Skip to content
Milind Soorya
Go back

Installing TensorFlow on M1 MacBook Air with GPU (Metal)

Installing TensorFlow on M1 MacBook Air with GPU banner

You can now leverage Apple’s tensorflow-metal PluggableDevice in TensorFlow v2.5 for accelerated training on Mac GPUs directly with Metal. Learn more here.

OS Requirements

macOS 12.0+ (latest beta)

Currently Not Supported

Multi-GPU support
Acceleration for Intel GPUs
V1 TensorFlow Networks

Table of contents

Open Table of contents

What is tensorflow?

TensorFlow is open source deep learning framework created by developers at Google and released in 2015. It is actively used at Google both for research and production needs.

Please check references if you would like to read more about TensorFlow and other popular deep learning libraries.

What is miniforge?

Miniforge is the community (conda-forge) driven minimalistic conda installer. Subsequent package installations come thus from conda-forge channel.

What is miniconda?

Miniconda is the Anaconda (company) driven minimalistic conda installer. Subsequent package installations come from the anaconda channels (default or otherwise).

Note: Uninstall Anaconda/Anaconda Navigator and other related previously installed version of conda-based installations. Anaconda and Miniforge cannot co-exist together.

Uninstalling existing anaconda/conda from macOS

Install the Anaconda-Clean package from Anaconda Prompt

conda install anaconda-clean
anaconda-clean --yes

Remove entire anaconda installation directory from macOS

which anaconda

After running above command you should see the directory where anaconda is installed.

now recursively remove that folder

rm -rf YOUR-ANACONDA-DIRECTORY

Step 1: Install Miniforge

Install miniforge - https://github.com/conda-forge/miniforge

Please note that this tutorial is for arm64 : Apple Silicon

Download and install Conda env

chmod +x ~/Downloads/Miniforge3-MacOSX-arm64.sh
sh ~/Downloads/Miniforge3-MacOSX-arm64.sh
source ~/miniforge3/bin/activate

Create an anaconda environment

conda create -n tf python=3.8

Activate the environment

conda activate tf

Step 2: Install TensorFlow

Install TensorFlow dependencies

conda install -c apple tensorflow-deps

Install base TensorFlow

python -m pip install tensorflow-macos

Install tensorflow-metal plugin

python -m pip install tensorflow-metal

### Install common Data Science packages

pip install tensorflow-datasets pandas jupyterlab

References


Share this post on:

Previous Post
How to build a simple CNN based Image classifier using Keras
Next Post
OpenPose - Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields