Enables pointer manipulation for sharing CUDA memory between processes in EXLA.
In theory, IPC can also be supported for host-allocated memory, but this is not implemented yet.
It is also possible to use local (as in, "same process") pointers through these functions,
albeit there's not much gain besides automated testing.
Enables pointer manipulation for sharing CUDA memory between processes in EXLA. In theory, IPC can also be supported for host-allocated memory, but this is not implemented yet. It is also possible to use local (as in, "same process") pointers through these functions, albeit there's not much gain besides automated testing.