Detail publikace

Effective Mapping of Grammatical Evolution to CUDA Hardware Model

POSPÍCHAL, P. SCHWARZ, J.

Originální název

Effective Mapping of Grammatical Evolution to CUDA Hardware Model

Typ

článek ve sborníku mimo WoS a Scopus

Jazyk

angličtina

Originální abstrakt

Several papers have shown that symbolic regression is suitable for data analysis and prediction in finance markets. The Grammatical Evolution (GE) has been successfully applied in solvingvarious tasks including symbolic regression. However, performance of this method can limit the areaof possible applications. This paper deals with utilizing mainstream graphics processing unit (GPU)for acceleration of GE solving symbolic regression. With respect to various mentioned constrains,such as PCI-Express and main memory bandwidth bottleneck, we have designed effective mappingof the algorithm to the CUDA framework. Results indicate that for larger number of regression pointscan our algorithm run 636 or 39 times faster than GEVA library routine or a sequential C code, respectively. As a result, the ordinary GPU, if used properly, can offer interesting performance boostfor solution the symbolic regression by the GE.

Klíčová slova

GPU, Graphics Processing Units, Grammatical Evolution, CUDA, Symbolic Regression,Speedup, C

Autoři

POSPÍCHAL, P.; SCHWARZ, J.

Rok RIV

2011

Vydáno

28. 4. 2011

Nakladatel

Brno University of Technology

Místo

Brno

ISBN

978-80-214-4273-3

Kniha

Proceedings of the 17th Conference Student EEICT 2011 Volume 3

Strany od

574

Strany do

578

Strany počet

5

URL

https://www.fit.vut.cz/research/publication/9595/

BibTex

@inproceedings{BUT76335,
  author="Petr {Pospíchal} and Josef {Schwarz}",
  title="Effective Mapping of Grammatical Evolution to CUDA Hardware Model",
  booktitle="Proceedings of the 17th Conference Student EEICT 2011 Volume 3",
  year="2011",
  pages="574--578",
  publisher="Brno University of Technology",
  address="Brno",
  isbn="978-80-214-4273-3",
  url="https://www.fit.vut.cz/research/publication/9595/"
}