How To Use Durham TITAN Server

From AutonomousPrime Senior Design
Jump to: navigation, search

The durham server contains a Titan X which is the most powerful GPU we have. Multiple users can be logged in at once however it will slow the performance of the GPU.

1. ssh into the machine

ssh -t /bin/bash
  • Make sure you are on the iastate VPN if not on campus
  • Replace net-id with your id

2. Install keras/tensorflow/opencv/flash locally run,

pip3 install --user flask imageio opencv-python==3.1.0 keras==1.2.2 tensorflow-gpu==1.0.1
  • Only need to install once.

3. Install tmux

wget -O - | bash -e
  • Only need to install once.

4. Set up environment for CUDA. Add the `.bashrc` file(Found in this directory) to your linux home directory

  • Only need to run once

5. Run your python code!!

6. exit the ssh session `exit`

Return to Main Page