“Parallel multi-view rendering on multi-core processor systems” by Ohlenburg and Broll

  • ©Jan Ohlenburg

Conference:


Type:


Title:

    Parallel multi-view rendering on multi-core processor systems

Presenter(s)/Author(s):



Abstract:


    Dual–core processors have now become a standard in modern desk- top and notebook computers and it is only the beginning of a new  trend leading to massive multi–core processor systems. In order to utilize multi–core systems applications have to be highly multi– threaded. Multi–threaded rendering is still a non trivial process, since update and rendering processes have to be synchronized very carefully. Failure to synchronize correctly will result in a drastic performance loss.  We present an approach to multi–threaded rendering taking advan- tage of the power of today’s and future multi–core systems using  OpenGL. This approach supports a high number of parallel ren- dering threads, simply limited by the capabilities of the computer,  where each rendering thread may have its own camera, render state and refresh rate while sharing resources like textures and display lists. In order to achieve the maximum performance one thread – called the manager– performs the updates of the scene graph and blocking all render threads during this time period.  

     


ACM Digital Library Publication:



Overview Page: