GPU-Ether: GPU-native packet I/O for GPU applications on commodity ethernet

Changue Jung; Suhwan Kim; Ikjun Yeom; Honguk Woo; Younghoon Kim

doi:10.1109/INFOCOM42981.2021.9488699

GPU-Ether: GPU-native packet I/O for GPU applications on commodity ethernet

Changue Jung, Suhwan Kim, Ikjun Yeom, Honguk Woo, Younghoon Kim

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

4 Scopus citations

Abstract

Despite the advent of various network enhancement technologies, it is yet a challenge to provide high-performance networking for GPU-accelerated applications on commodity Ethernet. Kernel-bypass I/O, such as DPDK or netmap, which is normally optimized for host memory-based CPU applications, has limitations on improving the performance of GPU-accelerated applications due to the data transfer overhead between host and GPU. In this paper, we propose GPU-Ether, GPU-native packet I/O on commodity Ethernet, which enables direct network access from GPU via dedicated persistent kernel threads. We implement GPU-Ether prototype on a commodity Ethernet NIC and perform extensive testing to evaluate it. The results show that GPU-Ether can provide high throughput and low latency for GPU applications.

Original language	English
Title of host publication	INFOCOM 2021 - IEEE Conference on Computer Communications
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9780738112817
DOIs	https://doi.org/10.1109/INFOCOM42981.2021.9488699
State	Published - May 10 2021
Externally published	Yes
Event	40th IEEE Conference on Computer Communications, INFOCOM 2021 - Vancouver, Canada Duration: May 10 2021 → May 13 2021

Publication series

Name	Proceedings - IEEE INFOCOM
Volume	2021-May
ISSN (Print)	0743-166X

Conference

Conference	40th IEEE Conference on Computer Communications, INFOCOM 2021
Country/Territory	Canada
City	Vancouver
Period	05/10/21 → 05/13/21

Access to Document

10.1109/INFOCOM42981.2021.9488699

Cite this

@inproceedings{57b0dbf747ff4b1db90b46ca3ccbf358,

title = "GPU-Ether: GPU-native packet I/O for GPU applications on commodity ethernet",

abstract = "Despite the advent of various network enhancement technologies, it is yet a challenge to provide high-performance networking for GPU-accelerated applications on commodity Ethernet. Kernel-bypass I/O, such as DPDK or netmap, which is normally optimized for host memory-based CPU applications, has limitations on improving the performance of GPU-accelerated applications due to the data transfer overhead between host and GPU. In this paper, we propose GPU-Ether, GPU-native packet I/O on commodity Ethernet, which enables direct network access from GPU via dedicated persistent kernel threads. We implement GPU-Ether prototype on a commodity Ethernet NIC and perform extensive testing to evaluate it. The results show that GPU-Ether can provide high throughput and low latency for GPU applications.",

author = "Changue Jung and Suhwan Kim and Ikjun Yeom and Honguk Woo and Younghoon Kim",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 40th IEEE Conference on Computer Communications, INFOCOM 2021 ; Conference date: 10-05-2021 Through 13-05-2021",

year = "2021",

month = may,

day = "10",

doi = "10.1109/INFOCOM42981.2021.9488699",

language = "English",

series = "Proceedings - IEEE INFOCOM",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "INFOCOM 2021 - IEEE Conference on Computer Communications",

}

Jung, C, Kim, S, Yeom, I, Woo, H & Kim, Y 2021, GPU-Ether: GPU-native packet I/O for GPU applications on commodity ethernet. in INFOCOM 2021 - IEEE Conference on Computer Communications., 9488699, Proceedings - IEEE INFOCOM, vol. 2021-May, Institute of Electrical and Electronics Engineers Inc., 40th IEEE Conference on Computer Communications, INFOCOM 2021, Vancouver, Canada, 05/10/21. https://doi.org/10.1109/INFOCOM42981.2021.9488699

GPU-Ether: GPU-native packet I/O for GPU applications on commodity ethernet. / Jung, Changue; Kim, Suhwan; Yeom, Ikjun et al.
INFOCOM 2021 - IEEE Conference on Computer Communications. Institute of Electrical and Electronics Engineers Inc., 2021. 9488699 (Proceedings - IEEE INFOCOM; Vol. 2021-May).

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

TY - GEN

T1 - GPU-Ether

T2 - 40th IEEE Conference on Computer Communications, INFOCOM 2021

AU - Jung, Changue

AU - Kim, Suhwan

AU - Yeom, Ikjun

AU - Woo, Honguk

AU - Kim, Younghoon

PY - 2021/5/10

Y1 - 2021/5/10

N2 - Despite the advent of various network enhancement technologies, it is yet a challenge to provide high-performance networking for GPU-accelerated applications on commodity Ethernet. Kernel-bypass I/O, such as DPDK or netmap, which is normally optimized for host memory-based CPU applications, has limitations on improving the performance of GPU-accelerated applications due to the data transfer overhead between host and GPU. In this paper, we propose GPU-Ether, GPU-native packet I/O on commodity Ethernet, which enables direct network access from GPU via dedicated persistent kernel threads. We implement GPU-Ether prototype on a commodity Ethernet NIC and perform extensive testing to evaluate it. The results show that GPU-Ether can provide high throughput and low latency for GPU applications.

AB - Despite the advent of various network enhancement technologies, it is yet a challenge to provide high-performance networking for GPU-accelerated applications on commodity Ethernet. Kernel-bypass I/O, such as DPDK or netmap, which is normally optimized for host memory-based CPU applications, has limitations on improving the performance of GPU-accelerated applications due to the data transfer overhead between host and GPU. In this paper, we propose GPU-Ether, GPU-native packet I/O on commodity Ethernet, which enables direct network access from GPU via dedicated persistent kernel threads. We implement GPU-Ether prototype on a commodity Ethernet NIC and perform extensive testing to evaluate it. The results show that GPU-Ether can provide high throughput and low latency for GPU applications.

UR - http://www.scopus.com/inward/record.url?scp=85111943406&partnerID=8YFLogxK

U2 - 10.1109/INFOCOM42981.2021.9488699

DO - 10.1109/INFOCOM42981.2021.9488699

M3 - Conference contribution

AN - SCOPUS:85111943406

T3 - Proceedings - IEEE INFOCOM

BT - INFOCOM 2021 - IEEE Conference on Computer Communications

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 10 May 2021 through 13 May 2021

ER -

GPU-Ether: GPU-native packet I/O for GPU applications on commodity ethernet

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this