this is of course correct (even cegui will probably need a lot of work) but you need to use hardware acceleration.
Example: CeGUI uses the following renders:
There are currently CEGUI renderer modules for Microsoft® DirectX® 9, 10 and 11, OpenGL, the Ogre engine, and the Irrlicht engine (CEGUI 0.7).
This is just muuuuuch faster and better than using standard windows apis
