Evaluating contexts in OpenSHMEM-X reference implementation

Aurelien Bouteiller; Swaroop Pophale; Swen Boehm; Matthew B. Baker; Manjunath Gorentla Venkata

doi:10.1007/978-3-319-73814-7_4

Evaluating contexts in OpenSHMEM-X reference implementation

Aurelien Bouteiller, Swaroop Pophale, Swen Boehm, Matthew B. Baker, Manjunath Gorentla Venkata

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

6 Scopus citations

Abstract

Many-core processors are now ubiquitous in supercomputing. This evolution pushes toward the adoption of mixed models in which cores are exploited with threading models (and related programming abstractions, such as OpenMP), while communication between distributed memory domains employ a communication Application Programming Interface (API). OpenSHMEM is a partitioned global address space communication specification that exposes one-sided and synchronization operations. As the threaded semantics of OpenSHMEM are being fleshed out by its standardization committee, it is important to assess the soundness of the proposed concepts. This paper implements and evaluate the “context” extension in relation to threaded operations. We discuss the implementation challenges of the context and the associated API in OpenSHMEM-X. We then evaluate its performance in threaded situations on the Infiniband network using micro-benchmarks and the Random Access benchmark and see that adding communication contexts significantly improves message rate achievable by the executing multi-threaded PEs.

Original language	English
Title of host publication	OpenSHMEM and Related Technologies
Subtitle of host publication	Big Compute and Big Data Convergence - 4th Workshop, OpenSHMEM 2017, Revised Selected Papers
Editors	Manjunath Gorentla Venkata, Neena Imam, Swaroop Pophale
Publisher	Springer Verlag
Pages	50-62
Number of pages	13
ISBN (Print)	9783319738130
DOIs	https://doi.org/10.1007/978-3-319-73814-7_4
State	Published - 2018
Event	4th Workshop on OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence, OpenSHMEM 2017 - Annapolis, United States Duration: Aug 7 2017 → Aug 9 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10679 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	4th Workshop on OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence, OpenSHMEM 2017
Country/Territory	United States
City	Annapolis
Period	08/7/17 → 08/9/17

Funding

A. Bouteiller and S. Pophale—Contributed equally. This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy 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). Acknowledgment. This work is supported by the United States Department of Defense and used resources of the Extreme Scale Systems Center located at the Oak Ridge National Laboratory.

Keywords

Interoperability
PGAS
Shared memory

Access to Document

10.1007/978-3-319-73814-7_4

Cite this

Bouteiller, A., Pophale, S., Boehm, S., Baker, M. B., & Venkata, M. G. (2018). Evaluating contexts in OpenSHMEM-X reference implementation. In M. Gorentla Venkata, N. Imam, & S. Pophale (Eds.), OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence - 4th Workshop, OpenSHMEM 2017, Revised Selected Papers (pp. 50-62). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10679 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-73814-7_4

Bouteiller, Aurelien ; Pophale, Swaroop ; Boehm, Swen et al. / Evaluating contexts in OpenSHMEM-X reference implementation. OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence - 4th Workshop, OpenSHMEM 2017, Revised Selected Papers. editor / Manjunath Gorentla Venkata ; Neena Imam ; Swaroop Pophale. Springer Verlag, 2018. pp. 50-62 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{f22b8816fe89470f9e7e6063a1966381,

title = "Evaluating contexts in OpenSHMEM-X reference implementation",

abstract = "Many-core processors are now ubiquitous in supercomputing. This evolution pushes toward the adoption of mixed models in which cores are exploited with threading models (and related programming abstractions, such as OpenMP), while communication between distributed memory domains employ a communication Application Programming Interface (API). OpenSHMEM is a partitioned global address space communication specification that exposes one-sided and synchronization operations. As the threaded semantics of OpenSHMEM are being fleshed out by its standardization committee, it is important to assess the soundness of the proposed concepts. This paper implements and evaluate the “context” extension in relation to threaded operations. We discuss the implementation challenges of the context and the associated API in OpenSHMEM-X. We then evaluate its performance in threaded situations on the Infiniband network using micro-benchmarks and the Random Access benchmark and see that adding communication contexts significantly improves message rate achievable by the executing multi-threaded PEs.",

keywords = "Interoperability, PGAS, Shared memory",

author = "Aurelien Bouteiller and Swaroop Pophale and Swen Boehm and Baker, {Matthew B.} and Venkata, {Manjunath Gorentla}",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG 2018.; 4th Workshop on OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence, OpenSHMEM 2017 ; Conference date: 07-08-2017 Through 09-08-2017",

