Many applications contain loops with an undetermined number of iterations. These loops have to be parallelized in order to increase the throughput when executed on an embedded multiprocessor platform.