A GPU parallelized finite volume scheme which solves the two dimensional Shallow Water Equations with multi-resolution is presented. Different levels of resolution are introduced by means of a novel type of grid called Block-Uniform Quadtree (BUQ), which is able to exploit the computational capability of GPUs with negligible overheads, allowing at the same time to reproduce small scale effects. The model has been tested by simulating floods propagation on a 38 km long reach of the Parma River (Italy) with a remarkable reduction of computational times and allocated memory, if compared with a Cartesian high-resolution grid.
Efficient non-uniform grid for GPU-parallel shallow water equations models / Vacondio, Renato; Ferrari, Alessia; Mignosa, Paolo; Aureli, Francesca; DAL PALU', Alessandro. - (2016), pp. 281-288. (Intervento presentato al convegno International Conference on Fluvial Hydraulics, RIVER FLOW 2016 tenutosi a usa nel 2016).
Efficient non-uniform grid for GPU-parallel shallow water equations models
VACONDIO, Renato;FERRARI, ALESSIA;MIGNOSA, Paolo;AURELI, Francesca;DAL PALU', Alessandro
2016-01-01
Abstract
A GPU parallelized finite volume scheme which solves the two dimensional Shallow Water Equations with multi-resolution is presented. Different levels of resolution are introduced by means of a novel type of grid called Block-Uniform Quadtree (BUQ), which is able to exploit the computational capability of GPUs with negligible overheads, allowing at the same time to reproduce small scale effects. The model has been tested by simulating floods propagation on a 38 km long reach of the Parma River (Italy) with a remarkable reduction of computational times and allocated memory, if compared with a Cartesian high-resolution grid.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
