Hi Markus, it's difficult to figure out while you want exactly compute shaders, but with OpenCL you might have several OpenCL threads - one which will communicate with OpenGl and another one which will perform independent computations <https://software.intel.com/en-us/articles/opencl-and-opengl-interoperability-tutorial> Regards Sergey
On Sun, Nov 9, 2014 at 4:16 PM, Markus Hein <[email protected]> wrote: > Hello All, > > maybe someone can point me to example or osg-users discussion where data > computation is done in parallel to scene rendering. Maybe someone has done > similar stuff before, I could not find the right answer after searching in > osg example code. > > I want to realize a pure OSG based application , where a compute thread > is using compute shaders, updating data stored in BufferObjects stored in > GPU memory. This thread must run very straight at high and predictable > framerate for numerical precision. > > In addition, the app will have one or multiple viewer threads, > rendering the scene at fastest possible framerate. The visual scene will > use the data stored in GPU memory that is updated by the compute thread > for rendering (avoiding data transfer to host and back). > > I see that osg viewer.frame() is dispatching rendering of the scene , but > fastest possible framerate is given by viewers framerate. Is there a way > to realize a thread that doesn't depend on viewers framerate for fast and > predictable computation ? I expect that my viewer threads will run at 20-60 > Hz, but the compute thread must run > 60 Hz, eventually locked to some > given framerate. > > Thanks for some info! > > regards, Markus > _______________________________________________ > osg-users mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org >
_______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
