On Unity’s roadmap for 5.4 it mentions in “Graphics: Multi-threaded Rendering Phase 2” that they are going to add “Parallel command list generation”.
I’m not sure what is meant by that.
This looks like it describes what that is.
In short, every time you send the videocard a task there is overhead. Thats why you never send particles to a video card individually. You send them as one big pile of triangles. This takes that concept further by breaking these tasks into chunks that can be handed to different cpus, making complex scenes run much faster. Sounds easy, but its actually quite hard to do right.