At this stage, we recommend installing DGL from
Currently DGL is tested on
- Ubuntu 16.04
- macOS X
- Windows 10
DGL is expected to work on all Linux distributions later than Ubuntu 16.04, macOS X, and Windows 10.
DGL also requires the Python version to be 3.5 or later. Python 3.4 or less is not tested, and Python 2 support is coming.
DGL supports multiple tensor libraries (e.g. PyTorch, MXNet) as backends; refer Working with different backends for requirements on backends and how to select a backend.
Install from conda¶
Once the conda environment is activated, run
conda install -c dglteam dgl
Install from pip¶
One can simply run the following command to install via
pip install dgl
Working with different backends¶
Currently DGL supports PyTorch and MXNet.
The backend is controlled by
DGLBACKEND environment variable, which defaults to
pytorch. Currently it supports the following values:
|pytorch||PyTorch||Requires 0.4.1 or later; see official website|
Requires nightly build; run the following command to install:
pip install --pre mxnet
or cuda version (e.g. for cuda 9.0)
pip install --pre mxnet-cu90
|numpy||NumPy||Does not support gradient computation|
Install from source¶
First, download the source files from GitHub:
git clone --recursive https://github.com/dmlc/dgl.git
One can also clone the repository first and run the following:
git submodule init git submodule update
Install the system packages for building the shared library, for Debian/Ubuntu users, run:
sudo apt-get update sudo apt-get install -y build-essential build-dep python3-dev make cmake
For Fedora/RHEL/CentOS users, run:
sudo yum install -y gcc-c++ python3-devel make cmake
Build the shared library. Use the configuration template
Copy it to either the project directory or the build directory and change the
configuration as you wish. For example, change
enable cuda build. You could also pass
-DKEY=VALUE to the cmake command
for the same purpose.
- CPU-only build:
mkdir build cd build cmake .. make -j4
- Cuda build:
mkdir build cd build cmake -DUSE_CUDA=ON .. make -j4
Finally, install the Python binding.
cd ../python python setup.py install
Installation on macOS is similar to Linux. But macOS users need to install building tools like clang, GNU Make, cmake first.
Tools like clang and GNU Make are packaged in Command Line Tools for macOS. To install:
To install other needed packages like cmake, we recommend first installing Homebrew, which is a popular package manager for macOS. Detailed instructions can be found on its homepage.
After installation of Homebrew, install cmake by:
brew install cmake
Then go to root directory of DGL repository, build shared library and install Python binding for DGL:
mkdir build cd build cmake .. make -j4 cd ../python python setup.py install
We tested installation on macOS X with clang 10.0.0, GNU Make 3.81, and cmake 3.13.1.
Currently Windows source build is tested with CMake and MinGW/GCC. We highly recommend using CMake and GCC from conda installations. To do so, run
conda install cmake m2w64-gcc m2w64-make
Then build the shared library and install the Python binding:
md build cd build cmake -DCMAKE_CXX_FLAGS="-DDMLC_LOG_STACK_TRACE=0 -DDGL_EXPORTS" -DCMAKE_MAKE_PROGRAM=mingw32-make .. -G "MSYS Makefiles" mingw32-make cd ..\python python setup.py install
MD build CD build cmake -DCMAKE_CXX_FLAGS="/DDGL_EXPORTS" -DCMAKE_CONFIGURATION_TYPES="Release" .. -G "Visual Studio 15 2017 Win64" msbuild dgl.sln cd ..\python python setup.py install