New computational developments on chemistry module of BRAMS numerical weather prediction

Roberto P. Souto, Maria E. S. Welter, Mateus S. Melo, Carla Osthoff, Renato J. P. Borseti, Luiz Flávio Rodrigues, Pedro Leite da Silva Dias, Franck Vigilant

Resumo


In this work, a restructuring of the calls of the direct model solver was carried out to remove redundant executions of the symbolic and numerical factorization stages present in the original implementation of the chemistry module from BRAMS regional numerical weather prediction. In the case study chosen, this change resulted in a gain of approximately 37.5% on average of the chemistry iteration serial time.


Palavras-chave


Numerical weather prediction, sparse linear systems, direct sparse method, LU factorization

Texto completo:

PDF


DOI: https://doi.org/10.5540/03.2018.006.02.0305

Apontamentos

  • Não há apontamentos.


SBMAC - Sociedade de Matemática Aplicada e Computacional
Edifício Medical Center - Rua Maestro João Seppe, nº. 900, 16º. andar - Sala 163 | São Carlos/SP - CEP: 13561-120
 


Normas para publicação | Contato