Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations

Jiri Jaros; Bradley E. Treeby; Alistair P. Rendell

Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations

Jiri Jaros^*, Bradley E. Treeby, Alistair P. Rendell

^*Corresponding author for this work

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

8 Citations (Scopus)

Abstract

This paper outlines our effort to migrate a compute intensive application of ultrasound propagation being developed in Matlab to a cluster computer where each node has seven GPUs. Our goal is to perform realistic simulations in hours and minutes instead of weeks and days. In order to reach this goal we investigate architecture characteristics of the target system focusing on the PCI-Express subsystem and new features proposed in CUDA version 4.0, especially simultaneous host to device, device to host and peer-to-peer transfers that the application is going to highly benefit from. We also present the results from a CPU based implementation and discuss future directions to exploit multiple GPUs.

Original language	English
Title of host publication	Parallel and Distributed Computing 2012 - Proceedings of the Tenth Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012
Pages	43-52
Number of pages	10
Publication status	Published - 2012
Event	10th Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012 - Melbourne, VIC, Australia Duration: 31 Jan 2012 → 3 Feb 2012

Publication series

Name	Conferences in Research and Practice in Information Technology Series
Volume	127
ISSN (Print)	1445-1336

Conference

Conference	10th Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012
Country/Territory	Australia
City	Melbourne, VIC
Period	31/01/12 → 3/02/12

Cite this

Jaros, J., Treeby, B. E., & Rendell, A. P. (2012). Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations. In Parallel and Distributed Computing 2012 - Proceedings of the Tenth Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012 (pp. 43-52). (Conferences in Research and Practice in Information Technology Series; Vol. 127).

Jaros, Jiri ; Treeby, Bradley E. ; Rendell, Alistair P. / Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations. Parallel and Distributed Computing 2012 - Proceedings of the Tenth Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012. 2012. pp. 43-52 (Conferences in Research and Practice in Information Technology Series).

@inproceedings{7ef044a6027440a5ba372a67972e38f9,

title = "Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations",

abstract = "This paper outlines our effort to migrate a compute intensive application of ultrasound propagation being developed in Matlab to a cluster computer where each node has seven GPUs. Our goal is to perform realistic simulations in hours and minutes instead of weeks and days. In order to reach this goal we investigate architecture characteristics of the target system focusing on the PCI-Express subsystem and new features proposed in CUDA version 4.0, especially simultaneous host to device, device to host and peer-to-peer transfers that the application is going to highly benefit from. We also present the results from a CPU based implementation and discuss future directions to exploit multiple GPUs.",

keywords = "7-GPU system, Bandwidth, CUDA, FFT, Matlab, Multi-core, PCI-Express, Ultrasound simulation",

author = "Jiri Jaros and Treeby, {Bradley E.} and Rendell, {Alistair P.}",

year = "2012",

language = "English",

isbn = "9781921770081",

series = "Conferences in Research and Practice in Information Technology Series",

pages = "43--52",

booktitle = "Parallel and Distributed Computing 2012 - Proceedings of the Tenth Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012",

note = "10th Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012 ; Conference date: 31-01-2012 Through 03-02-2012",

}

Jaros, J, Treeby, BE & Rendell, AP 2012, Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations. in Parallel and Distributed Computing 2012 - Proceedings of the Tenth Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012. Conferences in Research and Practice in Information Technology Series, vol. 127, pp. 43-52, 10th Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012, Melbourne, VIC, Australia, 31/01/12.

Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations. / Jaros, Jiri; Treeby, Bradley E.; Rendell, Alistair P.
Parallel and Distributed Computing 2012 - Proceedings of the Tenth Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012. 2012. p. 43-52 (Conferences in Research and Practice in Information Technology Series; Vol. 127).

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

TY - GEN

T1 - Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations

AU - Jaros, Jiri

AU - Treeby, Bradley E.

AU - Rendell, Alistair P.

PY - 2012

Y1 - 2012

N2 - This paper outlines our effort to migrate a compute intensive application of ultrasound propagation being developed in Matlab to a cluster computer where each node has seven GPUs. Our goal is to perform realistic simulations in hours and minutes instead of weeks and days. In order to reach this goal we investigate architecture characteristics of the target system focusing on the PCI-Express subsystem and new features proposed in CUDA version 4.0, especially simultaneous host to device, device to host and peer-to-peer transfers that the application is going to highly benefit from. We also present the results from a CPU based implementation and discuss future directions to exploit multiple GPUs.

AB - This paper outlines our effort to migrate a compute intensive application of ultrasound propagation being developed in Matlab to a cluster computer where each node has seven GPUs. Our goal is to perform realistic simulations in hours and minutes instead of weeks and days. In order to reach this goal we investigate architecture characteristics of the target system focusing on the PCI-Express subsystem and new features proposed in CUDA version 4.0, especially simultaneous host to device, device to host and peer-to-peer transfers that the application is going to highly benefit from. We also present the results from a CPU based implementation and discuss future directions to exploit multiple GPUs.

KW - 7-GPU system

KW - Bandwidth

KW - CUDA

KW - FFT

KW - Matlab

KW - Multi-core

KW - PCI-Express

KW - Ultrasound simulation

UR - http://www.scopus.com/inward/record.url?scp=84869098403&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9781921770081

T3 - Conferences in Research and Practice in Information Technology Series

SP - 43

EP - 52

BT - Parallel and Distributed Computing 2012 - Proceedings of the Tenth Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012

T2 - 10th Australasian Symposium on Parallel and Distributed Computing, AusPDC 2012

Y2 - 31 January 2012 through 3 February 2012

ER -

Use of multiple GPUs on shared memory multiprocessors for ultrasound propagation simulations

Abstract

Publication series

Conference

Other files and links

Fingerprint

Cite this