PR #8035 ggml-cuda: Adding support for unified memory

Adding support for unified memory

matteoserva committed 2 years ago

adding again the documentation about unified memory

matteoserva committed 2 years ago

refactoring: Moved the unified memory code in the correct location.

matteoserva committed 2 years ago

Fixed compilation error when using hipblas

matteoserva committed 2 years ago

cleaning up the documentation

matteoserva committed 2 years ago

Updating the documentation

matteoserva committed 2 years ago

adding one more case where the PR should not be enabled

matteoserva committed 2 years ago

llama.cpp ggml-cuda: Adding support for unified memory #8035 Merged