Performance portability study of epistasis detection using SYCL on NVIDIA GPU

Zheming Jin; Jeffrey S. Vetter

doi:10.1145/3535508.3545591

Performance portability study of epistasis detection using SYCL on NVIDIA GPU

Zheming Jin, Jeffrey S. Vetter

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

13 Scopus citations

Abstract

We describe the experience of converting a CUDA implementation of a high-order epistasis detection algorithm to SYCL. The goals are for our work to be useful to application and compiler developers with a detailed description of migration paths between CUDA and SYCL. Evaluating the CUDA and SYCL applications on an NVIDIA V100 GPU, we find that the optimization of loop unrolling needs to be applied manually to the SYCL kernel for obtaining comparable performance. The performance of the SYCL group reduce function, an alternative to the CUDA warp-based reduction, depends on the problem and work group sizes. The 64-bit popcount operation implemented with tree of adders is slightly faster than the built-in popcount operation. When the number of OpenMP threads is four, the highest performance of the SYCL and CUDA applications are comparable.

Original language	English
Title of host publication	Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022
Publisher	Association for Computing Machinery, Inc
ISBN (Electronic)	9781450393867
DOIs	https://doi.org/10.1145/3535508.3545591
State	Published - Aug 7 2022
Event	13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022 - Chicago, United States Duration: Aug 7 2022 → Aug 8 2022

Publication series

Name	Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022

Conference

Conference	13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022
Country/Territory	United States
City	Chicago
Period	08/7/22 → 08/8/22

Funding

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).

Keywords

Epistasis
GPU
Portability
Programming model

Access to Document

10.1145/3535508.3545591

Cite this

Jin, Z., & Vetter, J. S. (2022). Performance portability study of epistasis detection using SYCL on NVIDIA GPU. In Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022 Article 69 (Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022). Association for Computing Machinery, Inc. https://doi.org/10.1145/3535508.3545591

Jin, Zheming ; Vetter, Jeffrey S. / Performance portability study of epistasis detection using SYCL on NVIDIA GPU. Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022. Association for Computing Machinery, Inc, 2022. (Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022).

@inproceedings{953d71049f3849d0aefc0269335a14cd,

title = "Performance portability study of epistasis detection using SYCL on NVIDIA GPU",

abstract = "We describe the experience of converting a CUDA implementation of a high-order epistasis detection algorithm to SYCL. The goals are for our work to be useful to application and compiler developers with a detailed description of migration paths between CUDA and SYCL. Evaluating the CUDA and SYCL applications on an NVIDIA V100 GPU, we find that the optimization of loop unrolling needs to be applied manually to the SYCL kernel for obtaining comparable performance. The performance of the SYCL group reduce function, an alternative to the CUDA warp-based reduction, depends on the problem and work group sizes. The 64-bit popcount operation implemented with tree of adders is slightly faster than the built-in popcount operation. When the number of OpenMP threads is four, the highest performance of the SYCL and CUDA applications are comparable.",

keywords = "Epistasis, GPU, Portability, Programming model",

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

note = "Publisher Copyright: {\textcopyright} 2022 ACM.; 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022 ; Conference date: 07-08-2022 Through 08-08-2022",

year = "2022",

month = aug,

day = "7",

doi = "10.1145/3535508.3545591",

language = "English",

series = "Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022",

publisher = "Association for Computing Machinery, Inc",

booktitle = "Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022",

}

Jin, Z & Vetter, JS 2022, Performance portability study of epistasis detection using SYCL on NVIDIA GPU. in Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022., 69, Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022, Association for Computing Machinery, Inc, 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022, Chicago, United States, 08/7/22. https://doi.org/10.1145/3535508.3545591

Performance portability study of epistasis detection using SYCL on NVIDIA GPU. / Jin, Zheming; Vetter, Jeffrey S.
Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022. Association for Computing Machinery, Inc, 2022. 69 (Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022).

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

TY - GEN

T1 - Performance portability study of epistasis detection using SYCL on NVIDIA GPU

AU - Jin, Zheming

AU - Vetter, Jeffrey S.

PY - 2022/8/7

Y1 - 2022/8/7

N2 - We describe the experience of converting a CUDA implementation of a high-order epistasis detection algorithm to SYCL. The goals are for our work to be useful to application and compiler developers with a detailed description of migration paths between CUDA and SYCL. Evaluating the CUDA and SYCL applications on an NVIDIA V100 GPU, we find that the optimization of loop unrolling needs to be applied manually to the SYCL kernel for obtaining comparable performance. The performance of the SYCL group reduce function, an alternative to the CUDA warp-based reduction, depends on the problem and work group sizes. The 64-bit popcount operation implemented with tree of adders is slightly faster than the built-in popcount operation. When the number of OpenMP threads is four, the highest performance of the SYCL and CUDA applications are comparable.

AB - We describe the experience of converting a CUDA implementation of a high-order epistasis detection algorithm to SYCL. The goals are for our work to be useful to application and compiler developers with a detailed description of migration paths between CUDA and SYCL. Evaluating the CUDA and SYCL applications on an NVIDIA V100 GPU, we find that the optimization of loop unrolling needs to be applied manually to the SYCL kernel for obtaining comparable performance. The performance of the SYCL group reduce function, an alternative to the CUDA warp-based reduction, depends on the problem and work group sizes. The 64-bit popcount operation implemented with tree of adders is slightly faster than the built-in popcount operation. When the number of OpenMP threads is four, the highest performance of the SYCL and CUDA applications are comparable.

KW - Epistasis

KW - GPU

KW - Portability

KW - Programming model

UR - http://www.scopus.com/inward/record.url?scp=85137369147&partnerID=8YFLogxK

U2 - 10.1145/3535508.3545591

DO - 10.1145/3535508.3545591

M3 - Conference contribution

AN - SCOPUS:85137369147

T3 - Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022

BT - Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022

PB - Association for Computing Machinery, Inc

T2 - 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022

Y2 - 7 August 2022 through 8 August 2022

ER -

Jin Z, Vetter JS. Performance portability study of epistasis detection using SYCL on NVIDIA GPU. In Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022. Association for Computing Machinery, Inc. 2022. 69. (Proceedings of the 13th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB 2022). doi: 10.1145/3535508.3545591

Performance portability study of epistasis detection using SYCL on NVIDIA GPU

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this