摘要
DVCPRO HD标准广泛应用于广播电影电视前端的视频压缩,在普通的PC平台上每帧解码耗时需80到100ms,经过汇编优化后城需40到60ms,在优化比较好的情况下能达到实时解码播放的需求。GPU具有强大的并行处理能力,而DVCPRO HD的视频单元间耦合性小,适应于并行处理。充分利用GPU的并行性和强大的浮点运算功能,在CUDA开发环境下对其进行并行解码优化,可远远超过实时解码的要求。
DVCPRO HD standard is widely used in the foreside of movie and television, it takes about 80 to 100 milliseconds to decode a frame on a normal PC platform, and also takes 40 to 60 milliseconds after assembly optimization, which can decode and broadcast real-time with excellent commercial decoders. Parallel processing and powerful floating-point operations of GPU are fully used in this article, and after optimization under the CUDA environment, a much higher frame rate is reached.
出处
《信息技术》
2009年第7期72-74,共3页
Information Technology
