Integration of Mpeg-4 video tools onto multi-DSP architectures using AVSynDEx fast prototyping methodology

Mpeg-4 is a response to the growing need for coding method that can facilitate access to visual objects in natural and synthetic moving pictures. Future real time audio-visual applications using Mpeg-4 will have very important time constraints, that can be achieved with the use of several calculation units. Sequential software solutions actually developed for single processors can hardly be projected onto multiprocessor architectures, leading to extra load of source code and calculations, but also to a sub-optimal use of the architecture parallelism. A functional data flow description of the application is then a well suited front-end for optimal multi-components implementation. This paper presents an Mpeg-4 decoder with such description formalism, allowing incremental building, and easy handing-over up to date of the algorithms. Furthermore, we show that the use of our AVSynDEx methodology enables its optimized implementation onto a multi-C6X platform.