According to an Apple engineer on the machine learning team, the newest beta updates for macOS Ventura, iOS 16, and iPadOS will improve performance for the Stable Diffusion AI art generating routines.
AI on a Mac
Apple released macOS Ventura 13.1 beta four for developers on Thursday, as well as betas for the company's other operating systems.
A Twitter user named Atila, whose bio lists him as a machine learning tinkerer at Apple, tweeted about the new macOS Ventura beta.
Today's release of macOS Ventura 13.1 Beta 4 and iOS and iPadOS 16.2 Beta 4 include optimizations that let Stable Diffusion run with improved efficiency on the Apple Neural Engine as well as on Apple Silicon GPU
-- Atila @NeurIPS (@atiorh) December 1, 2022
He says the latest betas include optimizations that let Stable Diffusion run with improved efficiency on the Apple Neural Engine and the Apple Silicon GPU.
Apple has shared sample code for neural network model conversions from PyTorch to the company's Core ML system. They tested a baseline configuration of an M2 MacBook Air with 8GB of RAM, and it completed 50 iterations of the model in 18 seconds.
For a "distilled" version of Stable Diffusion 2, an M2 device can generate an image in less than one second.
Stable Diffusion is a text-to-speech machine-learning model that can be used for generating digital artwork. Users can type something like "cat piloting a submarine" and the program will generate an image based on the text prompt.
The speed of the generation varies on the prompt and the hardware that the code is being run on.
Apple has a code repository on GitHub to run Stable Diffusion on Apple Silicon with Core ML.