Integer Sum Reduction with OpenMP on an AMD MI100 GPU

Zheming Jin; Jeffrey S. Vetter

doi:10.1109/IPDPSW55747.2022.00088

Integer Sum Reduction with OpenMP on an AMD MI100 GPU

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

Abstract

Sum reduction is a primitive operation in parallel computing. Device offload support allows a user to use OpenMP directives to take advantage of a highly capable GPU. In this paper, we present the integer sum reduction annotated with the OpenMP directives and evaluate the performance impacts of tunable parameters with the AOMP and GCC compilers on an AMD MI100 GPU. In addition, we explain the implementations of the OpenMP reduction by the compilers. Sweeping over the pruned parameter space, we find that the speedup is approximately 20 with AOMP, and the reduction performance using AOMP is approximately 11% higher than that using GCC. However, the OpenMP offload performance is approximately 30% lower compared to the performance of the reductions written with rocThrust or hipCUB.

Original language	English
Title of host publication	Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	496-499
Number of pages	4
ISBN (Electronic)	9781665497473
DOIs	https://doi.org/10.1109/IPDPSW55747.2022.00088
State	Published - 2022
Event	36th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022 - Virtual, Online, France Duration: May 30 2022 → Jun 3 2022

Publication series

Name	Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022

Conference

Conference	36th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022
Country/Territory	France
City	Virtual, Online
Period	05/30/22 → 06/3/22

Funding

ACKNOWLEDGMENT We sincerely appreciate the reviewers for their comments and suggestions. This research used resources of the Experimental Computing Lab at Oak Ridge National Laboratory. This research was supported by the US Department of Energy Advanced Scientific Computing Research program under Contract No. DE-AC05-00OR22725. This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Funders	Funder number
US Department of Energy Advanced Scientific Computing Research	DE-AC05-00OR22725
U.S. Department of Energy

Keywords

AMD GPU
OpenMP target offload
Reduction

Access to Document

10.1109/IPDPSW55747.2022.00088

Cite this

Jin, Z., & Vetter, J. S. (2022). Integer Sum Reduction with OpenMP on an AMD MI100 GPU. In Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022 (pp. 496-499). (Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IPDPSW55747.2022.00088

Jin, Zheming ; Vetter, Jeffrey S. / Integer Sum Reduction with OpenMP on an AMD MI100 GPU. Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022. Institute of Electrical and Electronics Engineers Inc., 2022. pp. 496-499 (Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022).

@inproceedings{117ae32d62574dccb587e57770c535ce,

title = "Integer Sum Reduction with OpenMP on an AMD MI100 GPU",

abstract = "Sum reduction is a primitive operation in parallel computing. Device offload support allows a user to use OpenMP directives to take advantage of a highly capable GPU. In this paper, we present the integer sum reduction annotated with the OpenMP directives and evaluate the performance impacts of tunable parameters with the AOMP and GCC compilers on an AMD MI100 GPU. In addition, we explain the implementations of the OpenMP reduction by the compilers. Sweeping over the pruned parameter space, we find that the speedup is approximately 20 with AOMP, and the reduction performance using AOMP is approximately 11% higher than that using GCC. However, the OpenMP offload performance is approximately 30% lower compared to the performance of the reductions written with rocThrust or hipCUB.",

keywords = "AMD GPU, OpenMP target offload, Reduction",

author = "Zheming Jin and Vetter, {Jeffrey S.}",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 36th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022 ; Conference date: 30-05-2022 Through 03-06-2022",

year = "2022",

doi = "10.1109/IPDPSW55747.2022.00088",

language = "English",

series = "Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022",

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

pages = "496--499",

booktitle = "Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022",

}

Jin, Z & Vetter, JS 2022, Integer Sum Reduction with OpenMP on an AMD MI100 GPU. in Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022. Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022, Institute of Electrical and Electronics Engineers Inc., pp. 496-499, 36th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022, Virtual, Online, France, 05/30/22. https://doi.org/10.1109/IPDPSW55747.2022.00088

Integer Sum Reduction with OpenMP on an AMD MI100 GPU. / Jin, Zheming; Vetter, Jeffrey S.
Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022. Institute of Electrical and Electronics Engineers Inc., 2022. p. 496-499 (Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022).

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

TY - GEN

T1 - Integer Sum Reduction with OpenMP on an AMD MI100 GPU

AU - Jin, Zheming

AU - Vetter, Jeffrey S.

PY - 2022

Y1 - 2022

N2 - Sum reduction is a primitive operation in parallel computing. Device offload support allows a user to use OpenMP directives to take advantage of a highly capable GPU. In this paper, we present the integer sum reduction annotated with the OpenMP directives and evaluate the performance impacts of tunable parameters with the AOMP and GCC compilers on an AMD MI100 GPU. In addition, we explain the implementations of the OpenMP reduction by the compilers. Sweeping over the pruned parameter space, we find that the speedup is approximately 20 with AOMP, and the reduction performance using AOMP is approximately 11% higher than that using GCC. However, the OpenMP offload performance is approximately 30% lower compared to the performance of the reductions written with rocThrust or hipCUB.

AB - Sum reduction is a primitive operation in parallel computing. Device offload support allows a user to use OpenMP directives to take advantage of a highly capable GPU. In this paper, we present the integer sum reduction annotated with the OpenMP directives and evaluate the performance impacts of tunable parameters with the AOMP and GCC compilers on an AMD MI100 GPU. In addition, we explain the implementations of the OpenMP reduction by the compilers. Sweeping over the pruned parameter space, we find that the speedup is approximately 20 with AOMP, and the reduction performance using AOMP is approximately 11% higher than that using GCC. However, the OpenMP offload performance is approximately 30% lower compared to the performance of the reductions written with rocThrust or hipCUB.

KW - AMD GPU

KW - OpenMP target offload

KW - Reduction

UR - http://www.scopus.com/inward/record.url?scp=85136160970&partnerID=8YFLogxK

U2 - 10.1109/IPDPSW55747.2022.00088

DO - 10.1109/IPDPSW55747.2022.00088

M3 - Conference contribution

AN - SCOPUS:85136160970

T3 - Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022

SP - 496

EP - 499

BT - Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 36th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022

Y2 - 30 May 2022 through 3 June 2022

ER -

Jin Z, Vetter JS. Integer Sum Reduction with OpenMP on an AMD MI100 GPU. In Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022. Institute of Electrical and Electronics Engineers Inc. 2022. p. 496-499. (Proceedings - 2022 IEEE 36th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2022). doi: 10.1109/IPDPSW55747.2022.00088

Integer Sum Reduction with OpenMP on an AMD MI100 GPU

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this