In Settings there is an option to dynamically optimise memory usage. This is not selected by default and I wonder why as having it switched on seems the natural default.. What is the downside of selecting it?
Looks like when this option is on, AI models are unloaded from graphic card memory after they have been used.
This solves problems for “small” stations (due to graphic memory capacity), but takes more time since thoses models have to be loaded again for next use.
Thank you. That makes a lot of sense 