2. Getting started guide¶

2.1.1.1. pip¶

pip is a package manager for other Python dependencies. Once you have pip, installing additional python dependencies should be straightforward. pip comes with setuptools, which is required for installing Bifrost. The detailed instructions for pip can be found here, but the basics are as follows:

1. Download `get-pip.py <https://bootstrap.pypa.io/get-pip.py>`__

2. Navigate to the download directory, and run python get-pip.py --user, which will install a local copy of pip.

3. Check pip is working with pip list, which will give the versions of pip and setuptools.

2.1.1.2. numpy, matplotlib, contextlib2, simplejson, pint, graphviz, ctypesgen¶

If you have already installed pip, this step should be as simple as pip install --user numpy matplotlib contextlib2 simplejson pint graphviz ctypesgen==1.0.2.

2.1.2.1. CUDA¶

CUDA allows you to program your GPU from C and C++. You will need an NVIDIA GPU to do this. If this is your first time trying out Bifrost, and you don’t have CUDA yet, we recommend that you skip this step, and try out a CPU-only version of Bifrost. Then, once you have that first experience, you can come back here for a speedup.

If you are ready to work with a GPU, you will want to get the newest CUDA toolkit. Follow the operating system-specific instructions to install.

2.1.2.2. Other Dependencies¶

• exuberant-ctags

• Basic build tools (make, gcc, etc.)

On Ubuntu, the following command should grab everything you need:

sudo apt-get install build-essential software-properties-common exuberant-ctags