Mapping Method Usable with Clustered Many-core Platforms for Simulink Model

Multi/many-core processors are being increasingly used to reduce power consumption and improve performance. The use of model-based development for embedded systems has also been increasing. Relative these trends, the parallelizer or MBP an essential role in parallelizing applications at model level. maps Simulink blocks cores using various types information such as block characteristics, C code, multi/many-core hardware implementation. However, does not consider many-core with cluster structures Kalray MPPA2-256 processor which contains 16 clusters 256 general-purpose total. This paper proposes algorithm that determines core allocations by considering structures. proposed combines two other algorithms: one uses allocation path analysis cluster-level considers effects from communication contention when determining allocations, results remaps allocations. produces better than its component algorithms could produce separately. Evaluations demonstrate obtained best among four methods terms execution time on models.