Installing Tensorflow on Apple M1 With the New Metal Plugin

How to enable GPU acceleration on Mac M1 and achieve a smooth installation

Photo by Wesson Wang on Unsplash

Since Apple abandoned Nvidia support, the advent of the M1 chip sparked new hope in the ML community. The chip uses Apple Neural Engine, a component that allows Mac to perform machine learning tasks blazingly fast and without thermal issues.

When Apple with M1 was released, the integration with Tensorflow was very difficult. The process involved downloading, among other packages, a pre-configured environment.yml file with specific dependencies in such a way that no dependency conflicts will arise.

Unfortunately, that was not always the case. This article discusses how to install Tensorflow on Miniforge by using the Metal plugin, a process that is more straightforward and less prone to errors.

Step 1: Install Xcode

The first component to install is Xcode, which can easily be downloaded from the App Store. Additionally, install the Command Line Tools:

$ xcode-select --install

Step 2: Install MiniForge

MiniForge is a minimalistic conda installer which uses by default the conda-forge channel and supports, among others…