Intel recently released the Intel Extension for TensorFlow, a plugin that allows TF DL workloads to run on Intel GPUs, including experimental support for the Intel Arc A-Series GPUs running on both native-Linux and Windows Subsystem for Linux 2 (WSL2). Installing the Intel Extension for TensorFlow on WSL2 with Ubuntu 20.04 is easy, and the extension can be used to run the popular Keras-CV Stable Diffusion capability with TF in a Jupyter notebook for text-to-image generation on an A770 GPU.
Running TensorFlow Stable Diffusion on Intel Arc GPUs | by Christopher Lishka | Intel Analytics Software | Medium
This post provides a quick walkthrough of how to get Stable Diffusion, a text-to-image AI model created by CompVis, Stability AI, and LAION, up and running on the Intel Arc A770 GPU. It explains how to install the Intel Arc drivers and the Intel oneAPI Base Toolkit, as well as how to set up the Hugging Face Diffusers library and PyTorch and Intel Extensions for PyTorch. Finally, it provides instructions on how to use the Accelerator library to reduce the CPU memory usage.
PyTorch Stable Diffusion Using Hugging Face and Intel Arc | by TonyM | Towards Data Science