Installing ComfyUI on Windows for Intel Arc GPUsInstalling ComfyUI on Windows for Intel Arc GPUsInstalling ComfyUI on Windows for Intel Arc GPUs

Introduction

Intel's entry into the dedicated GPU market with the Arc series has provided a new, competitive option for AI enthusiasts. Thanks to the efforts of the PyTorch and Intel teams, ComfyUI now supports native hardware acceleration on Intel Arc GPUs via the XPU backend.

This guide will show you how to set up ComfyUI on your Windows machine to get the most out of your Intel Arc graphics card.

Prerequisites

Before we start, make sure you have the following:

• Operating System: Windows 10 or 11.
• GPU: An Intel Arc GPU (e.g., A750, A770).
• Drivers: The latest Intel Arc Graphics drivers. Support is improving rapidly, so keeping drivers up to date is crucial.
• Python: Python 3.10 or newer (3.12 is a good choice).
• Git: To clone the repository.
• 7-Zip: Recommended for extracting files if needed.

Installation Steps

Unlike NVIDIA users who have a "portable" one-click package, Intel Arc users currently need to do a manual installation. Don't worry, it's straightforward!

1. Clone the ComfyUI Repository

Open your Command Prompt or PowerShell and run:

git clone https://github.com/comfyanonymous/ComfyUI.git
cd ComfyUI

2. Set Up a Virtual Environment (Recommended)

To avoid messing up your global Python installation, creating a virtual environment is best practice.

# Create the environment named 'venv'
python -m venv venv

# Activate it
.\venv\Scripts\activate

Note: You'll see (venv) appear at the start of your command line, indicating the environment is active.

3. Install PyTorch with Intel XPU Support

This is the most important step. You need to install the version of PyTorch that knows how to talk to your Intel GPU.

Stable Version: This is the safest bet for stability.

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/xpu

Nightly Version (Experimental): If you want the absolute latest performance improvements (which are frequent for Intel Arc), try the nightly build:

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/xpu

4. Install Dependencies

Now install the rest of the required libraries for ComfyUI:

pip install -r requirements.txt

5. Add Your Models

Before you can generate images, you need to place your model files.

• Checkpoints: Move your .ckpt or .safetensors files (Stable Diffusion models) into: ComfyUI\models\checkpoints\
• VAEs: Move your VAE files into: ComfyUI\models\vae\

6. Run ComfyUI

You are all set! Run the following command to start the server:

python main.py

Your terminal will show a URL (usually http://127.0.0.1:8188). Open that in your web browser to start creating!

Troubleshooting

• "Torch not compiled with XPU enabled": This error means you likely installed the standard CPU or CUDA version of PyTorch by mistake. Uninstall it with pip uninstall torch and re-run the installation command from Step 3.
• Slow First Run: The first time you run a generation, it might take longer as kernels are compiled. Subsequent runs should be much faster.
• Memory Issues: If you have an 8GB card and run into memory errors with large workflows, try launching with python main.py --lowvram.

Conclusion

Intel Arc is becoming a strong contender in the AI space, and native ComfyUI support makes it a viable platform for creative work. By following this guide, you've unlocked the power of XPU acceleration for your stable diffusion workflows.

Happy creating!