State of data science's Python stack on Apple Silicon (M1)

Hello Python community. I was searching for a unified place in which to find all latest news and the state of the game of Apple Silicon support for Python’s libraries for data science (like pandas, numpy, scikit-learn, scipy, matplotlib, seaborn, others?) and other python libraries that are useful in general for data science (multiprocessing, asyincio, threading, progress reporting etc). Since I didn’t find such a place I thought that it might be useful to create it. If it already exists please point to me to the relevant page!

I’ll start by adding what news I have and I will keep this main post updated with new info as it comes out. There are three states of a software, I believe, for Apple Silicon: supported natively, working under Rosetta 2, not supported (broken in any way during installation or usage). I will only track the most recent version of each software to avoid this list getting out of hand.

  1. Python 3.10 alpha: native 3.10.0a4 from python.org
  2. Python 3.9: native with 3.9.1 from python.org (specific download for ARM64)
  3. Python 3.8: native with 3.8.2 as part of Apple’s developer CLI tools

via pip:

  • Pandas: ?
  • Numpy: native 1.19.4
  • Matplotlib: ?
  • Seaborn: ?
  • Scikit-learn: not supported ?
  • Scipy: native 1.5.4
  • tensorflow: ?
  • keras: ?
  • multiprocessing: ?
  • asyincio: ?
  • threading: ?
  • tqdm: ?
  • rich: ?

other distribution methods: