Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application

Istvan Z. Reguly; Andrew M.B. Owenson; Archie Powell; Stephen A. Jarvis; Gihan R. Mudalige

doi:10.1007/978-3-030-78713-4_21

Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application

Istvan Z. Reguly^*, Andrew M.B. Owenson, Archie Powell, Stephen A. Jarvis, Gihan R. Mudalige

^*Corresponding author for this work

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

Abstract

As the computing hardware landscape gets more diverse, and the complexity of hardware grows, the need for a general purpose parallel programming model capable of developing (performance) portable codes have become highly attractive. Intel’s OneAPI suite, which is based on the SYCL standard aims to fill this gap using a modern C++ API. In this paper, we use SYCL to parallelize MG-CFD, an unstructured-mesh computational fluid dynamics (CFD) code, to explore current performance of SYCL. The code is benchmarked on several modern processor systems from Intel (including CPUs and the latest Xe LP GPU), AMD, ARM and Nvidia, making use of a variety of current SYCL compilers, with a particular focus on OneAPI and how it maps to Intel’s CPU and GPU architectures. We compare performance with other parallelizations available in OP2, including SIMD, OpenMP, MPI and CUDA. The results are mixed; the performance of this class of applications, when parallelized with SYCL, highly depends on the target architecture and the compiler, but in many cases comes close to the performance of currently prevalent parallel programming models. However, it still requires different parallelization strategies or code-paths be written for different hardware to obtain the best performance.

Original language	English
Title of host publication	High Performance Computing
Subtitle of host publication	36th International Conference, ISC High Performance 2021, Virtual Event, June 24 – July 2, 2021, Proceedings
Editors	Bradford L. Chamberlain, Ana-Lucia Varbanescu, Hatem Ltaief, Piotr Luszczek
Publisher	Springer
Pages	391-410
Number of pages	20
Edition	1
ISBN (Electronic)	9783030787134
ISBN (Print)	9783030787127
DOIs	https://doi.org/10.1007/978-3-030-78713-4_21
Publication status	Published - 17 Jun 2021
Event	36th International Conference on High Performance Computing, ISC High Performance 2021 - Virtual, Online Duration: 24 Jun 2021 → 2 Jul 2021

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12728
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	36th International Conference on High Performance Computing, ISC High Performance 2021
City	Virtual, Online
Period	24/06/21 → 2/07/21

Bibliographical note

Funding Information:
Acknowledgment. This research is supported by Rolls-Royce plc., and by the UK Engineering and Physical Sciences Research Council (EPSRC): (EP/S005072/1 – Strategic Partnership in Computational Science for Advanced Simulation and Modelling of Engineering Systems – ASiMoV). Gihan Mudalige was supported by the Royal Society Industry Fellowship Scheme(INF/R1/1800 12). István Reguly was supported by National Research, Development and Innovation Fund of Hungary, project PD 124905, financed under the PD 17 funding scheme.

Publisher Copyright:
© 2021, Springer Nature Switzerland AG.

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-030-78713-4_21

Cite this

Reguly, I. Z., Owenson, A. M. B., Powell, A., Jarvis, S. A., & Mudalige, G. R. (2021). Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application. In B. L. Chamberlain, A.-L. Varbanescu, H. Ltaief, & P. Luszczek (Eds.), High Performance Computing: 36th International Conference, ISC High Performance 2021, Virtual Event, June 24 – July 2, 2021, Proceedings (1 ed., pp. 391-410). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12728). Springer. https://doi.org/10.1007/978-3-030-78713-4_21

Reguly, Istvan Z. ; Owenson, Andrew M.B. ; Powell, Archie et al. / Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application. High Performance Computing: 36th International Conference, ISC High Performance 2021, Virtual Event, June 24 – July 2, 2021, Proceedings. editor / Bradford L. Chamberlain ; Ana-Lucia Varbanescu ; Hatem Ltaief ; Piotr Luszczek. 1. ed. Springer, 2021. pp. 391-410 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{b6fb2a1b3f8c4d60a6a954f36086e58b,

title = "Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application",

abstract = "As the computing hardware landscape gets more diverse, and the complexity of hardware grows, the need for a general purpose parallel programming model capable of developing (performance) portable codes have become highly attractive. Intel{\textquoteright}s OneAPI suite, which is based on the SYCL standard aims to fill this gap using a modern C++ API. In this paper, we use SYCL to parallelize MG-CFD, an unstructured-mesh computational fluid dynamics (CFD) code, to explore current performance of SYCL. The code is benchmarked on several modern processor systems from Intel (including CPUs and the latest Xe LP GPU), AMD, ARM and Nvidia, making use of a variety of current SYCL compilers, with a particular focus on OneAPI and how it maps to Intel{\textquoteright}s CPU and GPU architectures. We compare performance with other parallelizations available in OP2, including SIMD, OpenMP, MPI and CUDA. The results are mixed; the performance of this class of applications, when parallelized with SYCL, highly depends on the target architecture and the compiler, but in many cases comes close to the performance of currently prevalent parallel programming models. However, it still requires different parallelization strategies or code-paths be written for different hardware to obtain the best performance.",

author = "Reguly, {Istvan Z.} and Owenson, {Andrew M.B.} and Archie Powell and Jarvis, {Stephen A.} and Mudalige, {Gihan R.}",

note = "Funding Information: Acknowledgment. This research is supported by Rolls-Royce plc., and by the UK Engineering and Physical Sciences Research Council (EPSRC): (EP/S005072/1 – Strategic Partnership in Computational Science for Advanced Simulation and Modelling of Engineering Systems – ASiMoV). Gihan Mudalige was supported by the Royal Society Industry Fellowship Scheme(INF/R1/1800 12). Istv{\'a}n Reguly was supported by National Research, Development and Innovation Fund of Hungary, project PD 124905, financed under the PD 17 funding scheme. Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; 36th International Conference on High Performance Computing, ISC High Performance 2021 ; Conference date: 24-06-2021 Through 02-07-2021",

