The model took a full 10 minutes to load. Is it the same for you guys? I have 32GB of RAM and 16GB of VRAM.
Is there something wrong with my computer settings? It also takes about 120s to generate a picture, and this is the case for both V5 and V11. . .
took a full 10 minutes to load. Is it the same for you guys?
No
Is there something wrong with my computer settings? It also takes about 120s to generate a picture
Yes
Is there something wrong with my computer settings? It also takes about 120s to generate a picture, and this is the case for both V5 and V11. . .
loading from hdd, 10min, maybe.
else
try --disable-smart-memory option
also some system specs could help, is it windows? where is swapspace located? ssd/hdd.
for me on linux without gui so headless ist fits barely into the 32gb ram, if you have windows it needs definively swap of doom to run.
First load needs, no matter how well you're configured, it's normal.
There is a method to isolate the CLIP component within the AIO model and run it on a remote PC or Docker environment. This enables the AIO model to run smoothly on any device with at least 8GB of GPU memory and a minimum of 32GB of RAM. https://github.com/Goldlionren/ComfyUI_RemoteQwenSuite