|Speaker||Dr. Christian Haubelt
University of Erlangen-Nuremberg, Germany
|Location||Engineering Tower, Room 331, University of California, Irvine|
|Date & Time||June 13 , 2008
Talk begins 11:00am
Refreshments at 10:30am
|Abstract||80% of all design decisions are made in the first 20% of the design time. Thus, a substantiated knowledge about possible solutions is mandatory in early design phases in order to design high quality products. SystemCoDesigner is a software tool for automatic design space exploration at the electronic system level and for platform-based prototyping of hardware/software systems. By exploring the design space, a designer becomes more confident in decisions to be done. Starting from a behavioral SystemC description different solutions are generated automatically by using software synthesis techniques and Forte Design Systems’ Cynthesizer for behavioral hardware synthesis. In order to allow such a degree of automation, SystemCoDesigner requires the behavioral SystemC model to be representable in a well-defined Model of Computation, called FunState. As a case study, a model of a Motion-JPEG decoder was automatically optimized and implemented using SystemCoDesigner. Several synthesized SoC variants based on this model show different tradeoffs between required hardware costs and achieved system throughput, ranging from software only solutions up to pure hardware implementations which reach real-time performance for QCIF streams on a 50MHz FPGA. In this talk, the SystemCoDesigner overall design flow, the FunState Model of Computation, and some automatic transformation are presented using the Motion-JPEG case study.|
Christian Haubelt received his diploma degree in Electrical Engineering from the University of Paderborn, Germany, in 2001. He received his Ph.D. in Computer Science from the University of Erlangen-Nuremberg, Germany, in 2005. Since 2004, he is head of the System-Level Design Automation Group at the Chair of Hardware/Software Co-Design, University of Erlangen-Nuremberg, Germany. Dr. Haubelt’s research interests include electronic system level design, system level design automation, and multi-objective optimization.