Highly-Efficient Parallel Convolution Acceleration by Using Multiple GPUs

被引：0

作者：

Sun, Kuangyuan ^{[1
]}

Li, Shuai ^{[1
]}

Luo, Yukui ^{[1
]}

Renteria, Raul ^{[1
]}

Choi, Ken ^{[1
]}

机构：

[1] IIT, Dept Elect & Comp Engn, VLSI Design & Automat Lab, 3301 S Dearborn St, Chicago, IL 60616 USA

来源：

PROCEEDINGS INTERNATIONAL SOC DESIGN CONFERENCE 2017 (ISOCC 2017) | 2017年

关键词：

Convolutional neural network; parallel acceleration; multiple GPUs;

D O I：

暂无

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

Convolutional Neural Network (CNN) is a powerful tool in machine learning area. However, the convolution computation is time-consuming, which limited the application on embedded system. In this paper, we introduce a parallel convolution acceleration implementation by using multiple GPUs Mali-T628 MP6 on embedded system Odroid XU4 and have tested its time reduction and GPU utilization. The result show that the execution time is reduced 25.8% on average.

引用

页码：300 / 301

页数：2

共 50 条

[31] Enabling Efficient Fast Convolution Algorithms on GPUs via MegaKernels
Jia, Liancheng
Liang, Yun
Li, Xiuhong
Lu, Liqiang
Yan, Shengen
IEEE TRANSACTIONS ON COMPUTERS, 2020, 69 (07) : 986 - 997
[32] Petascale turbulence simulation using a highly parallel fast multipole method on GPUs
Yokota, Rio
Barba, L. A.
Narumi, Tetsu
Yasuoka, Kenji
COMPUTER PHYSICS COMMUNICATIONS, 2013, 184 (03) : 445 - 455
[33] LIGHT-INTENSITY STABILIZATION USING HIGHLY-EFFICIENT FARADAY ROTATOR
YOSHINO, T
UMEGAKI, S
INOUE, H
KUROSAWA, K
JAPANESE JOURNAL OF APPLIED PHYSICS PART 1-REGULAR PAPERS SHORT NOTES & REVIEW PAPERS, 1982, 21 (04): : 612 - 616
[34] Parallel Acceleration on Simulation of a 2D Takeuchi Electrophysiology Cardiac Model Using GPUs
Qiu, Feng
Liu, Baohua
Shen, Wenfeng
Shen, Yanghua
Hu, Yaopeng
Zhu, Xin
PROCEEDINGS OF 2017 3RD IEEE INTERNATIONAL CONFERENCE ON COMPUTER AND COMMUNICATIONS (ICCC), 2017, : 2665 - 2668
[35] Memory-Efficient Parallel Simulation of Electron Beam Dynamics Using GPUs
Arumugam, Kamesh
Godunov, Alexander
Ranjan, Desh
Terzic, Balsa
Zubair, Mohammad
PROCEEDINGS OF 2016 IEEE 23RD INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE COMPUTING (HIPC), 2016, : 212 - 221
[36] A Winograd-Based Highly-Parallel Convolution Engine for 8-bit CNN Acceleration
Chen, Yong-Tai
Ou, Yu-Feng
Huang, Chao-Tsung
2022 IEEE INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE CIRCUITS AND SYSTEMS (AICAS 2022): INTELLIGENT TECHNOLOGY IN THE POST-PANDEMIC ERA, 2022, : 395 - 398
[37] A highly-efficient transparent online memory test
Thaller, K
INTERNATIONAL TEST CONFERENCE 2001, PROCEEDINGS, 2001, : 230 - 239
[38] A Highly-Efficient, Unidirectional Miniaturized Slot Antenna
Al-Joumayly, Mudar A.
Behdad, Nader
2008 IEEE ANTENNAS AND PROPAGATION SOCIETY INTERNATIONAL SYMPOSIUM, VOLS 1-9, 2008, : 1128 - 1131
[39] A Highly Parallel Reuse Distance Analysis Algorithm on GPUs
Cui, Huimin
Yi, Qing
Xue, Jingling
Wang, Lei
Yang, Yang
Feng, Xiaobing
2012 IEEE 26TH INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS), 2012, : 1080 - 1092
[40] Highly Parallel Transformation and Quantization for HEVC Encoder on GPUs
Igarashi, Hiroaki
Takano, Fumiyo
Moriyoshi, Tatsuji
2016 30TH ANNIVERSARY OF VISUAL COMMUNICATION AND IMAGE PROCESSING (VCIP), 2016,

← 1 2 3 4 5 →