comfyanonymous

https://github.com/comfyanonymous/ComfyUI/blob/master/comfy/sampler_helpers.py#L64 If you change: `memory_required = model.memory_required([noise_shape[0] * 2] + list(noise_shape[1:])) + inference_memory` to: `memory_required = model.memory_required([noise_shape[0] * 4] + list(noise_shape[1:])) + inference_memory` Does that fix it?

There has been some optimizations to the lowvram mode which should speed things up for most people. If you have issues you need to post your hardware and which model...

Does the change I made you do above have any effect on the number of frames you can do?

The problem is most likely related to what's in: https://github.com/phazei/ConfyUI-node-prompt-stash-saver/tree/main/js This sounds like a frontend issue that should be reported on: https://github.com/Comfy-Org/ComfyUI_frontend

 I'm not sure if returning the full paths is a good idea, it might be better to return just the custom node name.

There is no safety checker this is an issue with the model itself and how it was trained.

any example lora file?

That sounds like a hardware or driver issue. Since there are no other reports of the software making GPUs crash I don't think it's related to the software. Software can...

GPUs have an MMU so memory fragmentation shouldn't be an issue at all. I know pytorch does some dumb things with memory but I don't think it's dumb enough for...

Are you sure this is with mixProV4_v4.safetensors because I tried it and it works. I also see a print in your log that only happens when a ckpt is loaded...