The Sony-Toshiba-IBM Cell Broadband Engine is a heterogeneous multicore architecture that consists of a traditional microprocessor (PPE) with eight SIMD coprocessing units (SPEs) integrated on-chip. While the Cell/B.E. processor is designed with multimedia applications in mind, there are currently no open-source, optimized implementations of such applications available. In this paper, we present the design and implementation behind the creation of an optimized MPEG-2 software decoder for this unique parallel architecture, and demonstrate its performance through an experimental study. This is the first parallelization of an MPEG-2 decoder for a commodity heterogeneous multicore processor such as the IBM Cell/B.E. While Drake et al. have recently parallelized MPEG-2 using StreamIt for a streaming architecture, our algorithm is quite different and is the first to address the new challenges related to the optimization and tuning of a multicore algorithm with DMA transfers and local stor...
David A. Bader, Sulabh Patel