Implementation and improvement of Wavefront Parallel Processing for HEVC encoding on many-core platform

摘要

In this paper, Wavefront Parallel Processing (WPP) is implemented effectively for real-time HEVC encoder on many-core platform, where ping-pang storage, data reuse and try-and-wait mechanism are designed to optimize data storage and communication among cores. Parallel scalability of WPP is analyzed and a mathematic model of WPP is presented. In addition, three effective approaches are presented to improve the parallel scalability of WPP. Experimental results show that the model is valuable and the proposed methods improve more than 40% maximum parallel scalability.