The Multi-objective Design Space Exploration of Multiprocessor SOC Architectures for Embedded Multimedia Applications (MULTICUBE, January 2008 to June 2010) project aims at increasing the competitiveness of European industries by optimizing the design of embedded computing systems while reducing design time and costs. The project defined an automatic multi-objective DSE framework to be used at design-time to find the best power/performance trade-offs while meeting system-level constraints and speeding up the exploration process. A set of heuristic optimization algorithms have been defined to reduce the exploration time, while a set of response surface modeling techniques have been defined to further speed up the process.
Based on the results of the design-time multi-objective exploration, the MULTICUBE project also defined a methodology to be used at run-time to optimize the run-time allocation and scheduling of different application tasks. The design exploration flow results in a Pareto-optimal set of design alternatives in terms of power/performance trade-offs. The set of operating points can then be used at run-time to decide how the system resources should be distributed over different tasks running on the multiprocessor system on chip.
The MULTICUBE DSE framework leverages a set of open-source and proprietary exploration, modeling and simulation tools to guarantee a wide exploitation of the MULTICUBE project results in the embedded system design community. The MULTICUBE project has been strongly industry-driven: industrial partners (STMicroelectronics and DS2) have defined the design tools requirements and then validated the tools to design some industrial use cases.More information on the project is provided in the Project summary: a two-page Executive Summary followed by a description of Exploitable Results of the project.
Please find some time to read it and then just a few more minutes to fill in the feedback form.