year = "2021",

month = jun,

day = "17",

doi = "10.1007/978-3-030-78713-4_21",

language = "English",

isbn = "9783030787127",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "391--410",

editor = "Chamberlain, {Bradford L.} and Ana-Lucia Varbanescu and Hatem Ltaief and Piotr Luszczek",

booktitle = "High Performance Computing",

edition = "1",

}

Reguly, IZ, Owenson, AMB, Powell, A, Jarvis, SA & Mudalige, GR 2021, Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application. in BL Chamberlain, A-L Varbanescu, H Ltaief & P Luszczek (eds), High Performance Computing: 36th International Conference, ISC High Performance 2021, Virtual Event, June 24 – July 2, 2021, Proceedings. 1 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12728, Springer, pp. 391-410, 36th International Conference on High Performance Computing, ISC High Performance 2021, Virtual, Online, 24/06/21. https://doi.org/10.1007/978-3-030-78713-4_21

Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application. / Reguly, Istvan Z.; Owenson, Andrew M.B.; Powell, Archie et al.
High Performance Computing: 36th International Conference, ISC High Performance 2021, Virtual Event, June 24 – July 2, 2021, Proceedings. ed. / Bradford L. Chamberlain; Ana-Lucia Varbanescu; Hatem Ltaief; Piotr Luszczek. 1. ed. Springer, 2021. p. 391-410 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12728).

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

TY - GEN

T1 - Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application

AU - Reguly, Istvan Z.

AU - Owenson, Andrew M.B.

AU - Powell, Archie

AU - Jarvis, Stephen A.

AU - Mudalige, Gihan R.

N1 - Funding Information: Acknowledgment. This research is supported by Rolls-Royce plc., and by the UK Engineering and Physical Sciences Research Council (EPSRC): (EP/S005072/1 – Strategic Partnership in Computational Science for Advanced Simulation and Modelling of Engineering Systems – ASiMoV). Gihan Mudalige was supported by the Royal Society Industry Fellowship Scheme(INF/R1/1800 12). István Reguly was supported by National Research, Development and Innovation Fund of Hungary, project PD 124905, financed under the PD 17 funding scheme. Publisher Copyright: © 2021, Springer Nature Switzerland AG.

PY - 2021/6/17

Y1 - 2021/6/17

N2 - As the computing hardware landscape gets more diverse, and the complexity of hardware grows, the need for a general purpose parallel programming model capable of developing (performance) portable codes have become highly attractive. Intel’s OneAPI suite, which is based on the SYCL standard aims to fill this gap using a modern C++ API. In this paper, we use SYCL to parallelize MG-CFD, an unstructured-mesh computational fluid dynamics (CFD) code, to explore current performance of SYCL. The code is benchmarked on several modern processor systems from Intel (including CPUs and the latest Xe LP GPU), AMD, ARM and Nvidia, making use of a variety of current SYCL compilers, with a particular focus on OneAPI and how it maps to Intel’s CPU and GPU architectures. We compare performance with other parallelizations available in OP2, including SIMD, OpenMP, MPI and CUDA. The results are mixed; the performance of this class of applications, when parallelized with SYCL, highly depends on the target architecture and the compiler, but in many cases comes close to the performance of currently prevalent parallel programming models. However, it still requires different parallelization strategies or code-paths be written for different hardware to obtain the best performance.

AB - As the computing hardware landscape gets more diverse, and the complexity of hardware grows, the need for a general purpose parallel programming model capable of developing (performance) portable codes have become highly attractive. Intel’s OneAPI suite, which is based on the SYCL standard aims to fill this gap using a modern C++ API. In this paper, we use SYCL to parallelize MG-CFD, an unstructured-mesh computational fluid dynamics (CFD) code, to explore current performance of SYCL. The code is benchmarked on several modern processor systems from Intel (including CPUs and the latest Xe LP GPU), AMD, ARM and Nvidia, making use of a variety of current SYCL compilers, with a particular focus on OneAPI and how it maps to Intel’s CPU and GPU architectures. We compare performance with other parallelizations available in OP2, including SIMD, OpenMP, MPI and CUDA. The results are mixed; the performance of this class of applications, when parallelized with SYCL, highly depends on the target architecture and the compiler, but in many cases comes close to the performance of currently prevalent parallel programming models. However, it still requires different parallelization strategies or code-paths be written for different hardware to obtain the best performance.

UR - http://www.scopus.com/inward/record.url?scp=85111467524&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-78713-4_21

DO - 10.1007/978-3-030-78713-4_21

M3 - Conference contribution

AN - SCOPUS:85111467524

SN - 9783030787127

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 391

EP - 410

BT - High Performance Computing

A2 - Chamberlain, Bradford L.

A2 - Varbanescu, Ana-Lucia

A2 - Ltaief, Hatem

A2 - Luszczek, Piotr

PB - Springer

T2 - 36th International Conference on High Performance Computing, ISC High Performance 2021

Y2 - 24 June 2021 through 2 July 2021

ER -

Reguly IZ, Owenson AMB, Powell A, Jarvis SA, Mudalige GR. Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application. In Chamberlain BL, Varbanescu AL, Ltaief H, Luszczek P, editors, High Performance Computing: 36th International Conference, ISC High Performance 2021, Virtual Event, June 24 – July 2, 2021, Proceedings. 1 ed. Springer. 2021. p. 391-410. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-78713-4_21

Under the hood of SYCL – an initial performance analysis with an unstructured-mesh CFD application

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this