Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream

Jeff R Hammon; Tom Deakin; Jim H Cownie; Simon N McIntosh-Smith

doi:10.1109/PMBS56514.2022.00013

Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream

Jeff R Hammon^*, Tom Deakin^*, Jim H Cownie, Simon N McIntosh-Smith

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

8 Citations (Scopus)

739 Downloads (Pure)

Abstract

Fortran DO CONCURRENT has emerged as a new
way to achieve parallel execution of loops on CPUs and GPUs.
This paper studies the performance portability of this construct
on a range of processors and compares it with the incumbent
models: OpenMP, OpenACC and CUDA. To do this study
fairly, we implemented the BabelStream memory bandwidth
benchmark from scratch, entirely in modern Fortran, for all of
the models considered, which include Fortran DO CONCURRENT,
as well as two variants of OpenACC, four variants of OpenMP
(2 CPU and 2 GPU), CUDA Fortran, and both loop- and
array-based references. BabelStream Fortran matches the C++
implementation as closely as possible, and can be used to make
language-based comparisons. This paper represents one of the
first detailed studies of the performance of Fortran support on
heterogeneous architectures; we include results for AArch64 and
x86 64 CPUs as well as AMD, Intel and NVIDIA GPU platforms.

Original language	English
Title of host publication	2022 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS)
Subtitle of host publication	PMBS22
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	1-18
ISBN (Print)	978-1-6654-5185-7
DOIs	https://doi.org/10.1109/PMBS56514.2022.00013
Publication status	Published - 18 Nov 2022
Event	SC 2022 Workshops International Conference for High Performance Computing, Networking, Storage and Analysis - Dallas, United States Duration: 13 Nov 2022 → 18 Nov 2022

Conference

Conference	SC 2022 Workshops International Conference for High Performance Computing, Networking, Storage and Analysis
Country/Territory	United States
City	Dallas
Period	13/11/22 → 18/11/22

Access to Document

10.1109/PMBS56514.2022.00013

Full-text PDF (accepted author manuscript)
This is the accepted author manuscript (AAM). The final published version (version of record) is available online via IEEE at 10.1109/PMBS56514.2022.00013. Please refer to any applicable terms of use of the publisher.
Accepted author manuscript, 350 KB

https://scworkshoppub22:conf22%2F%[email protected]/scworkshoppub/#!/toc/21

Handle.net

Persistent link

Cite this

Hammon, J. R., Deakin, T., Cownie, J. H., & McIntosh-Smith, S. N. (2022). Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream. In 2022 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS): PMBS22 (pp. 1-18). Institute of Electrical and Electronics Engineers (IEEE). https://doi.org/10.1109/PMBS56514.2022.00013

@inproceedings{fb4d0fe4f0b64e5b81ae21c97f9f7e1d,

title = "Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream",

abstract = "Fortran DO CONCURRENT has emerged as a newway to achieve parallel execution of loops on CPUs and GPUs.This paper studies the performance portability of this constructon a range of processors and compares it with the incumbentmodels: OpenMP, OpenACC and CUDA. To do this studyfairly, we implemented the BabelStream memory bandwidthbenchmark from scratch, entirely in modern Fortran, for all ofthe models considered, which include Fortran DO CONCURRENT,as well as two variants of OpenACC, four variants of OpenMP(2 CPU and 2 GPU), CUDA Fortran, and both loop- andarray-based references. BabelStream Fortran matches the C++implementation as closely as possible, and can be used to makelanguage-based comparisons. This paper represents one of thefirst detailed studies of the performance of Fortran support onheterogeneous architectures; we include results for AArch64 andx86 64 CPUs as well as AMD, Intel and NVIDIA GPU platforms.",

author = "Hammon, {Jeff R} and Tom Deakin and Cownie, {Jim H} and McIntosh-Smith, {Simon N}",

year = "2022",

month = nov,

day = "18",

doi = "10.1109/PMBS56514.2022.00013",

language = "English",

isbn = "978-1-6654-5185-7",

pages = "1--18",

booktitle = "2022 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS)",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

address = "United States",

note = "SC 2022 Workshops International Conference for High Performance Computing, Networking, Storage and Analysis ; Conference date: 13-11-2022 Through 18-11-2022",

