Coordinated research of architecture driven code transformations for multimedia applications

1 January, 2002 to 31 December, 2005


  • Maurice Bruynooghe
  • Gerda Janssens

During the nineties, one of the most important evolutions within computer architecture was the spectacular development of digital consumer electronics: mobile telephony, electronic diaries, portable computers, game players, mp3-players, digital cameras and so on. Characteristic for all these devices is their low weight, their portability, their dependency on batteries, their wireless communication (very soon for all of them) and not at least their ever increasing functionality. Together with the pressure to shorten development time, the development of these products creates huge challenges to the development teams. Meeting the requested functionality (multimedia, wireless communication, ...) requires hundreds of thousands of lines of code (mpeg decoders, speech recognition, protocol stacks), many of them to be executed in real time. This requires heavy processors consuming lots of battery power, leading to expensive, large and heavy devices with limited autonomy. However, the consumer expects autonomy spans a working day so that batteries need to be reloaded only at night. This is definitely out of reach for the current generation of portable PC's. One solution is to wait for the next generation which can be expected to have circuits with higher density and lower power consumption. However, the requested functionality tends to increase faster than the power of the processors notwithstanding Moore's law predicts a 60 percent increase a year. So, to meet better the consumers expectations ---and to obtain a competitive advantage-- developers should also consider the optimisation of the software. This constitutes the goal of this project.