Will the Conv3D operation be supported on MPS through PyTorch?

I am new here. I’ve been trying to use stable video diffusion in ComfyUI on an Intel Core i9 MacBook Pro running Sonoma 14.3.1 with AMD Radeon Pro 5500M 8 GB. I come up against this error: RuntimeError: Conv3D is not supported on MPS. I can use the CPU instead, if I want to wait for a few hours instead of minutes, but this isn’t practical. I have tried the PyTorch Nightly release as well as some of the other suggestions on the Github page - https://github.com/comfyanonymous/ComfyUI/issues/2044 but have not been able to get it to work using MPS. Are there plans for the Conv3D to be supported on MPS through PyTorch?

1 Like


I have the same problem, did you find a solution?