Mostrar el registro sencillo del ítem
Efficient and portable GEMM-based convolution operators for deep neural network training on multicore processors
dc.contributor.author | Barrachina Mir, Sergio | |
dc.contributor.author | Dolz, Manuel F. | |
dc.contributor.author | San Juan, Pablo | |
dc.contributor.author | Quintana-Orti, Enrique S. | |
dc.date.accessioned | 2022-06-22T12:01:16Z | |
dc.date.available | 2022-06-22T12:01:16Z | |
dc.date.issued | 2022-05-30 | |
dc.identifier.citation | Barrachina, S., Dolz, M. F., San Juan, P., & Quintana-Ortí, E. S. (2022). Efficient and Portable GEMM-based Convolution Operators for Deep Neural Network Training on Multicore Processors. Journal of Parallel and Distributed Computing. | ca_CA |
dc.identifier.issn | 0743-7315 | |
dc.identifier.uri | http://hdl.handle.net/10234/198101 | |
dc.description.abstract | Convolutional Neural Networks (CNNs) play a crucial role in many image recognition and classification tasks, recommender systems, brain-computer interfaces, etc. As a consequence, there is a notable interest in developing high performance realizations of the convolution operators, which concentrate a significant portion of the computational cost of this type of neural networks. In a previous work, we introduced a portable, high performance convolution algorithm, based on the BLIS realization of matrix multiplication, which eliminates most of the runtime and memory overheads that impair the performance of the convolution operators appearing in the forward training pass, when performed via explicit im2col transform. In this paper, we extend our ideas to the full training process of CNNs on multicore processors, proposing new high performance strategies to tackle the convolution operators that are present in the more complex backward pass of the training process, while maintaining the portability of the realizations. In addition, we conduct a full integration of these algorithms into a framework for distributed training of CNNs on clusters of computers, providing a complete experimental evaluation of the actual benefits in terms of both performance and memory consumption. Compared with baseline implementation, the use of the new convolution operators using pre-allocated memory can accelerate the training by a factor of about 6%–25%, provided there is sufficient memory available. In comparison, the operator variants that do not rely on persistent memory can save up to 70% of memory. | ca_CA |
dc.description.sponsorShip | Funding for open access charge: CRUE-Universitat Jaume I | |
dc.format.extent | 15 p. | ca_CA |
dc.format.mimetype | application/pdf | ca_CA |
dc.language.iso | eng | ca_CA |
dc.publisher | Elsevier | ca_CA |
dc.publisher | Academic Press | ca_CA |
dc.relation.isPartOf | Journal of Parallel and Distributed Computing 167 (2022) 240–254 | ca_CA |
dc.rights | 0743-7315/© 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). | ca_CA |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/4.0/ | ca_CA |
dc.subject | convolutional neural networks | ca_CA |
dc.subject | distributed training | ca_CA |
dc.subject | high performance | ca_CA |
dc.subject | Python | ca_CA |
dc.subject | clusters of multicore processors | ca_CA |
dc.title | Efficient and portable GEMM-based convolution operators for deep neural network training on multicore processors | ca_CA |
dc.type | info:eu-repo/semantics/article | ca_CA |
dc.identifier.doi | https://doi.org/10.1016/j.jpdc.2022.05.009 | |
dc.rights.accessRights | info:eu-repo/semantics/openAccess | ca_CA |
dc.type.version | info:eu-repo/semantics/publishedVersion | ca_CA |
project.funder.name | Generalitat Valenciana | ca_CA |
oaire.awardNumber | PID2020-113656RB-C21/C22 | ca_CA |
oaire.awardNumber | MCIN/AEI/10.13039/501100011033 | ca_CA |
oaire.awardNumber | Prometeo/2019/109 | ca_CA |
oaire.awardNumber | CDEIGENT/2018/014 | ca_CA |
Ficheros en el ítem
Este ítem aparece en la(s) siguiente(s) colección(ones)
-
ICC_Articles [427]