The design of future systems on chips requires a scalable hardware-software design approach.
In the SHAPES
(scalable software hardware architecture platform for embedded systems) project,
a new hardware-software architecture paradigm is investigated.
Within the SHAPES project, we are dealing with the challenge of providing an environment for
programming parallel applications and mapping them to heterogeneous distributed architectures.
In principle, we try to tackle this challenge by:
- leveraging an appropriate model of computation, namely Kahn process networks,
- using a dedicated application programming interface, and
- applying appropriate methods for mapping optimization.
Being under steady development, the current environment consists of the following parts:
Distributed Operation Layer:
The distributed operation layer (DOL) is a
software development framework to program parallel applications. The DOL allows
to specify applications based on the Kahn process network model of computation and
features a simulation engine based on SystemC.
Moreover, the DOL provides an XML-based specification format to describe the
implementation of a parallel application on a multi-processor systems, including
binding and mapping.
Moses: Moses (modeling, simulation, and evaluation of systems) is a
graphical front-end to the DOL and allows to "draw" rather than write parts of an
Modular Performance Analysis:
For real-time analysis of multi-processor systems, DOL specifcations can be
converted into models for modular performance analysis (MPA).
Within the SHAPES project, our environment is used to develop streaming applications from such diverse
areas as medical imaging, audio wave field synthesis, and physical modeling of complex systems.
This webpage is no longer maintained after the end of the SHAPES project in 2009. For further information please contact Prof. Dr. Lothar Thiele.