The main purpose of this master’s thesis is the development of a multi-GPU version of the CUTEFLOW finite volume solver for the Shallow Water Equations (SWE). We use both MPI and CUDA-Fortran to use as many GPUs as we need. The METIS library is used in order to do a domain decomposition on the 2D unstructured triangular meshes. A CUDA-Aware OpenMPI version is used in order to speed up the messages between the MPI processes. Finally, a study of both speed-up and efficiency is carried over first, a classic onedimensional Dam-Break case, and then the Mille Îles river. In both cases, meshes with up to 13 million cells are used. Using 24 to 28 GPUs on these meshes lead to more than 80% efficacy.
| Date | 25 Aug 2020 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Azzeddine Soulaïmani (Supervisor) |
|---|
Delmas, V. (Author),
Soulaïmani (Supervisor),
25 Aug 2020Student thesis: Master's thesis › Master in Engineering: Mechanical Engineering