2013 | OriginalPaper | Buchkapitel
On the Effectiveness of Using the GPU for Numerical Solution of Stochastic Collection Equation
verfasst von : Nikita O. Raba, Elena N. Stankova
Erschienen in: Computational Science and Its Applications – ICCSA 2013
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Effective parallel algorithm is presented for numerical solution of stochastic collection equation on graphical processors. System of stochastic collection equations describes evolution of spectrum of drops and ice crystals due to the process of coalescence in the numerical models of natural convective clouds. They present the most computationally expensive part of such models aimed for forecasting dangerous weather phenomena such as thunderstorm, heavy rain and hail. Our results show that use of GPU can accelerate calculations in 15-20 times and thus provide operational forecast of such disaster events. Parallel algorithm is specially developed for calculations on GPU using CUDA technology. It is quadratic in time and allows using more than 1000 threads for effective parallelization. Special methods for optimization of memory access are also described.