Multi-GPU simulation of sediment transport
General-purpose GPUs offer tremendous computing power. While it is already difficult to program single-GPU codes, coupling multiple GPUs adds an additional challenge.
It is becoming more and more common that one desktop computer has several GPUs inside
- Aim of project: To implement a fully explicit numerical scheme for simulating sediment transport, capable of using multiple GPUs.
- Two programming approaches: (1) Use of MPI (via the host-CPU) for inter-GPU communication (2) Use of the latest programming features from CUDA Toolkit for interaction between multiple GPUs.
- Learning outcome: The master student will learn simple numerical strategies for sedimentation simulation, basic knowledge of parallel computing, CUDA programming, and use of multiple GPUs for scientific computing.