Multiple Algorithms Against Multiple Hardware Architectures: Data-Driven Exploration on Deep Convolution Neural Network

Chongyang Xu; Zhongzhi Luan; Lan Gao; Rui Wang; Han Zhang; Lianyi Zhang; Yi Liu; Depei Qian

doi:10.1007/978-3-030-30709-7_36

Multiple Algorithms Against Multiple Hardware Architectures: Data-Driven Exploration on Deep Convolution Neural Network

Chongyang Xu
, Zhongzhi Luan
, Lan Gao
, Rui Wang^*
, Han Zhang
, Lianyi Zhang
, Yi Liu
, Depei Qian

^*Corresponding author for this work

School of Computer Science and Engineering

Beihang University
Beijing Simulation Center

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

With the rapid development of deep learning (DL), various convolution neural network (CNN) models have been developed. Moreover, to execute different DL workloads efficiently, many accelerators have been proposed. To guide the design of both CNN models and hardware architectures for a high-performance inference system, we choose five types of CNN models and test them on six processors and measure three metrics. With our experiments, we get two observations and conduct two insights for the design of CNN algorithms and hardware architectures.

Original language	English
Title of host publication	Network and Parallel Computing - 16th IFIP WG 10.3 International Conference, NPC 2019, Proceedings
Editors	Xiaoxin Tang, Quan Chen, Pradip Bose, Weiming Zheng, Jean-Luc Gaudiot
Publisher	Springer
Pages	371-375
Number of pages	5
ISBN (Print)	9783030307080
DOIs	https://doi.org/10.1007/978-3-030-30709-7_36
State	Published - 2019
Event	16th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2019 - Hohhot, China Duration: 23 Aug 2019 → 24 Aug 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11783 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	16th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2019
Country/Territory	China
City	Hohhot
Period	23/08/19 → 24/08/19

Keywords

Convolutional neural network
Hardware architecture
Performance evaluation

Access to Document

10.1007/978-3-030-30709-7_36

Cite this

Xu, C., Luan, Z., Gao, L., Wang, R., Zhang, H., Zhang, L., Liu, Y., & Qian, D. (2019). Multiple Algorithms Against Multiple Hardware Architectures: Data-Driven Exploration on Deep Convolution Neural Network. In X. Tang, Q. Chen, P. Bose, W. Zheng, & J.-L. Gaudiot (Eds.), Network and Parallel Computing - 16th IFIP WG 10.3 International Conference, NPC 2019, Proceedings (pp. 371-375). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11783 LNCS). Springer. https://doi.org/10.1007/978-3-030-30709-7_36

Xu, Chongyang ; Luan, Zhongzhi ; Gao, Lan et al. / Multiple Algorithms Against Multiple Hardware Architectures : Data-Driven Exploration on Deep Convolution Neural Network. Network and Parallel Computing - 16th IFIP WG 10.3 International Conference, NPC 2019, Proceedings. editor / Xiaoxin Tang ; Quan Chen ; Pradip Bose ; Weiming Zheng ; Jean-Luc Gaudiot. Springer, 2019. pp. 371-375 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{911a33a0f45241e1b009b0ec93cbff6b,

title = "Multiple Algorithms Against Multiple Hardware Architectures: Data-Driven Exploration on Deep Convolution Neural Network",

abstract = "With the rapid development of deep learning (DL), various convolution neural network (CNN) models have been developed. Moreover, to execute different DL workloads efficiently, many accelerators have been proposed. To guide the design of both CNN models and hardware architectures for a high-performance inference system, we choose five types of CNN models and test them on six processors and measure three metrics. With our experiments, we get two observations and conduct two insights for the design of CNN algorithms and hardware architectures.",

keywords = "Convolutional neural network, Hardware architecture, Performance evaluation",

author = "Chongyang Xu and Zhongzhi Luan and Lan Gao and Rui Wang and Han Zhang and Lianyi Zhang and Yi Liu and Depei Qian",

note = "Publisher Copyright: {\textcopyright} 2019, IFIP International Federation for Information Processing.; 16th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2019 ; Conference date: 23-08-2019 Through 24-08-2019",

