GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models

Tom Deakin; James Price; Matt Martineau; Simon McIntosh-Smith

doi:10.1007/978-3-319-46079-6_34

GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models

Tom Deakin^*, James Price, Matt Martineau, Simon McIntosh-Smith

^*Corresponding author for this work

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

62 Citations (Scopus)

1109 Downloads (Pure)

Abstract

Many scientific codes consist of memory bandwidth bound kernels | the dominating factor of the runtime is the speed at which data can be loaded from memory into the Arithmetic Logic Units, before results are written back to memory. One major advantage of General Purpose Graphics Processing Units (GPGPUs) and other many-core devices such as the Intel Xeon Phi is their focus on providing increased memory bandwidth over traditional CPU architectures. However, as with CPUs, this peak memory bandwidth is usually unachievable in practice and so benchmarks are required to measure a practical upper bound on expected performance.

The choice of one programming model over another should ideally not limit the performance that can be achieved on a device. GPU-STREAM has been updated to incorporate a wide variety of the latest parallel programming models, all implementing the same parallel scheme. As such this tool can be used as a kind of Rosetta Stone which provides both a cross-platform and cross-programming model array of results of achievable memory bandwidth.

Original language	English
Title of host publication	High Performance Computing - ISC High Performance 2016 International Workshops ExaComm, E-MuCoCoS, HPC-IODC, IXPUG, IWOPH, P^3MA, VHPC, WOPSSS, Revised Selected
Editors	Michela Taufer, Bernd Mohr, Julian M Kunkel
Publisher	Springer, Cham
Pages	489-507
Number of pages	19
ISBN (Electronic)	9783319460796
ISBN (Print)	9783319460789
DOIs	https://doi.org/10.1007/978-3-319-46079-6_34
Publication status	Published - 6 Oct 2016
Event	International Workshops on High Performance Computing, ISC High Performance 2016 and Workshop on 2nd International Workshop on Communication Architectures at Extreme Scale, ExaComm 2016, Workshop on Exascale Multi/Many Core Computing Systems, E-MuCoCoS 2016, HPC I/O in the Data Center, HPC-IODC 2016, Application Performance on Intel Xeon Phi – Being Prepared for KNL and Beyond, IXPUG 2016, International Workshop on OpenPOWER for HPC, IWOPH 2016, International Workshop on Performance Portable Programming Models for Accelerators, P^3MA 2016, Workshop on Virtualization in High-Performance Cloud Computing, VHPC 2016, Workshop on Performance and Scalability of Storage Systems, WOPSSS 2016 - Frankfurt, Germany Duration: 19 Jun 2016 → 23 Jun 2016

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9945 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	International Workshops on High Performance Computing, ISC High Performance 2016 and Workshop on 2nd International Workshop on Communication Architectures at Extreme Scale, ExaComm 2016, Workshop on Exascale Multi/Many Core Computing Systems, E-MuCoCoS 2016, HPC I/O in the Data Center, HPC-IODC 2016, Application Performance on Intel Xeon Phi – Being Prepared for KNL and Beyond, IXPUG 2016, International Workshop on OpenPOWER for HPC, IWOPH 2016, International Workshop on Performance Portable Programming Models for Accelerators, P^3MA 2016, Workshop on Virtualization in High-Performance Cloud Computing, VHPC 2016, Workshop on Performance and Scalability of Storage Systems, WOPSSS 2016
Country/Territory	Germany
City	Frankfurt
Period	19/06/16 → 23/06/16

Access to Document

10.1007/978-3-319-46079-6_34

Full-text PDF (accepted author manuscript)
This is the author accepted manuscript (AAM). The final published version (version of record) is available online via Springer at http://link.springer.com/chapter/10.1007%2F978-3-319-46079-6_34. Please refer to any applicable terms of use of the publisher.
Accepted author manuscript, 476 KB

Handle.net

Persistent link

Cite this

Deakin, T., Price, J., Martineau, M., & McIntosh-Smith, S. (2016). GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models. In M. Taufer, B. Mohr, & J. M. Kunkel (Eds.), High Performance Computing - ISC High Performance 2016 International Workshops ExaComm, E-MuCoCoS, HPC-IODC, IXPUG, IWOPH, P^3MA, VHPC, WOPSSS, Revised Selected (pp. 489-507). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9945 LNCS). Springer, Cham. https://doi.org/10.1007/978-3-319-46079-6_34

