An Optimized OGRE 3D Render Engine Based By Multi-Core Platform

With computer architecture development, the multi-core platform has been widely utilized. Combined with the parallel computing theory, multi-core platform greatly improves the application's performance by the multi-thread parallelization.However, multi-core platform has not been expanded to the graphic field. So this paper threads the OGRE 3D render engine which is widely used in the graphic field, and verifies the engine in the multi-core platform. When we get a optimized engine, we proof the multi-core platform and the parallel computing theory's advantage.This paper first analyses the OGRE system's overall framework from the point of view of parallelism. Then it abstracts and analyses the 3D render system modules in detail, and summarize the 3 ways of parallelizing the OGRE 3D render system, which are multi-frames rendering parallelization, loop iterations parallelization, call of functions parallelization.In addition, this paper designs the methods when using OpenMP to parallelize, which include date protection, threads amount setting, threads scheduling. In order to avoid the conflict between threads when accessing the same data in the same time, This paper uses the OpenMP clauses to protect and synchronize the data, such as the private clause, barrier clause, critical clause, and so on. And this paper set the proper threads amount and the best threads scheduling strategies, which maximizes the engine's performance.This paper employs the Win32 API to parallelize the render frames, and uses the OpenMP to parallelize the loop iterations and call of functions. At last, this paper exploits the OGRE benchmark programs to test the optimized engine's performance, and compares it with the original engine. The results show that the programs upgraded ratio from 27.70% to -3.40%, averaged 12.16%.In summary, this paper successfully optimizes the OGRE 3D render engine by threading in the multi-core platform that combined with the parallel computing theory, and effectively upgrades the engine's performance. So this paper accurately verifies the multi-core platform and the parallel computing theory's advantage, and achieves the research purpose.