Mapping applications onto cluster-based MPSOCS

• Ano de defesa: 2014
• Autor(a) principal: Longhi, Oliver Bellaver
• Orientador(a): Hessel, Fabiano Passuelo
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Pontifícia Universidade Católica do Rio Grande do Sul; Porto Alegre
• Palavras-chave em Português: INFORMÁTICA; ARQUITETURA DE COMPUTADOR; MULTIPROCESSADORES
• Link de acesso: http://hdl.handle.net/10923/5829

The industry for decades has increased the clock rate to answer the need of performance. Reaching a physical limitations in terms of heat, the new chosen axis to increase performance is to scale the number of processing elements. To deal with that scaling number of processing elements, more and more important are the methodologies to support the design of MPSoCs. Approaches like simulation and FPGA-based prototyping are too expensive and timing consuming. Therefore, techniques like Analytical Models represent important alternatives to the previous consuming approaches. However, these architecture models are difficult to build and characterize. In addition, emerging MPSoC topologies lack analytical models. Due to that, this work proposes an analytical model to support designers in common tasks of the design process like application mapping and prototypes generation.