Home News People Research Study Search

Institute for Computing Systems Architecture

Eccentric SoC architectures as the future norm

Gordon Brebner

Xilinx Research Labs

PLEASE NOTE USUAL DAY
FRIDAY 10 OCTOBER 2003
JCMB, ROOM 2511
3.30 P.M.

Reconfigurable system-on-chip (SoC) platforms are now a physical reality. For example, the Xilinx Virtex-II Pro family offers a combination of programmable logic, embedded processors, embedded memory and specialized ASIC blocks, in a sea of programmable interconnect. This gives a system substrate which, physically at least, is fairly neutral with respect to use models and system architectures. As an intermediate step in system design, it is usually helpful to envisage domain-specific soft platforms provided on the physical fabric, as bases for application-specific system implementation. Assumptions built into such platforms may have a radical impact on the quality and performance of implemented systems.

When embedded processors are present, the most obvious use model to follow is to consider the system-on-chip as a scaled-down version of a conventional processor/memory/peripheral-device architecture: we call this a "processor-centric" model. Particular side-effects of this model include extensive use of buses in the architecture and general sequentialization of processing. We feel that while this can offer immediate routes to implementation, the adherence to fifty-plus years of computing tradition can obscure many of the benefits of the new SoC devices. Particular benefits arise through concurrency, topology and programmability, and the challenge is how to make these available to applications. Therefore, we shall discuss an alternative "interface-centric" model, where the environment of a system is the driving force behind its architecture. In particular, the role of embedded processors is just to assist the majority processing being carried out in logic and in input/output interfaces. We see such "eccentric" architectures as the norm for the future, particularly given the vision of the "disappearing computer" and the rise of "ambient intelligence".

We shall illustrate the general discussion with examples drawn from our current research into systems designed for "message processing" - a term we use for network/packet processing occurring in a wide range of potential applications, with widely-differing quality and performance requirements. Such systems represent an intermediate point between two well-known models - signal processing and data processing - that seems well-suited to interface-centric architecture.


Home : Colloquium 

Please contact our webadmin with any comments or changes.
Unless explicitly stated otherwise, all material is copyright © The University of Edinburgh.