/ Troubleshooting/ GPU Driver Issues/ Why Is the T4 GPU Display Abnormal?
Updated on 2024-08-15 GMT+08:00
View PDF
Share

Why Is the T4 GPU Display Abnormal?

Symptom

An ECS, for example, of the PI2 or G6 flavor, is using NVIDIA Tesla T4 GPUs. However, when you run nvidia-smi to check the GPU usage, the following information is displayed: 
No devices were found

Possible Causes

NVIDIA Tesla T4 GPU is a new version of NVIDIA. By default, the GSP firmware is enabled and used. As a result, the GPU cannot be identified.

Method 1

The following settings will become invalid after the ECS is restarted.

  1. Remove the NVIDIA kernel module.

    rmmod nvidia_drm

    rmmod nvidia_modeset

    rmmod nvidia

  2. Disable GSP firmware and load the NVIDIA kernel module.

    modprobe nvidia NVreg_EnableGpuFirmware=0

    modprobe nvidia_drm

    modprobe nvidia_modeset

  3. If the fault persists, contact customer service.

Method 2

  1. Run the following command to open the /etc/modprobe.d/nvidia.conf file:

    vim /etc/modprobe.d/nvidia.conf

    Press i to enter the editing mode.

  2. Add the following information to /etc/modprobe.d/nvidia.conf: 
    options nvidia NVreg_EnableGpuFirmware=0

    Press Esc, enter :wq!, and exit.

  3. Run the following command to restart the ECS:

    reboot

  4. If the fault persists, contact customer service.
Parent topic: GPU Driver Issues