HPC (high perfomance computing) based on clusters of multicores is one of the main research lines in parallel programming. It is important to study the impact of programming paradigms of shared memory, message passing or a combination of both on these architectures in order to efficiently exploit the power of these architectures. The Smith-Waterman algorithm is used as study case for the local alignment of DNA sequences, which allows establishing the similarity degree between two sequences. In this paper, the Smith-Waterman algorithm is parallelized by means of a pipeline scheme due to the data dependencies that are inherent to the problem, using the various communication/synchronization models mentioned above and then carrying out a comparative analysis. Finally, experimental results are presented, as well as future research lines.Facultad de Informática