Skip to content

Latest commit

 

History

History
83 lines (57 loc) · 2.62 KB

azure.md

File metadata and controls

83 lines (57 loc) · 2.62 KB

Microsoft Azure Sponsorship

We would like to thank Microsoft Azure Sponsorship 2 for their generous donation of GPU computational resources for this course, as well as, Michael Thomas, Andrew Webber and Lee Stott for their precious help.

Specifications

Each student is given a VM instance with the following specifications:

  • Name: Standard NC6 (East US)
  • CPU: 6 cores
  • RAM: 56 GB
  • GPU: NVIDIA Tesla K80

Connect to your server

By this point you should have received an IP and a password in your college email. Using the ip and password provided you can connect to your server using ssh deepnlp2017@<your server ip>

Session managment

tmux allows access multiple separate terminal sessions inside a remote terminal session. Using tmux you can keep sessions alive in the background (e.g. training your network) even when disconnecting, and reconnect to them the next time you login. Basic shortcuts:

  • list sessions tmux ls

  • detach from session ctrl+b - d

  • change name of session ctrl+b - shift+4

  • reconnect to session tmux attach -t <session_name>

Install Cuda 8.0

sudo apt-get install build-essential
cd ~
wget https://developer.nvidia.com/compute/cuda/8.0/Prod2/local_installers/cuda_8.0.61_375.26_linux-run
chmod +x cuda_8.0.61_375.26_linux-run
sudo ./cuda_8.0.61_375.26_linux-run

Answer the questions with y(es) or enter (default settings) for paths. To confirm successful installation run nvidia-smi.

Install CuDNN 5.1

cd ~
wget http://developer.download.nvidia.com/compute/redist/cudnn/v5.1/cudnn-8.0-linux-x64-v5.1.tgz
sudo tar -xzf cudnn-8.0-linux-x64-v5.1.tgz -C /usr/local
rm cudnn-8.0-linux-x64-v5.1.tgz
sudo ldconfig

Add in references

append the following lines to your ~/.bashrc. You can edit it with nano ~/.bashrc.

export CUDA_HOME=/usr/local/cuda-8.0
export PATH=/usr/local/cuda-8.0/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

Install framework

You are ready to install python and your favourite framework:

  • PyTorch
  • Tensorflow
  • CNTK

Instructions of how to set up an iPython notebook server remotely can be found here.

Troubleshooting

  • ensure uninstallation of previous NVIDIA drivers sudo apt-get remove --purge nvidia-*

Support

For further information for Univerisity of Oxford students please contact [email protected].