SCIARA is a deterministic Macroscopic Cellular Automata model for simulating lava flows.
In this paper we presents a CUDA parallelization of the deterministic Macroscopic Cellular Automata model SCIARA for simulating lavaflows. Furthermore, we present the results of an experimental analysiswe conducted in order to evaluate performance-wise impact. Specifically, given benchmarks of our parallel implementation with respect toserial algorithm, we observe how parallel implementation is very com-petitive and even faster.
📚 Report: HERE
To run Sciara-fv2 CUDA backend:
$ make run-$VERSION
# Where $VERSION can be 'cuda-straightforward', 'cuda-tiled-halo', 'cuda-tiled-no-halo', 'cuda-multi-gpu'NOTE: Building and running requires CUDA compatible devices with 5.2 or higher compute capabilities.
 |
 |
 |
 |
 |
Sciara-fv2 CUDA uses and depends on third-party open-source tools and libraries which are outside of this repository.
Copyright (c) Ransome CA. All rights reserved.
Licensed under the GPL-3.0 license.