Hi! I'm developing a simple input-lag sensitive application as part of my masters thesis. I need to use vsync but I also need to sample mouse input as close to monitor refresh as possible. It seems to me triple buffering would solve my problem. I was doing some testing with this code from Lazy Foo' Productions: lazyfoo.net/tutorials/SDL/51_SDL_and_modern_opengl/index.php lazyfoo.net/tutorials/SDL/51_SDL_and_modern_opengl/51_SDL_and_modern_opengl.zip (tutorial source code) It renders a quad on screen, no projection or shading or anything fancy. On my Radeon HD 7950 I enabled vsync and triple buffering in Catalyst settings and printed out render loop durations. Unfortunately the loop durations are ~1/60 s. I was expecting something below 1ms for rendering a quad. It seems SDL_GL_SwapWindow(gWindow) blocks until the monitor is updated. It seems the OpenGL Triple buffering in Catalyst settings is of the "flip queue/render ahead queue" kind that doesn't drop frames when flipping (back) buffers. Instead it's blocking when the flip queue is full like with double buffering with vsync. See a detailed discussion here: forum.guru3d.com/showthread.php?t=376645 Also see an old Anandtech article for the difference if my explanation was unclear: anandtech.com/show/2794 . Question: Is it possible to implement parallel (true) triple buffering in OpenGL, Direct3D or Mantle? I'm planning on using OpenGL (modern pipeline). I'm not familiar with Direct3D or Mantle, but I might consider switching if they have what I need. Madness: "To be able to post links or images your post count must be 10 or greater. You currently have 0 posts.":bang:
