CAB420_docker

This is a docker setup for CAB420, This will not be maintained passed end of Semester 1, 2022.

Use this setup at your own risk, I'm neither a emulation or security expert.

Requirements

Currently this is running on a Ubuntu 20.04.4 LTS system with:

• Nvidia Driver 510,
• Docker version 20.10.12 (build e91ed57),
• nvidia-docker2/bionic, now 2.9.1-1 all.

Setup

Install Ubuntu.

Install Required Nvidia Driver

Install Docker

Install Nvidia-docker2

Install git

sudo apt install git

Clone Repo in the folder you want it to live.

git clone https://github.com/kj87au/CAB420_docker.git

Using it

Place all your work(file and folders) in the folder where the launch.sh file lives.

Launch using:

bash launch.sh

Once the jupyter notebook is runnning open the links provided.

Problems or Errors

It may spew some errors relating to GPU placement, would look like this...

2022-03-02 02:34:01.159200: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:939] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero

This can be ignored (at least from what I can tell)