Thanks for your help Robert. I think I will take your advice as to seeing why my app is so cpu limited. I am doing a 600 particle simulation using the bullet physics library. Therefore, I have 1 geometry, 600 geodes, 600 materials, and 600 Positionattitudetransforms. The shared geometry and the materials are attached to the geodes which are then attached to the PAT's. The application runs in real time, and every frame each sphere position and material color changes. I do all the physics stuff and the updating of the positions and materials in a separate thread. So the main OSG thread just runs viewer.frame() with no node update callbacks. I am running the viewer in single threaded mode. I profiled the main thread and find that most of the time spent is in osg::group::computebound and osg::matrixd::premult, premulttranslate, postmult, etc. Is there maybe another structure other than Positionattitudetransform that I can use where I can do all the matrix computations in my other thread so that viewer.frame() doesn't have to do them? Any other suggestions?
... Thank you Brian ------------------ Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=31639#31639 _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
