M1 Tensorflow
GPU accelerated TensorFlow on M1
See here for the official documentation.
Table of contents
Install Miniforge
Download and install Miniforge3 conda:
curl -L -O https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-MacOSX-arm64.sh
chmod +x Miniforge3-MacOSX-arm64.sh
sh Miniforge3-MacOSX-arm64.sh
# Optional
conda config --set auto_activate_base false
For information on conda see here.
Installing Miniconda3 macOS Apple M1 64-bit bash (Py38 conda 4.10.1 2021-11-08) did not work. Miniforge’s arm64 was the only thing that worked.
Install TensorFlow dependencies
Create a new environment:
conda create -n tf-m1 python=3.9
You can use either Python 3.8 or 3.9.
Install tensorflow-deps:
conda install -c apple tensorflow-deps
Install base TensorFlow and tensorflow-metal
Making sure the environment is activated:
Check environment is indeed activated by which pip.
Especially if you’re doing it in a VSCode terminal.
pip install tensorflow-macos
pip install tensorflow-metal
To upgrade to a new TensorFlow version
From the official doc, the recommended way is:
# uninstall existing tensorflow-macos and tensorflow-metal
python -m pip uninstall tensorflow-macos
python -m pip uninstall tensorflow-metal
# Upgrade tensorflow-deps
conda install -c apple tensorflow-deps --force-reinstall
# or point to specific conda environment
conda install -c apple tensorflow-deps --force-reinstall -n my_env