Is there any page to describe the dependency of nvidia-docker2?

It concludes to the following questions I think:

  1. Is it necessary to install nvidia-driver in linux system before start a nvidia-docker2 using tensorflow-gpu?
  2. Is it necessary to install cuda in linux system before start a nvidia-docker2 using tensorflow-gpu?
  3. Is it necessary to install tensorrt in linux system before start a nvidia-docker2 using tensorflow-gpu with tensorRT support?
    If there is any similar topics, please notify me. Thank you.

I too have the same set of questions. Pls help us

Hello!

Is there any page to describe the dependency of nvidia-docker2?

Yep, this is fairly well documented on the Readme of the GitHub project: https://github.com/NVIDIA/nvidia-docker/blob/master/README.md

To install nvidia-docker 2 you’ll need to have the NVIDIA driver, cuda and docker installed on your system.

Which maps in your questions to 1 and 2 being required but 3 isn’t.

Thanks!