year = "2019",

doi = "10.1007/978-3-030-30709-7\_36",

language = "英语",

isbn = "9783030307080",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "371--375",

editor = "Xiaoxin Tang and Quan Chen and Pradip Bose and Weiming Zheng and Jean-Luc Gaudiot",

booktitle = "Network and Parallel Computing - 16th IFIP WG 10.3 International Conference, NPC 2019, Proceedings",

address = "德国",

}

Xu, C, Luan, Z, Gao, L, Wang, R, Zhang, H, Zhang, L, Liu, Y & Qian, D 2019, Multiple Algorithms Against Multiple Hardware Architectures: Data-Driven Exploration on Deep Convolution Neural Network. in X Tang, Q Chen, P Bose, W Zheng & J-L Gaudiot (eds), Network and Parallel Computing - 16th IFIP WG 10.3 International Conference, NPC 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11783 LNCS, Springer, pp. 371-375, 16th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2019, Hohhot, China, 23/08/19. https://doi.org/10.1007/978-3-030-30709-7_36

Multiple Algorithms Against Multiple Hardware Architectures: Data-Driven Exploration on Deep Convolution Neural Network. / Xu, Chongyang; Luan, Zhongzhi; Gao, Lan et al.
Network and Parallel Computing - 16th IFIP WG 10.3 International Conference, NPC 2019, Proceedings. ed. / Xiaoxin Tang; Quan Chen; Pradip Bose; Weiming Zheng; Jean-Luc Gaudiot. Springer, 2019. p. 371-375 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11783 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Multiple Algorithms Against Multiple Hardware Architectures

T2 - 16th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2019

AU - Xu, Chongyang

AU - Luan, Zhongzhi

AU - Gao, Lan

AU - Wang, Rui

AU - Zhang, Han

AU - Zhang, Lianyi

AU - Liu, Yi

AU - Qian, Depei

PY - 2019

Y1 - 2019

N2 - With the rapid development of deep learning (DL), various convolution neural network (CNN) models have been developed. Moreover, to execute different DL workloads efficiently, many accelerators have been proposed. To guide the design of both CNN models and hardware architectures for a high-performance inference system, we choose five types of CNN models and test them on six processors and measure three metrics. With our experiments, we get two observations and conduct two insights for the design of CNN algorithms and hardware architectures.

AB - With the rapid development of deep learning (DL), various convolution neural network (CNN) models have been developed. Moreover, to execute different DL workloads efficiently, many accelerators have been proposed. To guide the design of both CNN models and hardware architectures for a high-performance inference system, we choose five types of CNN models and test them on six processors and measure three metrics. With our experiments, we get two observations and conduct two insights for the design of CNN algorithms and hardware architectures.

KW - Convolutional neural network

KW - Hardware architecture

KW - Performance evaluation

UR - https://www.scopus.com/pages/publications/85076111302

U2 - 10.1007/978-3-030-30709-7_36

DO - 10.1007/978-3-030-30709-7_36

M3 - 会议稿件

AN - SCOPUS:85076111302

SN - 9783030307080

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 371

EP - 375

BT - Network and Parallel Computing - 16th IFIP WG 10.3 International Conference, NPC 2019, Proceedings

A2 - Tang, Xiaoxin

A2 - Chen, Quan

A2 - Bose, Pradip

A2 - Zheng, Weiming

A2 - Gaudiot, Jean-Luc

PB - Springer

Y2 - 23 August 2019 through 24 August 2019

ER -

Xu C, Luan Z, Gao L, Wang R, Zhang H, Zhang L et al. Multiple Algorithms Against Multiple Hardware Architectures: Data-Driven Exploration on Deep Convolution Neural Network. In Tang X, Chen Q, Bose P, Zheng W, Gaudiot JL, editors, Network and Parallel Computing - 16th IFIP WG 10.3 International Conference, NPC 2019, Proceedings. Springer. 2019. p. 371-375. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-30709-7_36

Multiple Algorithms Against Multiple Hardware Architectures: Data-Driven Exploration on Deep Convolution Neural Network

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this