Research Article

Parallel Algorithms of Well-Balanced and Weighted Average Flux for Shallow Water Model Using CUDA

Table 3

Speedup compared to baseline program for various problems.

Rectangular dam breakCircular dam break on wet bedCircular dam break on dry bedDam break flows over three humps
16k1M16k1M16k1M16k1M

Serial: cell-basedBaselineBaselineBaselineBaselineBaselineBaselineBaselineBaseline
Serial: edge-based1.901.921.871.241.921.521.741.81
Parallel V112.4518.6720.9819.6619.0321.3913.7417.47
Parallel occupancy30.1852.2038.0439.7134.2343.8231.8743.58
Parallel memory pattern40.2452.5648.1940.2242.5544.1231.4944.02
Parallel unroll48.0763.3557.7249.3950.1953.6739.6450.60