Hi, first of all thank for your work.
Secondly since Teacache naturally requires a moderately large quantity of RAM to be used, it would be wise to explicitly mention it on the Readme since is something nobody seems to understand and doing so would avoid all the complaint about it not working or being slower than standard time.
Thirdly it would makes sense and make the whole thing works if the user could choose to store the cached data on the system RAM instead of VRAM.
As a reference on how to implement it take a look at https://github.com/kijai/ComfyUI-KJNodes/blob/1016861aec609d8b1f58c23fd8481edd5d9f8224/nodes/model_optimization_nodes.py