zero-copy with mmap was added to llama.cpp, but the way it was implemented sparked controversy.
The mmap change was just about mapping the model files in memory instead of copying them, which has less overhead.
[0]: https://developer.apple.com/documentation/metal/mtldevice/14...
It can still be beneficial on discrete GPUs because it avoids a copy inside the driver.
I brought it up because shuffling weights around takes lots of time, takes up more RAM, and saps memory bandwidth the IGP and CPU desparately need.