year = "2018",

doi = "10.1007/978-3-319-73814-7_4",

language = "English",

isbn = "9783319738130",

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

publisher = "Springer Verlag",

pages = "50--62",

editor = "{Gorentla Venkata}, Manjunath and Neena Imam and Swaroop Pophale",

booktitle = "OpenSHMEM and Related Technologies",

}

Bouteiller, A, Pophale, S , Boehm, S, Baker, MB & Venkata, MG 2018, Evaluating contexts in OpenSHMEM-X reference implementation. in M Gorentla Venkata, N Imam & S Pophale (eds), OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence - 4th Workshop, OpenSHMEM 2017, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10679 LNCS, Springer Verlag, pp. 50-62, 4th Workshop on OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence, OpenSHMEM 2017, Annapolis, United States, 08/7/17. https://doi.org/10.1007/978-3-319-73814-7_4

Evaluating contexts in OpenSHMEM-X reference implementation. / Bouteiller, Aurelien; Pophale, Swaroop ; Boehm, Swen et al.
OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence - 4th Workshop, OpenSHMEM 2017, Revised Selected Papers. ed. / Manjunath Gorentla Venkata; Neena Imam; Swaroop Pophale. Springer Verlag, 2018. p. 50-62 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10679 LNCS).

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

TY - GEN

T1 - Evaluating contexts in OpenSHMEM-X reference implementation

AU - Bouteiller, Aurelien

AU - Pophale, Swaroop

AU - Boehm, Swen

AU - Baker, Matthew B.

AU - Venkata, Manjunath Gorentla

N1 - Publisher Copyright: © Springer International Publishing AG 2018.

PY - 2018

Y1 - 2018

N2 - Many-core processors are now ubiquitous in supercomputing. This evolution pushes toward the adoption of mixed models in which cores are exploited with threading models (and related programming abstractions, such as OpenMP), while communication between distributed memory domains employ a communication Application Programming Interface (API). OpenSHMEM is a partitioned global address space communication specification that exposes one-sided and synchronization operations. As the threaded semantics of OpenSHMEM are being fleshed out by its standardization committee, it is important to assess the soundness of the proposed concepts. This paper implements and evaluate the “context” extension in relation to threaded operations. We discuss the implementation challenges of the context and the associated API in OpenSHMEM-X. We then evaluate its performance in threaded situations on the Infiniband network using micro-benchmarks and the Random Access benchmark and see that adding communication contexts significantly improves message rate achievable by the executing multi-threaded PEs.

AB - Many-core processors are now ubiquitous in supercomputing. This evolution pushes toward the adoption of mixed models in which cores are exploited with threading models (and related programming abstractions, such as OpenMP), while communication between distributed memory domains employ a communication Application Programming Interface (API). OpenSHMEM is a partitioned global address space communication specification that exposes one-sided and synchronization operations. As the threaded semantics of OpenSHMEM are being fleshed out by its standardization committee, it is important to assess the soundness of the proposed concepts. This paper implements and evaluate the “context” extension in relation to threaded operations. We discuss the implementation challenges of the context and the associated API in OpenSHMEM-X. We then evaluate its performance in threaded situations on the Infiniband network using micro-benchmarks and the Random Access benchmark and see that adding communication contexts significantly improves message rate achievable by the executing multi-threaded PEs.

KW - Interoperability

KW - PGAS

KW - Shared memory

UR - http://www.scopus.com/inward/record.url?scp=85041534671&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-73814-7_4

DO - 10.1007/978-3-319-73814-7_4

M3 - Conference contribution

AN - SCOPUS:85041534671

SN - 9783319738130

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

SP - 50

EP - 62

BT - OpenSHMEM and Related Technologies

A2 - Gorentla Venkata, Manjunath

A2 - Imam, Neena

A2 - Pophale, Swaroop

PB - Springer Verlag

T2 - 4th Workshop on OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence, OpenSHMEM 2017

Y2 - 7 August 2017 through 9 August 2017

ER -

Bouteiller A, Pophale S , Boehm S, Baker MB, Venkata MG. Evaluating contexts in OpenSHMEM-X reference implementation. In Gorentla Venkata M, Imam N, Pophale S, editors, OpenSHMEM and Related Technologies: Big Compute and Big Data Convergence - 4th Workshop, OpenSHMEM 2017, Revised Selected Papers. Springer Verlag. 2018. p. 50-62. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-73814-7_4

Evaluating contexts in OpenSHMEM-X reference implementation

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this