Run Stable Diffusion on Your M1 Mac’s GPU (replicate.com)

Article summary

The article provides a guide on how to run Stable Diffusion on an M1 Mac's GPU, including the prerequisites, setup, and installation of dependencies. It also mentions that the standard release of Stable Diffusion can work on Intel-based Macs with some tweaks. The guide is based on a fork of the stable-diffusion repository on GitHub. Running Stable Diffusion on an M1 Mac's GPU can take around 1-2 minutes for a 512x512 image.

Main themes

• Stable Diffusion on M1 Macs
• GPU performance
• Automation and labor
• PyTorch and backend support
• Dependency issues and troubleshooting
• Robotics and generative algorithms
• Docker compatibility and limitations
• Hardware comparisons and performance benchmarks

What commenters say

• Some users have successfully run Stable Diffusion on their Intel-based Macs with modifications, while others have encountered dependency issues.
• The use of PyTorch as a backend allows for running Stable Diffusion on Apple Silicon, serving as a 'glue' for different platforms.
• The performance of Stable Diffusion on M1 Macs varies greatly depending on the amount of RAM available, with 8GB machines being significantly slower than those with 16GB or more.
• There are concerns about the automation of tasks and the potential impact on human labor, with some arguing that certain tasks will always require human intervention.
• The combination of progress in generative algorithms and robotics could lead to significant advancements in automation.
• Some users have reported issues with getting CUDA to work on their Macs, due to Apple's lack of support for Nvidia GPUs.
• The article's guide is not compatible with Docker for Mac due to its inability to access the M1 GPU.
• The performance difference between M1 Macs and high-end GPUs like the 3070 is significant, with the latter being much faster and more powerful.