Hi, > I'm actually not sure if you need to clone it at all. At least with the > ATI driver you can just use the accelerant twice. If you > compile and run the SlimDemo (./graphics/SlimDemo) you should get a > black and white screen but the app_server will still be running and > switching to a workspace with a different resolution (or the > ctrl-shift-option-command-whatever f12 thing) will bring your desktop > back. > If it's not possible to use the accelerant directly, you should be able > to build something from the SlimDemo. > I don't remember how exactly, but I had the first accelerated Haiku > window right under R5 with the app_server still running. It either used > SlimDemo as basis or the AccelerantHWInterface directly. Ok, thanks for the info. I'll get to work then. I think it will be much easier to optimize Painter once it runs exactly like it would under Haiku. Only without rebooting. One thing that I'm still wondering about... now that I would use BDirectWindow under R5 myself, directly banging the HW but still honoring the clipping - how is this supposed to work with the asynchronous nature of client server communication? Is the app_server supposed to wait for an acknowledgement from all currently showing BDirectWindows before enforcing a new clipping state for _all_ windows? As long as clients have the app_server draw, this is no problem, but when they directly access the framebuffer themselves? How is the clipping synchronized? Any insights? Best regards, -Stephan