hrcak mascot   Srce   HID

Izvorni znanstveni članak
https://doi.org/10.32985/ijeces.11.1.1

Performance Assessment of Deep Learning Frameworks through Metrics of CPU Hardware Exploitation on an Embedded Platform

Delia Velasco-Montero   ORCID icon orcid.org/0000-0003-3487-1712 ; Instituto de Microelectrónica de Sevilla, Universidad de Sevilla-CSIC
Jorge Fernández-Berni ; Instituto de Microelectrónica de Sevilla, Universidad de Sevilla-CSIC
Ricardo Carmona-Galán ; Instituto de Microelectrónica de Sevilla, Universidad de Sevilla-CSIC
Ángel Rodríguez-Vázquez ; Instituto de Microelectrónica de Sevilla, Universidad de Sevilla-CSIC

Puni tekst: engleski, pdf (1 MB) str. 1-11 preuzimanja: 127* citiraj
APA 6th Edition
Velasco-Montero, D., Fernández-Berni, J., Carmona-Galán, R. i Rodríguez-Vázquez, Á. (2020). Performance Assessment of Deep Learning Frameworks through Metrics of CPU Hardware Exploitation on an Embedded Platform. International journal of electrical and computer engineering systems, 11 (1), 1-11. https://doi.org/10.32985/ijeces.11.1.1
MLA 8th Edition
Velasco-Montero, Delia, et al. "Performance Assessment of Deep Learning Frameworks through Metrics of CPU Hardware Exploitation on an Embedded Platform." International journal of electrical and computer engineering systems, vol. 11, br. 1, 2020, str. 1-11. https://doi.org/10.32985/ijeces.11.1.1. Citirano 22.09.2021.
Chicago 17th Edition
Velasco-Montero, Delia, Jorge Fernández-Berni, Ricardo Carmona-Galán i Ángel Rodríguez-Vázquez. "Performance Assessment of Deep Learning Frameworks through Metrics of CPU Hardware Exploitation on an Embedded Platform." International journal of electrical and computer engineering systems 11, br. 1 (2020): 1-11. https://doi.org/10.32985/ijeces.11.1.1
Harvard
Velasco-Montero, D., et al. (2020). 'Performance Assessment of Deep Learning Frameworks through Metrics of CPU Hardware Exploitation on an Embedded Platform', International journal of electrical and computer engineering systems, 11(1), str. 1-11. https://doi.org/10.32985/ijeces.11.1.1
Vancouver
Velasco-Montero D, Fernández-Berni J, Carmona-Galán R, Rodríguez-Vázquez Á. Performance Assessment of Deep Learning Frameworks through Metrics of CPU Hardware Exploitation on an Embedded Platform. International journal of electrical and computer engineering systems [Internet]. 2020 [pristupljeno 22.09.2021.];11(1):1-11. https://doi.org/10.32985/ijeces.11.1.1
IEEE
D. Velasco-Montero, J. Fernández-Berni, R. Carmona-Galán i Á. Rodríguez-Vázquez, "Performance Assessment of Deep Learning Frameworks through Metrics of CPU Hardware Exploitation on an Embedded Platform", International journal of electrical and computer engineering systems, vol.11, br. 1, str. 1-11, 2020. [Online]. https://doi.org/10.32985/ijeces.11.1.1

Sažetak
In this paper, we analyze heterogeneous performance exhibited by some popular deep learning software frameworks for visual inference on a resource-constrained hardware platform. Benchmarking of Caffe, OpenCV, TensorFlow, and Caffe2 is performed on the same set of convolutional neural networks in terms of instantaneous throughput, power consumption, memory footprint, and CPU utilization. To understand the resulting dissimilar behavior, we thoroughly examine how the resources in the processor are differently exploited by these frameworks. We demonstrate that a strong correlation exists between hardware events occurring in the processor and inference performance. The proposed hardware-aware analysis aims to find limitations and bottlenecks emerging from the joint interaction of frameworks and networks on a particular CPU-based platform. This provides insight into introducing suitable modifications in both types of components to enhance their global performance. It also facilitates the selection of frameworks and networks among a large diversity of these components available these days for visual understanding.

Ključne riječi
convolutional neural networks, deep learning, edge inference, embedded vision, hardware performance, software frameworks

Hrčak ID: 242926

URI
https://hrcak.srce.hr/242926

Posjeta: 246 *