Based On Cpu + Gpu In H. 264 Encoder Parallel Code Design

In recent decades, the multimedia industries benefitted in the sound video frequency standard promotion obtained the sufficient development, compared with the tradition video compression technology,the H.264standard might save about50%of the code rate, and thus favored the network transmission as well as local saves. With multimedia equipment performance increasing, people request video quality promoting too. People need clearer video signal, but the CPU basic frequency promotion meets the bottleneck, and CPU also changed the multiple-core direction to develop. In this aspect, GPU has its born that way superiority:formidable parallel processing ability and high efficiency’s data transmission ability. At the beginning of its design, it is to carry on the science computation to the video signal, with more cores on the hardware.This degree thesis unifies the CUDA programming characteristic, will move the CPU programs to GPU running. Because of CPU and the GPU construction is different, the programs cannot simple run with compiling, only then after the process optimizing the procedure can better display GPU potency. The construction of GPU decided that it suits running the large-scale parallel program, becoming the key of procedure transplant the conventional encoding method parallelization. Unifying X264source code, in addition optimizing the strategy and the CUDA compiler, may realize the new video frequency encoder.Finally, testing the performance of the encoder with experiments, the data obtained show that; the new Encoder can meet the requirements of real-time encoding of H.264HD.