Using genetic algorithms for solving partitioning problem in codesign

Partitioning problem in codesign is of critical importance since it has big impact on. cost/performance characteristics of the final product. It is an NP-Complete problem that deals with the different constraints relative to the system and the underlying target architecture. The reported partitioning approaches have several drawbacks (they axe often dedicated to a particular application or target architecture, they operate at a unique granularity level, most of them are manual and impossible to apply for complex systems, the number of constraints they deal with is generally limited...). This paper introduces an automatic approach using genetic algorithms to solve partitioning in codesign. This approach is totally independent of target architecture. Another advantage of this approach is that it allows determining dynamically the granularity of the objects to partition, making it possible to browse more efficiently solution space.