Ctrl+K

Build and Install

Build and Install#

Install from PyPI#

MLX Data can be installed from PyPI. For most Linuxes and Apple silicon Macs one can simply,

pip install mlx-data

This should install the package together with the necessary dependencies to read audio, images, video and remote content from S3.

Building from source#

MLX Data consists of a C++ library with shallow python bindings and some Python helpers. You can choose to build and use only the C++ library or simply pip install locally the Python library which transparently builds the required C++ backend.

Dependencies#

Handling data requires reading various filetypes over various protocols. This means that MLX Data has a couple of dependencies e.g. for loading images, audio and videos or fetching files from an S3 bucket.

The following dependencies are optional but without them some functionality of MLX data will not be built. Below we enumerate the list of dependencies and subsequently we provide example shell commands to install them using homebrew or on Ubuntu using apt.

  • libsndfile to load audio files.

  • libsamplerate for audio samplerate conversion.

  • ffmpeg for loading video files.

  • libjpegturbo to load JPG images.

  • zlib, bzip2 and liblzma to process compressed streams.

  • aws-sdk for accessing S3 buckets.

To install the above on a Mac using homebrew one needs simply to run:

brew install libsndfile libsamplerate ffmpeg jpeg-turbo zlib bzip2 xz aws-sdk-cpp

For an Ubuntu machine, on the other hand:

sudo apt install libsndfile1-dev libsamplerate0-dev ffmpeg libjpeg-turbo8-dev \
    zlib1g-dev libbz2-dev liblzma-dev

# We have to build the AWS SDK from source :-(
# None of this is needed if installing MLX Data from a prebuilt binary or if
# you don't care about fetching from S3

sudo apt install libcurl4-openssl-dev libssl-dev
git clone --depth 1 --recurse-submodules https://github.com/aws/aws-sdk-cpp.git
cd aws-sdk-cpp
mkdir build
cd build
cmake .. -DCMAKE_BUILD_TYPE=Release -DBUILD_ONLY="s3" -DBUILD_SHARED_LIBS=OFF
make -j
sudo make install

Finally before building the python bindings you need to make sure that you have pybind11 installed and a relatively modern version of CMake. For example, both be installed with pip,

pip install pybind11[global] cmake

Build Python bindings from source#

After installing the dependencies you care about as described above, you can directly install MLX data using pip.

cd /path/to/mlx/data
pip install .  # or pip install -e . for an editable install

Building the standalone C++ library#

MLX data can also be installed as a standalone C++ static library that you can link your projects against.

mkdir build && cd build
cmake ..
make -j
sudo make install

Subsequently, you can use CMake’s find_package(MLXData) which defines MLX_DATA_FOUND, MLX_DATA_INCLUDE_DIRS and MLX_DATA_LIBRARIES or simply link with -lmlxdata as it is normally installed in /usr/local/lib.

Contents