Optimization techniques for a DSP based H.264 embedded system

This paper presents the various optimization techniques that are used to fine-tune the performance of a H.264 encoder implemented on a DSP based embedded platform. The approach includes the use of software tool to perform data flow analysis and to identify the critical codes that most affect the encoder performance. Packed data processing, judicious choice of data/code memory allocation, compiler intrinsic and optimizations are then employed to fine-tune the program to extract the maximum performance achievable of the DSP based embedded platform. Actual measurement results are presented to illustrate the incremental improvements in the system performance with each technique employed, and the tradeoffs in PSNR and compression efficiency in attaining the performance.