设计了一种语法元素指令流驱动的全流水CABAC(Context-based Adaptive Binary Arithmetic Coding)熵编码VLSI结构,并对提出的语法元素级分组并行算术编码器的体系结构进行了设计和开销评估.该并行方法可以与现有符号级并行算法正交,可同时使用,适合大规模片上并行视频编码器;相比标准CABAC,增加约55%的晶体管即可实现2倍以上的符号处理加速比和>1Gbin/s的吞吐率.%This paper proposes a new CABAC pipeline driving by syntax element instructions and a GABAC-based parallel arithmetic entropy coder for on-chip large-scale parallel H.264 video coders is also presented in this paper. Furthermore, the hardware architecture and transistor cost of the new parallel entropy coder are evaluated. This new parallel entropy coder can cooperate with the traditional symbol-level parallel algorithms and it suits the manycore platform well.Compared with the traditional CABAC hardware,the new parallel entropy coder can double its throughput to > 1Gbins/s by a cost of 55% transistors.
展开▼