Does ggml supports Model splitting to gpu and cpu ??

[forrackun]

i have a 2gb mx150 Nvidia gpu and ofcourse intel gpu in laptop?

can i make use of mx150 by loading some layers to gpu alsong side cpu?

can i do similar with intel cpu and gpu ?? in this case they are pointing to same memory right? even though kernel allocates seprate chunk for gpu .. i remembering seeing some way to read same memory in intel opencl doc or sth does that help ??