A cost-benefit analysis of GPU-based EC2 instances for a deep learning algorithm
Resumo
This paper analyzes the cost-benefit of using EC2 instances, specifically the p2 and p3 virtual machine types, which have GPU accelerators, to execute a machine learning algorithm. This analysis includes the runtime of a convolutional neural network executions, and it takes into consideration the necessary time to stabilize the accuracy value with different batch sizes. Also, we measure the cost of using each machine type, and we define a relation be- tween this cost and the execution time for each virtual machine. The results show that, although the price per hour of the p3 instance is three times bigger, it is faster and costs almost the same as the p2 instance type to train the deep learning algorithm.
Palavras-chave:
Arquiteturas Dedicadas e Específicas (GPUs, FPGAs e outras), Computação de Alto Desempenho, Computação em Nuvem, Grade, Aglomerado (Cluster) e Peer-to-Peer
Referências
Ben-Nun, T. and Hoefler, T. (2018). Demystifying parallel and distributed deep learning: An in-depth concurrency analysis. arXiv preprint arXiv:1802.09941.
He, K., Zhang, X., Ren, S., and Sun, J. (2016). Deep residual learning for image recogni- tion. In CVPR, pages 770–778. IEEE. Kaplunovich, A. and Yesha, Y. (2017). Cloud big data decision support system for ma- chine learning on AWS: Analytics of analytics. In Big Data, pages 3508–3516. IEEE.
Keskar, N. S., Mudigere, D., Nocedal, J., Smelyanskiy, M., and Tang, P. T. P. (2016). On large-batch training for deep learning: Generalization gap and sharp minima. arXiv preprint arXiv:1609.04836.
Krizhevsky, A., Nair, V., and Hinton, G. (2010). CIFAR-10 (Canadian Institute for Ad- vanced Research). http://www.cs.toronto.edu/ ̃kriz/cifar.html.
LeCun, Y., Boser, B., Denker, J., Howard, R., Habbard, W., Jackel, L., and Henderson, D. (1990). Handwritten digit recognition with a back-propagation network. In NIPS, pages 396–404.
Lee, K. and Son, M. (2017). DeepSpotCloud: leveraging cross-region gpu spot instances for deep learning. In CLOUD, pages 98–105. IEEE.
Miller, F. P., Vandome, A. F., and McBrewster, J. (2010). Amazon Web Services. Alpha Press.
He, K., Zhang, X., Ren, S., and Sun, J. (2016). Deep residual learning for image recogni- tion. In CVPR, pages 770–778. IEEE. Kaplunovich, A. and Yesha, Y. (2017). Cloud big data decision support system for ma- chine learning on AWS: Analytics of analytics. In Big Data, pages 3508–3516. IEEE.
Keskar, N. S., Mudigere, D., Nocedal, J., Smelyanskiy, M., and Tang, P. T. P. (2016). On large-batch training for deep learning: Generalization gap and sharp minima. arXiv preprint arXiv:1609.04836.
Krizhevsky, A., Nair, V., and Hinton, G. (2010). CIFAR-10 (Canadian Institute for Ad- vanced Research). http://www.cs.toronto.edu/ ̃kriz/cifar.html.
LeCun, Y., Boser, B., Denker, J., Howard, R., Habbard, W., Jackel, L., and Henderson, D. (1990). Handwritten digit recognition with a back-propagation network. In NIPS, pages 396–404.
Lee, K. and Son, M. (2017). DeepSpotCloud: leveraging cross-region gpu spot instances for deep learning. In CLOUD, pages 98–105. IEEE.
Miller, F. P., Vandome, A. F., and McBrewster, J. (2010). Amazon Web Services. Alpha Press.
Publicado
12/04/2019
Como Citar
MALTA, Eva; RODAMILANS, Charles; AVILA, Sandra; BORIN, Edson.
A cost-benefit analysis of GPU-based EC2 instances for a deep learning algorithm . In: ESCOLA REGIONAL DE ALTO DESEMPENHO DE SÃO PAULO (ERAD-SP), 10. , 2019, Campinas.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2019
.
p. 21-24.
DOI: https://doi.org/10.5753/eradsp.2019.13588.
