Analisando o Impacto do DVFS no Desempenho e Energia de Aplicações Paralelas em GPUs
Resumo
A capacidade de processamento paralelo de unidades de processamento gráfico (GPUs) tornou essencial seu uso em aceleração de aplicações de inteligência artificial. Com forte presença de multiplicação de matrizes nessas aplicações, novas estratégias são necessárias para obter melhor eficiência energética. Dessa maneira, analisamos o impacto de métricas da cache em uma GPU e mostramos uma diferença de 19,87% em gasto de energia com pequenos ganhos no desempenho.
Palavras-chave:
Algoritmos Paralelos e Distribuídos, Avaliação, Medição e Predição de Desempenho, Computação Heterogênea
Referências
Anzt, H., Tomov, S., and Dongarra, J. (2015). Energy efficiency and performance frontiers for sparse computations on gpu supercomputers. In Proceedings of the Sixth International Workshop on Programming Models and Applications for Multicores and Manycores, PMAM ’15, page 1–10, New York, NY, USA. Association for Computing Machinery.
Baji, T. (2017). GPU: the biggest key processor for AI and parallel processing. In Takehisa, K., editor, Symposium on Photomask and Next-Generation Lithography Mask Technology, volume 10454, page 1045406. International Society for Optics and Photonics, SPIE.
Fei, X., Li, K., Yang, W., and Li, K. (2020). Analysis of energy efficiency of a parallel aes algorithm for cpu-gpu heterogeneous platforms. Parallel Computing, 94-95:102621.
Jahanshahi, A., Sabzi, H. Z., Lau, C., and Wong, D. (2020). Gpu-nest: Characterizing energy efficiency of multi-gpu inference servers. IEEE Computer Architecture Letters, 19(2):139–142.
Jin, Z. and Vetter, J. S. (2023). A benchmark suite for improving performance portability of the sycl programming model. In 2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS), pages 325–327.
Li, D., Chen, X., Becchi, M., and Zong, Z. (2016). Evaluating the energy efficiency of deep convolutional neural networks on cpus and gpus. In IEEE international conferences on big data and cloud computing, pages 477–484. IEEE.
Sharma, R., M, V., and Moharir, M. (2016). Revolutionizing machine learning algorithms using gpus. In CSITSS, pages 318–323.
Wang, Y., Karimi, M., Xiang, Y., and Kim, H. (2021). Balancing energy efficiency and real-time performance in gpu scheduling. In 2021 IEEE Real-Time Systems Symposium (RTSS), pages 110–122.
Baji, T. (2017). GPU: the biggest key processor for AI and parallel processing. In Takehisa, K., editor, Symposium on Photomask and Next-Generation Lithography Mask Technology, volume 10454, page 1045406. International Society for Optics and Photonics, SPIE.
Fei, X., Li, K., Yang, W., and Li, K. (2020). Analysis of energy efficiency of a parallel aes algorithm for cpu-gpu heterogeneous platforms. Parallel Computing, 94-95:102621.
Jahanshahi, A., Sabzi, H. Z., Lau, C., and Wong, D. (2020). Gpu-nest: Characterizing energy efficiency of multi-gpu inference servers. IEEE Computer Architecture Letters, 19(2):139–142.
Jin, Z. and Vetter, J. S. (2023). A benchmark suite for improving performance portability of the sycl programming model. In 2023 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS), pages 325–327.
Li, D., Chen, X., Becchi, M., and Zong, Z. (2016). Evaluating the energy efficiency of deep convolutional neural networks on cpus and gpus. In IEEE international conferences on big data and cloud computing, pages 477–484. IEEE.
Sharma, R., M, V., and Moharir, M. (2016). Revolutionizing machine learning algorithms using gpus. In CSITSS, pages 318–323.
Wang, Y., Karimi, M., Xiang, Y., and Kim, H. (2021). Balancing energy efficiency and real-time performance in gpu scheduling. In 2021 IEEE Real-Time Systems Symposium (RTSS), pages 110–122.
Publicado
23/04/2025
Como Citar
GONÇALVES, Thiago dos S.; LORENZON, Arthur F..
Analisando o Impacto do DVFS no Desempenho e Energia de Aplicações Paralelas em GPUs. In: ESCOLA REGIONAL DE ALTO DESEMPENHO DA REGIÃO SUL (ERAD-RS), 25. , 2025, Foz do Iguaçu/PR.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2025
.
p. 17-20.
ISSN 2595-4164.
DOI: https://doi.org/10.5753/eradrs.2025.6824.
