Revisiting the parallel programming

• Autores: Aslla, A. P., Montalvo, R. A., Rivera, P. H.
• Formato: artículo
• Fecha de Publicación: 2011
• Institución: Universidad Nacional Mayor de San Marcos
• Repositorio: Revistas - Universidad Nacional Mayor de San Marcos
• Lenguaje: español
• OAI Identifier: oai:ojs.csi.unmsm:article/8540
• Enlace del recurso: https://revistasinvestigacion.unmsm.edu.pe/index.php/fisica/article/view/8540
• Nivel de acceso: acceso abierto
• Materia: OpenMPI; OpenMP; Cell BE; i7 980X; CUDA.

In the last decade the multicore processors had emerged with the comeback of the symmetric multiprocessing (SMP) after a long decade in which the based OpenMPI clusters had dominated the high performace processing world. Actually, we have in the market 2, 4 and 6 multicore processors; the Cell BE processors with a Power Processor Element (PPE) and 8 vectorial processors called Synergistic Processor Element (SPE), and graphics processors which have from 48 to 448 cores dedicated to numerical tasks. In the present work, we show our experience in the interaction with these processors, discussing the programming techniques and the future possibilities for making low cost numerical tasks.