Deakin, Tom ; Price, James ; Martineau, Matt et al. / GPU-STREAM v2.0 : Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models. High Performance Computing - ISC High Performance 2016 International Workshops ExaComm, E-MuCoCoS, HPC-IODC, IXPUG, IWOPH, P^3MA, VHPC, WOPSSS, Revised Selected. editor / Michela Taufer ; Bernd Mohr ; Julian M Kunkel. Springer, Cham, 2016. pp. 489-507 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{afd3849b6dee4191bcf154b4f55f60b9,

title = "GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models",

abstract = "Many scientific codes consist of memory bandwidth bound kernels | the dominating factor of the runtime is the speed at which data can be loaded from memory into the Arithmetic Logic Units, before results are written back to memory. One major advantage of General Purpose Graphics Processing Units (GPGPUs) and other many-core devices such as the Intel Xeon Phi is their focus on providing increased memory bandwidth over traditional CPU architectures. However, as with CPUs, this peak memory bandwidth is usually unachievable in practice and so benchmarks are required to measure a practical upper bound on expected performance.The choice of one programming model over another should ideally not limit the performance that can be achieved on a device. GPU-STREAM has been updated to incorporate a wide variety of the latest parallel programming models, all implementing the same parallel scheme. As such this tool can be used as a kind of Rosetta Stone which provides both a cross-platform and cross-programming model array of results of achievable memory bandwidth.",

author = "Tom Deakin and James Price and Matt Martineau and Simon McIntosh-Smith",

year = "2016",

month = oct,

day = "6",

doi = "10.1007/978-3-319-46079-6_34",

language = "English",

isbn = "9783319460789",

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

publisher = "Springer, Cham",

pages = "489--507",

editor = "Michela Taufer and Bernd Mohr and Kunkel, {Julian M}",

booktitle = "High Performance Computing - ISC High Performance 2016 International Workshops ExaComm, E-MuCoCoS, HPC-IODC, IXPUG, IWOPH, P^3MA, VHPC, WOPSSS, Revised Selected",

address = "Switzerland",

note = "International Workshops on High Performance Computing, ISC High Performance 2016 and Workshop on 2nd International Workshop on Communication Architectures at Extreme Scale, ExaComm 2016, Workshop on Exascale Multi/Many Core Computing Systems, E-MuCoCoS 2016, HPC I/O in the Data Center, HPC-IODC 2016, Application Performance on Intel Xeon Phi – Being Prepared for KNL and Beyond, IXPUG 2016, International Workshop on OpenPOWER for HPC, IWOPH 2016, International Workshop on Performance Portable Programming Models for Accelerators, P^3MA 2016, Workshop on Virtualization in High-Performance Cloud Computing, VHPC 2016, Workshop on Performance and Scalability of Storage Systems, WOPSSS 2016 ; Conference date: 19-06-2016 Through 23-06-2016",

}

Deakin, T, Price, J, Martineau, M & McIntosh-Smith, S 2016, GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models. in M Taufer, B Mohr & JM Kunkel (eds), High Performance Computing - ISC High Performance 2016 International Workshops ExaComm, E-MuCoCoS, HPC-IODC, IXPUG, IWOPH, P^3MA, VHPC, WOPSSS, Revised Selected. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9945 LNCS, Springer, Cham, pp. 489-507, International Workshops on High Performance Computing, ISC High Performance 2016 and Workshop on 2nd International Workshop on Communication Architectures at Extreme Scale, ExaComm 2016, Workshop on Exascale Multi/Many Core Computing Systems, E-MuCoCoS 2016, HPC I/O in the Data Center, HPC-IODC 2016, Application Performance on Intel Xeon Phi – Being Prepared for KNL and Beyond, IXPUG 2016, International Workshop on OpenPOWER for HPC, IWOPH 2016, International Workshop on Performance Portable Programming Models for Accelerators, P^3MA 2016, Workshop on Virtualization in High-Performance Cloud Computing, VHPC 2016, Workshop on Performance and Scalability of Storage Systems, WOPSSS 2016, Frankfurt, Germany, 19/06/16. https://doi.org/10.1007/978-3-319-46079-6_34

GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models. / Deakin, Tom; Price, James; Martineau, Matt et al.
High Performance Computing - ISC High Performance 2016 International Workshops ExaComm, E-MuCoCoS, HPC-IODC, IXPUG, IWOPH, P^3MA, VHPC, WOPSSS, Revised Selected. ed. / Michela Taufer; Bernd Mohr; Julian M Kunkel. Springer, Cham, 2016. p. 489-507 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9945 LNCS).

