Benchmarking the scalability of MPI-based parallel solvers for fluid dynamics in low-budget cloud infrastructure

  • Vanderlei M. Pereira Filho UFSC
  • Márcio Castro UFSC

Resumo


This ongoing study presents an analysis of the scaling of MPI-enabled partial differential equations (PDE) systems solver implementations with public cloud infrastructure. Shallow-Water equations are used as case study for analysis. Results indicate that public cloud can be a cost-effective solution for highly coupled problems, given a certain problem size threshold and appropriate infrastructure configuration. However, adequate software tooling and models are required for estimating and dimensioning optimal clusters.

Referências

Netto, M. A. S., Calheiros, R. N., Rodrigues, E. R., Cunha, R. L., and Buyya, R. (2018). HPC Cloud for Scientific and Bussiness Applications: Taxonomy, Vision, and Research Challenges. volume 51. Association for Computing Machinery, New York, NY, USA.
Publicado
18/04/2022
PEREIRA FILHO, Vanderlei M.; CASTRO, Márcio. Benchmarking the scalability of MPI-based parallel solvers for fluid dynamics in low-budget cloud infrastructure. In: ESCOLA REGIONAL DE ALTO DESEMPENHO DA REGIÃO SUL (ERAD-RS), 22. , 2022, Curitiba. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2022 . p. 77-78. ISSN 2595-4164. DOI: https://doi.org/10.5753/eradrs.2022.19170.