Memory Allocation for Neural Networks using Graph Coloring

Leonid Barenboim; Rami Drucker; Oleg Zatulovsky; Eli Levi

doi:10.1145/3491003.3498354

Memory Allocation for Neural Networks using Graph Coloring

Leonid Barenboim, Rami Drucker, Oleg Zatulovsky, Eli Levi

Computer Science discipline

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

Abstract

The Memory Allocation problem for neural networks can be represented as a two-dimensional optimization problem. The neural network is allocated into limited memory space while allocating as much data as possible into the low latency memory. Our solution is based on a generalization of graph coloring, edge-to-node transformation and considers the order in which the graph nodes are colored. We observed improvement of more than 40% in SRAM memory bandwidth in various neural networks.

Original language	English
Title of host publication	ICDCN 2022 - Proceedings of the 2022 International Conference on Distributed Computing and Networking
Publisher	Association for Computing Machinery
Pages	232-233
Number of pages	2
ISBN (Electronic)	9781450395601
DOIs	https://doi.org/10.1145/3491003.3498354
State	Published - 4 Jan 2022
Event	23rd International Conference on Distributed Computing and Networking, ICDCN 2022 - Virtual, Online, India Duration: 4 Jan 2022 → 7 Jan 2022

Publication series

Name	ACM International Conference Proceeding Series

Conference

Conference	23rd International Conference on Distributed Computing and Networking, ICDCN 2022
Country/Territory	India
City	Virtual, Online
Period	4/01/22 → 7/01/22

Bibliographical note

Publisher Copyright:
© 2022 Owner/Author.

Keywords

Graph Coloring
Memory Management
Neural Networks

Access to Document

10.1145/3491003.3498354

Cite this

@inproceedings{fb2c0c1f473a43fd8880d145ad9ec603,

title = "Memory Allocation for Neural Networks using Graph Coloring",

abstract = "The Memory Allocation problem for neural networks can be represented as a two-dimensional optimization problem. The neural network is allocated into limited memory space while allocating as much data as possible into the low latency memory. Our solution is based on a generalization of graph coloring, edge-to-node transformation and considers the order in which the graph nodes are colored. We observed improvement of more than 40% in SRAM memory bandwidth in various neural networks.",

keywords = "Graph Coloring, Memory Management, Neural Networks",

author = "Leonid Barenboim and Rami Drucker and Oleg Zatulovsky and Eli Levi",

note = "DBLP License: DBLP's bibliographic metadata records provided through http://dblp.org/ are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.; 23rd International Conference on Distributed Computing and Networking, ICDCN 2022 ; Conference date: 04-01-2022 Through 07-01-2022",

year = "2022",

month = jan,

day = "4",

doi = "10.1145/3491003.3498354",

language = "אנגלית",

series = "ACM International Conference Proceeding Series",

publisher = "Association for Computing Machinery",

pages = "232--233",

booktitle = "ICDCN 2022 - Proceedings of the 2022 International Conference on Distributed Computing and Networking",

}

Barenboim, L, Drucker, R, Zatulovsky, O & Levi, E 2022, Memory Allocation for Neural Networks using Graph Coloring. in ICDCN 2022 - Proceedings of the 2022 International Conference on Distributed Computing and Networking. ACM International Conference Proceeding Series, Association for Computing Machinery, pp. 232-233, 23rd International Conference on Distributed Computing and Networking, ICDCN 2022, Virtual, Online, India, 4/01/22. https://doi.org/10.1145/3491003.3498354

Memory Allocation for Neural Networks using Graph Coloring. / Barenboim, Leonid; Drucker, Rami; Zatulovsky, Oleg et al.
ICDCN 2022 - Proceedings of the 2022 International Conference on Distributed Computing and Networking. Association for Computing Machinery, 2022. p. 232-233 (ACM International Conference Proceeding Series).

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

TY - GEN

T1 - Memory Allocation for Neural Networks using Graph Coloring

AU - Barenboim, Leonid

AU - Drucker, Rami

AU - Zatulovsky, Oleg

AU - Levi, Eli

N1 - DBLP License: DBLP's bibliographic metadata records provided through http://dblp.org/ are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.

PY - 2022/1/4

Y1 - 2022/1/4

N2 - The Memory Allocation problem for neural networks can be represented as a two-dimensional optimization problem. The neural network is allocated into limited memory space while allocating as much data as possible into the low latency memory. Our solution is based on a generalization of graph coloring, edge-to-node transformation and considers the order in which the graph nodes are colored. We observed improvement of more than 40% in SRAM memory bandwidth in various neural networks.

AB - The Memory Allocation problem for neural networks can be represented as a two-dimensional optimization problem. The neural network is allocated into limited memory space while allocating as much data as possible into the low latency memory. Our solution is based on a generalization of graph coloring, edge-to-node transformation and considers the order in which the graph nodes are colored. We observed improvement of more than 40% in SRAM memory bandwidth in various neural networks.

KW - Graph Coloring

KW - Memory Management

KW - Neural Networks

UR - http://www.scopus.com/inward/record.url?scp=85123997013&partnerID=8YFLogxK

U2 - 10.1145/3491003.3498354

DO - 10.1145/3491003.3498354

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:85123997013

T3 - ACM International Conference Proceeding Series

SP - 232

EP - 233

BT - ICDCN 2022 - Proceedings of the 2022 International Conference on Distributed Computing and Networking

PB - Association for Computing Machinery

T2 - 23rd International Conference on Distributed Computing and Networking, ICDCN 2022

Y2 - 4 January 2022 through 7 January 2022

ER -

Memory Allocation for Neural Networks using Graph Coloring

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this