Installation¶
The CUDA Toolkit is required for all installation methods. Installation guides are available for Linux, MacOS, and Windows.
The simplest way to install the most recent release of SPORCO-CUDA is from
PyPI via pip
. The CUDAHOME
environment variable must be set so that the build process can find the local CUDA toolkit installation. For example, under Linux with the CUDA toolkit installed at /usr/local/cuda-10.1
CUDAHOME=/usr/local/cuda-10.1 pip install sporco-cuda
SPORCO-CUDA can also be manually installed from source, either from the development version from GitHub, or from a release source package downloaded from PyPI.
To install the development version from GitHub, do
git clone https://github.com/bwohlberg/sporco-cuda.git
followed by (again, assuming building under Linux with the CUDA toolkit installed at /usr/local/cuda-10.1
)
cd sporco-cuda
export CUDAHOME=/usr/local/cuda-10.1
python setup.py build
python setup.py test
python setup.py install
Please report any test failures. The install command will usually have to be performed with root permissions, e.g. under Linux
sudo -H CUDAHOME=/usr/local/cuda-10.1 pip install sporco-cuda
or
sudo python setup.py install
The procedure for installing from a source package downloaded from PyPI is similar.
A summary of the most significant changes between SPORCO-CUDA releases can
be found in the CHANGES.rst
file. It is strongly recommended to
consult this summary when updating from a previous version.
Requirements¶
The primary requirements are Python itself, Cython, the CUDA Toolkit, and modules future, numpy, and sporco.
Installation of these requirements is system dependent.
If the CUDA Toolkit is not already installed, install it following the instructions from Nvidia.
Under Ubuntu Linux, the remaining requirements can be installed via the package manager and pip. Under Ubuntu 18.04, the following commands should be sufficient for Python 2
sudo apt-get -y install cython python-numpy python-pip python-future python-pytest
sudo -H pip install pytest-runner sporco
or Python 3
sudo apt-get -y install cython python3-numpy python3-pip python3-future python3-pytest
sudo -H pip3 install pytest-runner sporco
Some additional dependencies are required for building the documentation from the package source, for which Python 3.3 or later is required. For example, under Ubuntu Linux 18.04, the following commands should be sufficient
sudo apt-get -y install python3-sphinx python3-numpydoc
sudo -H pip3 install sphinx_tabs sphinx_fontawesome
Not yet tested under Mac OS. If you build SPORCO-CUDA under Mac OS, please submit details for inclusion here.
Not yet tested under Windows. If you build SPORCO-CUDA under Windows, please submit details for inclusion here.