The ability to attach accelerators that have shared memory access that is coherent with the system's processors.

An implementation of AME on a hardware accelerator.

The ability to dispatch multiple virtual machines on a single core at the same time.

An alternative to memory lock base synchronization.