Multi-GPU inference

[HeathHose]

Thank you for your excellent work on this project.
I noticed that multi-GPU inference is currently not supported.
May I ask if there are any plans to support multi-GPU inference in the future?
Thank you very much for your time!

[wymanCV]

The original repository should support multi-GPU inference, but we may have accidentally broken the pipeline and haven’t checked it yet. If you’re in a hurry, you can use the ComfyUI from our README: it only requires ~10 GB of VRAM. We’ll verify the multi-GPU functionality later. Thanks.