Tensorflow Docker

This repository provides a dockerized tensorflow environment with cuda support for neural network training.

Install

Install docker

https://docs.docker.com/engine/install/ubuntu/

Run post-installation steps

https://docs.docker.com/engine/install/linux-postinstall/

You should be able to run docker run hello-world without using sudo. If sudo is still required, run sudo newgrp docker and restart your computer.

Install Nvidia drivers

Install Nvidia Driver Version: 525.125.06 or higher.

Install Nvidia container toolkit

https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#docker

Compile container

./compile.sh

Run livox_ros container

Run the container: ./run.sh In case of an error like this: docker: Error response from daemon: Unknown runtime specified nvidia.

Run the following commands

sudo apt install -y nvidia-docker2
sudo systemctl daemon-reload
sudo systemctl restart docker

In VSCode install Dev Container and Docker

The repository /data folder is not synced to github, but it is attached to the container at /home/appuser/data, the same goes with src.

Test the docker

Navigate to src folder and run the provided python script. python test-gpu.py The result should look similar to this: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /device:GPU:0 with 6439 MB memory: -> device: 0, name: NVIDIA GeForce GTX 1080, pci bus id: 0000:02:00.0, compute capability: 6.1 True

Let's go!