Parallelization of Harmonic Progression Sum with OpenMP and CUDA

  • Pedro C. Coleone UFSCar
  • Pietro M. Moralles UFSCar
  • Carlos H. R. Matos Filho UFSCar
  • Hélio C. Guardia UFSCar
DOI: https://doi.org/10.5753/eradsp.2024.239920

Abstract


This article examines parallel implementations of the harmonic progression sum problem using OpenMP in a multiprocessor shared memory environment and CUDA for GPU processing. The performance of these implementations is evaluated through speedup and efficiency metrics, highlighting runtime gains that demonstrate the importance of parallelization techniques. Calculation partitioning and partial result aggregation were essential to achieve good speedup, especially with the GPU, by avoiding unnecessary data copies between GPU memory and RAM.

References

Documentação oficial do CUDA Toolkit. NVIDIA.

Documentação oficial do OpenMP. OpenMP Architecture Review Board.

Bianchini, C. and Pillon, M. A. (2021). 16th marathon of parallel programming: Rules for remote contest. In Proceedings of the 16th Marathon of Parallel Programming, pages 1–1, Brazil. SBAC-PAD & WSCAD.

Knuth, D. E. (1968). The Art of Computer Programming, Volume 1: Fundamental Algorithms. Addison-Wesley.
Published
2024-05-16
How to Cite

Select a Format
COLEONE, Pedro C.; MORALLES, Pietro M.; MATOS FILHO, Carlos H. R.; GUARDIA, Hélio C.. Parallelization of Harmonic Progression Sum with OpenMP and CUDA. In: REGIONAL SCHOOL OF HIGH PERFORMANCE COMPUTING FROM SÃO PAULO (ERAD-SP), 15. , 2024, Rio Claro/SP. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2024 . p. 13-16. DOI: https://doi.org/10.5753/eradsp.2024.239920.