A Multi-GPU Program for Uncertainty-Aware Drainage Basin Delineation: Scalability benchmarking with country-wide data sets

Abstract

Processing high-resolution digital elevation models (DEMs) can be tedious due to the large size of the data. In uncertainty-aware drainage basin delineation, we apply a Monte Carlo (MC) simulation that further increases the processing demand by two to three orders of magnitude. Utilizing graphics processing units (GPUs) can speed up the programs, but their on-chip random access memory (RAM) limits the size of the DEMs that can be processed efficiently on one GPU. Here, we present a parallel uncertainty-aware drainage basin delineation algorithm and a multinode GPU compute unified device architecture (CUDA) implementation along with scalability benchmarking. All of the computations are run on the GPUs, and the parallel processes communicate using a message-passing interface (MPI) via the host central processing units (CPUs). The implementation can utilize any number of nodes, with one or many GPUs per node. The performance and scalability of the program have been tested with a 10-m DEM covering 390,905 km2, i.e., the entire area of Finland. Performing the drainage basin delineation for the DEM with different numbers of GPUs shows a nearly linear strong scalability.