}

Hammon, JR, Deakin, T, Cownie, JH & McIntosh-Smith, SN 2022, Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream. in 2022 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS): PMBS22. Institute of Electrical and Electronics Engineers (IEEE), pp. 1-18, SC 2022 Workshops International Conference for High Performance Computing, Networking, Storage and Analysis, Dallas, Texas, United States, 13/11/22. https://doi.org/10.1109/PMBS56514.2022.00013

Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream. / Hammon, Jeff R; Deakin, Tom; Cownie, Jim H et al.
2022 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS): PMBS22. Institute of Electrical and Electronics Engineers (IEEE), 2022. p. 1-18.

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

TY - GEN

T1 - Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream

AU - Hammon, Jeff R

AU - Deakin, Tom

AU - Cownie, Jim H

AU - McIntosh-Smith, Simon N

PY - 2022/11/18

Y1 - 2022/11/18

N2 - Fortran DO CONCURRENT has emerged as a newway to achieve parallel execution of loops on CPUs and GPUs.This paper studies the performance portability of this constructon a range of processors and compares it with the incumbentmodels: OpenMP, OpenACC and CUDA. To do this studyfairly, we implemented the BabelStream memory bandwidthbenchmark from scratch, entirely in modern Fortran, for all ofthe models considered, which include Fortran DO CONCURRENT,as well as two variants of OpenACC, four variants of OpenMP(2 CPU and 2 GPU), CUDA Fortran, and both loop- andarray-based references. BabelStream Fortran matches the C++implementation as closely as possible, and can be used to makelanguage-based comparisons. This paper represents one of thefirst detailed studies of the performance of Fortran support onheterogeneous architectures; we include results for AArch64 andx86 64 CPUs as well as AMD, Intel and NVIDIA GPU platforms.

AB - Fortran DO CONCURRENT has emerged as a newway to achieve parallel execution of loops on CPUs and GPUs.This paper studies the performance portability of this constructon a range of processors and compares it with the incumbentmodels: OpenMP, OpenACC and CUDA. To do this studyfairly, we implemented the BabelStream memory bandwidthbenchmark from scratch, entirely in modern Fortran, for all ofthe models considered, which include Fortran DO CONCURRENT,as well as two variants of OpenACC, four variants of OpenMP(2 CPU and 2 GPU), CUDA Fortran, and both loop- andarray-based references. BabelStream Fortran matches the C++implementation as closely as possible, and can be used to makelanguage-based comparisons. This paper represents one of thefirst detailed studies of the performance of Fortran support onheterogeneous architectures; we include results for AArch64 andx86 64 CPUs as well as AMD, Intel and NVIDIA GPU platforms.

U2 - 10.1109/PMBS56514.2022.00013

DO - 10.1109/PMBS56514.2022.00013

M3 - Conference Contribution (Conference Proceeding)

SN - 978-1-6654-5185-7

SP - 1

EP - 18

BT - 2022 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS)

PB - Institute of Electrical and Electronics Engineers (IEEE)

T2 - SC 2022 Workshops International Conference for High Performance Computing, Networking, Storage and Analysis

Y2 - 13 November 2022 through 18 November 2022

ER -

Hammon JR, Deakin T, Cownie JH, McIntosh-Smith SN. Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream. In 2022 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS): PMBS22. Institute of Electrical and Electronics Engineers (IEEE). 2022. p. 1-18 doi: 10.1109/PMBS56514.2022.00013

Benchmarking Fortran DO CONCURRENT on CPUs and GPUs Using BabelStream

Abstract

Conference

Access to Document

Handle.net

Fingerprint

Cite this