Skip to Main Content
Consumer electronic products are increasingly becoming more open and flexible, which is achieved by replacing dedicate, single-function hardware components by software components running on programmable platforms. Scalable video algorithms (SVAs) using the novel principle of priority processing can guarantee real-time performance on these platforms even with limited resources. In this paper we describe the mapping of a priority-processing application on an embedded consumer platform comprising a general purpose processor and multiple stream processing elements. An application comprises one or more independent, competing priority-processing algorithms. These SVAs have a single, dedicated streaming processor at their disposal. Dynamic resource allocation is required to maximize the overall output quality of SVAs that are executed on a shared platform. To enable real-time processing of individual SVAs, we compare the performance of different implementations for dynamic-resource-allocation mechanisms. Finally, we show that priority processing achieves real-time performance even under tight resource constraints.