Study of GPU scalability

Kyu Hyun Choi, Dongha Jung, Seon Wook Kim, Tae Ho Hwang, Jin San Kwon, Dong Sun Kim

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

General-purpose Graphic Processing Units (GPGPUs) have been widely used to accelerate heavy compute-intensive applications. In a market the number of GPU cores on one chip are increased to maximize performance by exploiting thread-level parallelism (TLP). But memory system could be a bottleneck, and assigning more cores does not help faster execution. In this paper, we observe GPU core scalability issue, and analyze major causes of the performance degradation under many core configuration in perspective of memory latency.

Original languageEnglish
Title of host publicationProceedings of the International Symposium on Consumer Electronics, ISCE
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Print)9781479945924
DOIs
Publication statusPublished - 2014 Jan 1
Event18th IEEE International Symposium on Consumer Electronics, ISCE 2014 - Jeju, Korea, Republic of
Duration: 2014 Jun 222014 Jun 25

Other

Other18th IEEE International Symposium on Consumer Electronics, ISCE 2014
CountryKorea, Republic of
CityJeju
Period14/6/2214/6/25

Fingerprint

Scalability
Data storage equipment
Degradation
Graphics processing unit

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Choi, K. H., Jung, D., Kim, S. W., Hwang, T. H., Kwon, J. S., & Kim, D. S. (2014). Study of GPU scalability. In Proceedings of the International Symposium on Consumer Electronics, ISCE [6884421] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISCE.2014.6884421

Study of GPU scalability. / Choi, Kyu Hyun; Jung, Dongha; Kim, Seon Wook; Hwang, Tae Ho; Kwon, Jin San; Kim, Dong Sun.

Proceedings of the International Symposium on Consumer Electronics, ISCE. Institute of Electrical and Electronics Engineers Inc., 2014. 6884421.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Choi, KH, Jung, D, Kim, SW, Hwang, TH, Kwon, JS & Kim, DS 2014, Study of GPU scalability. in Proceedings of the International Symposium on Consumer Electronics, ISCE., 6884421, Institute of Electrical and Electronics Engineers Inc., 18th IEEE International Symposium on Consumer Electronics, ISCE 2014, Jeju, Korea, Republic of, 14/6/22. https://doi.org/10.1109/ISCE.2014.6884421
Choi KH, Jung D, Kim SW, Hwang TH, Kwon JS, Kim DS. Study of GPU scalability. In Proceedings of the International Symposium on Consumer Electronics, ISCE. Institute of Electrical and Electronics Engineers Inc. 2014. 6884421 https://doi.org/10.1109/ISCE.2014.6884421
Choi, Kyu Hyun ; Jung, Dongha ; Kim, Seon Wook ; Hwang, Tae Ho ; Kwon, Jin San ; Kim, Dong Sun. / Study of GPU scalability. Proceedings of the International Symposium on Consumer Electronics, ISCE. Institute of Electrical and Electronics Engineers Inc., 2014.
@inproceedings{1f980a59b99c4384ab7d0fc4c1dd9bfe,
title = "Study of GPU scalability",
abstract = "General-purpose Graphic Processing Units (GPGPUs) have been widely used to accelerate heavy compute-intensive applications. In a market the number of GPU cores on one chip are increased to maximize performance by exploiting thread-level parallelism (TLP). But memory system could be a bottleneck, and assigning more cores does not help faster execution. In this paper, we observe GPU core scalability issue, and analyze major causes of the performance degradation under many core configuration in perspective of memory latency.",
keywords = "GPU, memory latency, scalability",
author = "Choi, {Kyu Hyun} and Dongha Jung and Kim, {Seon Wook} and Hwang, {Tae Ho} and Kwon, {Jin San} and Kim, {Dong Sun}",
year = "2014",
month = "1",
day = "1",
doi = "10.1109/ISCE.2014.6884421",
language = "English",
isbn = "9781479945924",
booktitle = "Proceedings of the International Symposium on Consumer Electronics, ISCE",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Study of GPU scalability

AU - Choi, Kyu Hyun

AU - Jung, Dongha

AU - Kim, Seon Wook

AU - Hwang, Tae Ho

AU - Kwon, Jin San

AU - Kim, Dong Sun

PY - 2014/1/1

Y1 - 2014/1/1

N2 - General-purpose Graphic Processing Units (GPGPUs) have been widely used to accelerate heavy compute-intensive applications. In a market the number of GPU cores on one chip are increased to maximize performance by exploiting thread-level parallelism (TLP). But memory system could be a bottleneck, and assigning more cores does not help faster execution. In this paper, we observe GPU core scalability issue, and analyze major causes of the performance degradation under many core configuration in perspective of memory latency.

AB - General-purpose Graphic Processing Units (GPGPUs) have been widely used to accelerate heavy compute-intensive applications. In a market the number of GPU cores on one chip are increased to maximize performance by exploiting thread-level parallelism (TLP). But memory system could be a bottleneck, and assigning more cores does not help faster execution. In this paper, we observe GPU core scalability issue, and analyze major causes of the performance degradation under many core configuration in perspective of memory latency.

KW - GPU

KW - memory latency

KW - scalability

UR - http://www.scopus.com/inward/record.url?scp=84907381434&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84907381434&partnerID=8YFLogxK

U2 - 10.1109/ISCE.2014.6884421

DO - 10.1109/ISCE.2014.6884421

M3 - Conference contribution

SN - 9781479945924

BT - Proceedings of the International Symposium on Consumer Electronics, ISCE

PB - Institute of Electrical and Electronics Engineers Inc.

ER -