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

TY - GEN

T1 - GPU-STREAM v2.0

T2 - International Workshops on High Performance Computing, ISC High Performance 2016 and Workshop on 2nd International Workshop on Communication Architectures at Extreme Scale, ExaComm 2016, Workshop on Exascale Multi/Many Core Computing Systems, E-MuCoCoS 2016, HPC I/O in the Data Center, HPC-IODC 2016, Application Performance on Intel Xeon Phi – Being Prepared for KNL and Beyond, IXPUG 2016, International Workshop on OpenPOWER for HPC, IWOPH 2016, International Workshop on Performance Portable Programming Models for Accelerators, P^3MA 2016, Workshop on Virtualization in High-Performance Cloud Computing, VHPC 2016, Workshop on Performance and Scalability of Storage Systems, WOPSSS 2016

AU - Deakin, Tom

AU - Price, James

AU - Martineau, Matt

AU - McIntosh-Smith, Simon

PY - 2016/10/6

Y1 - 2016/10/6

N2 - Many scientific codes consist of memory bandwidth bound kernels | the dominating factor of the runtime is the speed at which data can be loaded from memory into the Arithmetic Logic Units, before results are written back to memory. One major advantage of General Purpose Graphics Processing Units (GPGPUs) and other many-core devices such as the Intel Xeon Phi is their focus on providing increased memory bandwidth over traditional CPU architectures. However, as with CPUs, this peak memory bandwidth is usually unachievable in practice and so benchmarks are required to measure a practical upper bound on expected performance.The choice of one programming model over another should ideally not limit the performance that can be achieved on a device. GPU-STREAM has been updated to incorporate a wide variety of the latest parallel programming models, all implementing the same parallel scheme. As such this tool can be used as a kind of Rosetta Stone which provides both a cross-platform and cross-programming model array of results of achievable memory bandwidth.

AB - Many scientific codes consist of memory bandwidth bound kernels | the dominating factor of the runtime is the speed at which data can be loaded from memory into the Arithmetic Logic Units, before results are written back to memory. One major advantage of General Purpose Graphics Processing Units (GPGPUs) and other many-core devices such as the Intel Xeon Phi is their focus on providing increased memory bandwidth over traditional CPU architectures. However, as with CPUs, this peak memory bandwidth is usually unachievable in practice and so benchmarks are required to measure a practical upper bound on expected performance.The choice of one programming model over another should ideally not limit the performance that can be achieved on a device. GPU-STREAM has been updated to incorporate a wide variety of the latest parallel programming models, all implementing the same parallel scheme. As such this tool can be used as a kind of Rosetta Stone which provides both a cross-platform and cross-programming model array of results of achievable memory bandwidth.

UR - http://www.scopus.com/inward/record.url?scp=84992603317&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-46079-6_34

DO - 10.1007/978-3-319-46079-6_34

M3 - Conference Contribution (Conference Proceeding)

AN - SCOPUS:84992603317

SN - 9783319460789

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

SP - 489

EP - 507

BT - High Performance Computing - ISC High Performance 2016 International Workshops ExaComm, E-MuCoCoS, HPC-IODC, IXPUG, IWOPH, P^3MA, VHPC, WOPSSS, Revised Selected

A2 - Taufer, Michela

A2 - Mohr, Bernd

A2 - Kunkel, Julian M

PB - Springer, Cham

Y2 - 19 June 2016 through 23 June 2016

ER -

Deakin T, Price J, Martineau M, McIntosh-Smith S. GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models. In Taufer M, Mohr B, Kunkel JM, editors, High Performance Computing - ISC High Performance 2016 International Workshops ExaComm, E-MuCoCoS, HPC-IODC, IXPUG, IWOPH, P^3MA, VHPC, WOPSSS, Revised Selected. Springer, Cham. 2016. p. 489-507. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-46079-6_34

GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models

Abstract

Publication series

Conference

Access to Document

Handle.net

Other files and links

Fingerprint

HPC (High Performance Computing) and HTC (High Throughput Computing) Facilities

Cite this

GPU-STREAM v2.0: Benchmarking the achievable memory bandwidth of many-core processors across diverse parallel programming models

Abstract

Publication series

Conference

Access to Document

Handle.net

Other files and links

Fingerprint

Equipment

HPC (High Performance Computing) and HTC (High Throughput Computing) Facilities

Cite this