D11's kernel style window system architecture promises significant performance benefits over X11's current client/server architecture involving clients and servers running in separate address spaces. The performance benefits of D11 come from reduced context switching overhead, the elimination of local protocol packing and unpacking, the elimination of local transport overhead, and more efficient support for pixmap rendering.
D11 does require significant operating system support for active contexts, alternate personality processes, and protected procedure calls. The performance of protected procedure calls is vital to fulfilling D11's promise. Implementing the window system component of D11 is eased because most of the existing X11 client and server code can be reused in a D11 implementation.
Comparing D11 with other methods for improving window system performance, D11 reduces system overhead for local window system usage in all significant areas. D11 can be considered a logical extension of DGHA techniques to encompass the full range of X requests, not just rendering requests. D11 can also be considered the next logical step beyond a shared memory transport, where not only is the copy of data from the transport buffer eliminated, but so is the copy into the transport buffer and all associated protocol packing and unpacking.
No compromises in network transparency, X11 compatibility, or protection and reliability are made by D11. D11 favors local window system usage at a time when trends for high-bandwidth applications will likely result in greater reliance on and need for local window system performance. D11 optimizes local window system performance currently unharnessed by X11.