How do you set up Ubuntu 16.04 rigs for deep bio-molecular reinforcement learning?

A battle-tested guide to install Nvidia drivers, CUDA, cuDNN, conda, Tensorflow-GPU, PyMol, and OpenAI Gym on Linux!

Bion Howard
bitpharma.com blog
3 min readNov 17, 2018

--

Thanks for the advice, Elon…

install ubuntu 16.04

get a usb stick and plug it in (needs >2gb space)
download http://releases.ubuntu.com/xenial/ubuntu-16.04.5-desktop-amd64.iso
use startup disk creator to burn the iso to the usb
put the usb into the computer and restart it
press ESC, F1, F2, F8 or F10 to boot from usb
use the gui to install ubuntu

set files explorer to default as list

gsettings set org.gnome.nautilus.preferences default-folder-viewer ‘list-view’

install nvidia drivers

sudo apt-get purge nvidia* (might be unnecessary on fresh install 16.04)
sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt-get update && sudo apt-get install nvidia-396
nvidia-smi

install cuda 9.0
link http://developer.nvidia.com/cuda-downloads
avoid the runfile because it doesn’t work due to X server issues unless you switch to a text-only mode of ubuntu and deactivate stuff

sudo dpkg -i cuda-repo-ubuntu1604–9–0-local_9.0.176–1_amd64.deb
sudo apt-key add /var/cuda-repo-9–0-local/7fa2af80.pub
sudo apt-get update
sudo apt-get install nvidia-cuda-toolkit (*install cuda overwrites nvidia-396*)
install patches by double clicking in files
nvcc — version

install cudnn

guide https://docs.nvidia.com/deeplearning/sdk/cudnn-install/index.html
link https://developer.nvidia.com/rdp/cudnn-download (get 7.0)

sudo dpkg -i libcudnn7*
install with ubuntu installer
cat /usr/include/cudnn.h | grep CUDNN_MAJOR -A 2

install conda and configure the environment
link https://repo.continuum.io/archive/Anaconda3-5.1.0-Linux-x86_64.sh

cd ~/Downloads
bash Anaconda3–5.1.0-Linux-x86_64.sh
follow the prompts
close the terminal (alt-f4)
open a new terminal (ctrl-alt-t)
conda create -n tfg tensorflow-gpu=1.11
source activate tfg

install dependencies

conda install -c conda-forge slackclient
conda install -c schrodinger pymol
conda install -c anaconda pillow
conda install imageio
pip install gym

clone the repo and run neuromax

cd && mkdir hax && cd hax
git clone https://bionicle1337@bitbucket.org/bitpharma1337/neuromax.git
cd neuromax && git checkout kamelbranch
python neuromax.py (edited)

never underestimate what nerds can do in 15 hours of deep work

Copyright 2018 Bion Howard & bitpharma.com team

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

--

--