Hello! Now that I can render into a BDirectWindow I found another issue, the lack of hardware cursor support, at least in the radeon_hd driver, and the fact that BDirectWindow doesn't check it. According to the BeBook, BDirectWindow::SupportsWindowMode() should return false if the graphics driver does not have hardware cursor support... but in Haiku's implementation it only checks for B_PARALLEL_ACCESS (I opened bugticket for this, see [1]). When using the BDirectWindow without hardware cursor, drawing into the window draws even over the cursor, making it invisible. This can easily be tested with for example Chart from the demo applications... Set it to "Slow rotation" and DirectWindow as display mode. Now place the cursor into the drawing area in a place where stars are moving and do not move the cursor anymore -- you can see the stars that pass over the cursor partially erase it, leaving holes in the cursor. When you have an application redrawing the entire window every frame, the cursor is completely invisible, only leaving a flickering rectangle when moving the mouse. I wonder if it makes sense to use BDirectWindow at all if the radeon_hd accelerant doesn't support hardware cursors currently. Oddly enough, by the way, radeon_hd does report that it supports it, it uses the default flags from create_display_modes.cpp which include B_HARDWARE_CURSOR. Is this a bug and should I open a ticket for it too? Since I know only little about how modern radeon cards work... can it be expected that hardware cursor support will be integrated anytime soon or is it a bigger effort? I'm asking because, in the latter case, I'd rather stay with my alternate (already working) solution for the near future, i.e. drawing into a BBitmap and bringing it to screen via BView::DrawBitmap()... it's many times slower than BDirectWindow, but it would work on more machines then. [1] https://dev.haiku-os.org/ticket/9185 -- So long, Julian