New to Comfy and finally have a basic working knowledge thanks to some excellent tutorials from Pixaroma and others. Most things work on Mac but it’s SLOW. System is a Mac Studio M1 Ultra with 128 GB of RAM running Sequoia 15.3.1 and ComfyUI 04.23
Running the template files:
Image Generation - 9.28 seconds
Image to Image - 45.5 seconds
2 Pass Upscale: - 46 seconds
Flux Schnell - 81.5 seconds
So I’m curious to know how much faster these run on newer hardware. Are my speeds typical or did I botch my rather difficult installation a few weeks ago?
We all know that these models really want to run on nice Nvidia GPUs. For comparison today I ran a fairly complex Flux DevQ8 prompt for an 1888x1056 image on the cloud service Runninghub (not sure what the GPU was) and it completed in 43 seconds, but pretty much the same workflow on my machine took 9 minutes!
Anybody have any thoughts?