Modelos Paralelos para la Resolución de Problemas de Ingeniería Agrícola

Abstract

This work is in the field of parallel computing and, in particular, in the development and use of computational models in heterogeneous parallel architectures to solve applied problems. We address a number of problems that are related to the application of technology in the field of agricultural engineering. The problems are: landform representation, processing climate information (for example, temperature), and water resources management. The study and the solution of these problems in the geographical area in question may suppose an important economic and environmental impact. The problems are formulated in a mathematical model whose solution is computationally costly, and sometimes the solution in a reasonable time is not possible. Our work deals with the implementation of fast and efficient parallel algorithms to solve the associated mathematical problems in multicore and multi-GPU nodes. In addition, some techniques are proposed, studied and implemented to automatically adapt the routines to the parallel system where they are installed, with the aim of obtaining executions close to the optimum and at a low computational cost. The objective is to provide the users with portable software able to run efficiently on the target computer, regardless of the characteristics of the architecture and the user knowledge of it.

Algoritmos para agricultura de precisión utilizando computación de alto rendimiento

“…Las unidades de procesamiento gráfico (GPU) es un tipo de hardware nacido para mejorar el rendimiento de los juegos por computador. Con el avance tecnológico las GPU han evolucionado para ser utilizados en otras áreas incluida la agricultura de precisión (Murilo, 2014). Por su estructura computacional, las GPU pueden ofrecer un rendimiento superior en comparación con las clásicas Universidad Nacional de La Plata | Marco Pusdá-Chulde | Doctorado en Ciencias Informáticas unidades centrales de proceso (CPU) (HajiRassouliha et al, 2018) en aplicaciones que requieran grandes cargas de trabajo computacional.…”