Amazon Web Services (AWS) offer easy-to-use and cheap cloud computing on demand - the elastic cloud computing (EC2). I’ve been using it for several of my projects and I found that using spot instances is an extremely useful and effective way to test/experiment with your ideas.
In this post, I will show you how to set up iPython Notebook which you can then access from your web-browser. Please note, that I will be installing nVidia’s drivers and CUDA so that you can use Theano to speed up your calculations using GPU.
Installing Ubuntu Packages
After you start your GPU large instance and connect to it using SSH just install these packages
Then, we can add CUDA’s lib and bin to bashrc
And then test if CUDA works as it is supposed to be
I prefer to use Python’s virtual environments to keep each python environment separate.
First of all, create a password in python
Then, create a certificate for notebook
Followed by creating a iPython profile
Then, modify the profile .ipython/profile_quantServer
Run iPython Notebook
After all this, we can finally run iPython Notebook on our server by simply calling
And that’s it! After this, you can connect to your server on port 9999 and enjoy iPython. Once finished, you can create an amazon image (AMI) and use it in future to set up your spot instance and avoid running all these scripts over and over again.