Popis: |
Software implementation of compute-intensive applications in digital signal processing requires large computing power and has real-time performance requirements. Employing multicore architecture is usually the only means for solving the grand challenge of computational problems. Developing multicore-based systems requires a high degree of concurrency for optimizing performances of systems. For this purpose, this paper addresses the redesigning of MATLAB\Simulink models for efficient concurrent implementation using multiple processors. Our approach consists of translating a Simiuink model into discrete synchronous dataflow graph in order to treat them as concurrent system by exploiting task-level parallelism without alter the input-output behavior of the system. |