Uma abordagem para a execução paralela de modelos de simulação

• Autor(a) principal: Mazzonetto, Angela
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UPF
• Texto Completo: https://repositorio.upf.br/handle/123456789/1755
• topic: Processamento paralelo (Computadores); Simulação (Computadores); Programação paralela (Computação); Parallel processing (computers); Simulation (Computers); Parallel programming (Computer); CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO

In the current scenario of science and technology, many factors contribute to the increased volume of data used in the solution of real applications in di_erent areas, including data used in simulation models of crops and diseases. The results generated by these models are analyzed after with intent to assist in the decision-making process. However, with the increased coverage area or with the expansion of the simulation period, the amount of data needed can often derail the execution and therefore the processing of the results generated by the models. Thus, the use high-performance tools or computational techniques designed for optimizing and improving the execution of these models becomes essential. This work speci_es an e_cient parallelization method for the parallel execution of growth models of wheat's culture CSM-Cropsim: Wheat and statistical Model Output Calibration correction, applied in the correction of weather forecast generated by the Eta model of the CPTEC-INPE, objects of this research. For these models the master-slave approach, using MPI communication library, proved the most appropriate, since each of the executions are independent in both models evaluated. The results showed the tests were satisfactory and contributed to the e_cient parallel execution of the models and consequently to the extension of their coverage areas and/or the time series to be simulated.