TABLE OF CONTENTS Claudio Brunelli and Jari Nurmi
Tampere University of Technology
In this chapter, we describe a coprocessor approach to accelerating applications. We are concentrating on multimedia applications as an illustrative example of where this approach can be applied. In everyday life we all deal with a number of integrated digital systems which are present into almost all the products for consumer electronics. The heart of those systems is usually a microprocessor core, which is mainly used to control all the other components.
Many of the items that feature a microprocessor inside them are handheld devices, thus posing severe constraints both on the area available and on the energy budget and power dissipation. Despite of this, the will for conveniently including more and more complex and smart functionalities even on simplest devices has been pushing system designers to consider new ways of empowering their architectures with special solutions to guarantee the ability of executing the required functionalities while respecting the posed constraints.
In particular, users ask for their portable devices (PDAs) to support heavy multimedia and 3D graphics applications, besides other ones which are commonly referred to as general-purpose computations [213].
Besides others which need general-purpose computations, multimedia applications need to squeeze all the possible performance out of the computation system, and standard RISC architectures can hardly provide the necessary computational capabilities. For this reason, computer architectures shifted from the usage of conventional microprocessors to entire Systems on a Chip (SoC). Inside SoCs there is usually a main...
