How do I use webodm with gpu

  1. Apart from nvidia driver, which of the following packages do I need to install:
    a. Nvidia docker toolkit
    b. Cuda toolkit
    c. Nvidia container runtime
    d. Cuda driver

  2. Anytime I run ./webodm.sh start --gpu, only 3 of the 5 containers in the webodm application run automatically in docker desktop. The 2 that don’t run include GPU container. I have to always run the GPU image as a separate container. I would to know if this is normal.

  3. How do I really process images with --gpu. Do I process in webodm using the default node with as many containers running under webodm application in docker desktop OR use a created node in webodm node based on the gpu image in docker desktop.

OS: Ubuntu 22.04LTS

Thank you for the attention

2 Likes