In this paper we present a unique approach towards parallelizing the Software MPEG-2 Video Encoder. The approach involves the division of encoding process spatially, between four similar processors with distributed memory system, in such a way that each frame of the input video sequence is divided into four separate sections, each individually processed by a processor. The results thus produced were compared with those of the Sequential Encoder for a number of different input bitstreams and a speedup in the range of 3 to 3.5 with four processors was successfully obtained.
展开▼