System-level power optimizing data-flow transformations for multimedia applications realized on programmable multimedia processors

Abstract

In this paper a methodology is presented for the application of a class of power optimizing code transformations based on data flow modifications. The target domain consists of multimedia applications realized on (embedded) programmable processors. In this important context, the application of data flow transformations is a crucial step to reduce the power consumption due to data storage and transfers that forms a significant part of the total power budget of the system. Performance issues are also taken into consideration. The proposed methodology forms a part of a complete methodology for the performance conscious application of power optimizing code transformations in a programmable processor context. The usefulness of the proposed methodology is demonstrated using a complex real-life video processing application.