Slow generation speed

[healthyfat]

I am having a problem with very slow generation speed when using AutoCFG. Wonder if this might have anything to do with below warning or is it just my 6GB VRAM to little for this node?

Requested to load SDXL
Loading 1 new model
0%| | 0/24 [00:00<?, ?it/s]

C:\Programs\ComfyUI\python_embeded\Lib\site-packages\torchsde_brownian\brownian_interval.py:608: UserWarning: Should have tb<=t1 but got tb=14.614644050598145 and t1=14.614643.
warnings.warn(f"Should have {tb_name}<=t1 but got {tb_name}={tb} and t1={self._end}.")

25%|████████████████████▊ | 6/24 [06:50<20:29, 68.31s/it]

[Extraltodeus]

Hello! Unfortunately I can't really tell like this. You message is about something else and the iteration speed, while slow indeed, may come from anything.

[healthyfat]

You right. That warning is not due to AutoCFG. I will just mention that after I cancel the process and start it again then it works normal as on example below.

Loading 1 new model
0%| | 0/24 [00:00<?, ?it/s]

C:\Programs\ComfyUI\python_embeded\Lib\site-packages\torchsde_brownian\brownian_interval.py:608:
UserWarning: Should have tb<=t1 but got tb=14.614644050598145 and t1=14.614643.
warnings.warn(f"Should have {tb_name}<=t1 but got {tb_name}={tb} and t1={self._end}.")

4%|███▍ | 1/24 [01:38<37:34, 98.04s/it]
Processing interrupted
Prompt executed in 189.60 seconds

got prompt
100%|██████████████████████████████████████████████████████████████████████████████████| 24/24 [00:52<00:00, 2.17s/it]
Requested to load AutoencoderKL
Loading 1 new model
Prompt executed in 74.84 seconds

[Extraltodeus]

Sometimes while using comfy the first time that a set of parameters is used for some reason I get a slowdown too. Now I'm not sure about your setup.

[healthyfat]

Setup? You mean my hardware or Comfy? I use GeForce 2060 6GB.

[Extraltodeus]

ComfyUI can automatically switch into low vram mode.

try to start the UI with the --normalvram argument