An Approach for Algorithm Parallelization Oriented to a Many-core Implementation

The main goal of this work is to present a parallelization methodology oriented to shared memory many-core platforms. There are two important elements in the proposed method that facilitate the design process. One is the use of OpenMP as a programming paradigm since it is oriented to a shared memory environment. The other is the use of a virtual platform that enables the evaluation of performance and task execution time before a real hardware prototype is available and also helps in the design space exploration of generalized and configurable target platforms.