Characterizing the Performance of Executing Many-tasks on Summit

Matteo Turilli; Andre Merzky; Thomas Naughton; Wael Elwasif; Shantenu Jha

doi:10.1109/IPDRM49579.2019.00007

Characterizing the Performance of Executing Many-tasks on Summit

Matteo Turilli, Andre Merzky, Thomas Naughton, Wael Elwasif, Shantenu Jha

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

11 Scopus citations

Abstract

Many scientific workloads are comprised of many tasks, where each task is an independent simulation or analysis of data. The execution of millions of tasks on heterogeneous HPC platforms requires scalable dynamic resource management and multi-level scheduling. RADICAL-Pilot (RP) - an implementation of the Pilot abstraction, addresses these challenges and serves as an effective runtime system to execute workloads comprised of many tasks. In this paper, we characterize the performance of executing many tasks using RP when interfaced with JSM and PRRTE on Summit: RP is responsible for resource management and task scheduling on acquired resource; JSM or PRRTE enact the placement of launching of scheduled tasks. Our experiments provide lower bounds on the performance of RP when integrated with JSM and PRRTE. Specifically, for workloads comprised of homogeneous single-core, 15 minutes-long tasks we find that: PRRTE scales better than JSM for > O(1000) tasks; PRRTE overheads are negligible; and PRRTE supports optimizations that lower the impact of overheads and enable resource utilization of 63% when executing O(16K), 1-core tasks over 404 compute nodes.

Original language	English
Title of host publication	Proceedings of IPDRM 2019
Subtitle of host publication	3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	18-25
Number of pages	8
ISBN (Electronic)	9781728159935
DOIs	https://doi.org/10.1109/IPDRM49579.2019.00007
State	Published - Nov 2019
Event	3rd IEEE/ACM Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware, IPDRM 2019 - Denver, United States Duration: Nov 22 2019 → …

Publication series

Name	Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis

Conference

Conference	3rd IEEE/ACM Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware, IPDRM 2019
Country/Territory	United States
City	Denver
Period	11/22/19 → …

Funding

We would like the thank other members of the PMIx community, and Ralph Castain in particular, for the excellent work that we build upon. This research used resources of the Oak Ridge Leadership Computing Facility at the Oak Ridge National Laboratory, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC05-00OR22725. At Rutgers, this work was also supported by NSF “CAREER” ACI-1253644, RADICAL-Cybertools NSF 1440677 and 1931512, and DOE Award DE-SC0016280. We also acknowledge DOE INCITE awards for allocations on Summit.

Keywords

Data-Vortex;-irregular-application;-high-performance-computing

Access to Document

10.1109/IPDRM49579.2019.00007

Cite this

Turilli, M., Merzky, A., Naughton, T., Elwasif, W., & Jha, S. (2019). Characterizing the Performance of Executing Many-tasks on Summit. In Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis (pp. 18-25). Article 8956483 (Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IPDRM49579.2019.00007

Turilli, Matteo ; Merzky, Andre ; Naughton, Thomas et al. / Characterizing the Performance of Executing Many-tasks on Summit. Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 18-25 (Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis).

@inproceedings{2fdb615e3c2f4361809136be1cdaf894,

title = "Characterizing the Performance of Executing Many-tasks on Summit",

abstract = "Many scientific workloads are comprised of many tasks, where each task is an independent simulation or analysis of data. The execution of millions of tasks on heterogeneous HPC platforms requires scalable dynamic resource management and multi-level scheduling. RADICAL-Pilot (RP) - an implementation of the Pilot abstraction, addresses these challenges and serves as an effective runtime system to execute workloads comprised of many tasks. In this paper, we characterize the performance of executing many tasks using RP when interfaced with JSM and PRRTE on Summit: RP is responsible for resource management and task scheduling on acquired resource; JSM or PRRTE enact the placement of launching of scheduled tasks. Our experiments provide lower bounds on the performance of RP when integrated with JSM and PRRTE. Specifically, for workloads comprised of homogeneous single-core, 15 minutes-long tasks we find that: PRRTE scales better than JSM for > O(1000) tasks; PRRTE overheads are negligible; and PRRTE supports optimizations that lower the impact of overheads and enable resource utilization of 63% when executing O(16K), 1-core tasks over 404 compute nodes.",

keywords = "Data-Vortex;-irregular-application;-high-performance-computing",

author = "Matteo Turilli and Andre Merzky and Thomas Naughton and Wael Elwasif and Shantenu Jha",

note = "Publisher Copyright: {\textcopyright} 2019 IEEE.; 3rd IEEE/ACM Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware, IPDRM 2019 ; Conference date: 22-11-2019",

year = "2019",

month = nov,

doi = "10.1109/IPDRM49579.2019.00007",

language = "English",

series = "Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "18--25",

booktitle = "Proceedings of IPDRM 2019",

}

Turilli, M, Merzky, A, Naughton, T , Elwasif, W & Jha, S 2019, Characterizing the Performance of Executing Many-tasks on Summit. in Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis., 8956483, Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis, Institute of Electrical and Electronics Engineers Inc., pp. 18-25, 3rd IEEE/ACM Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware, IPDRM 2019, Denver, United States, 11/22/19. https://doi.org/10.1109/IPDRM49579.2019.00007

Characterizing the Performance of Executing Many-tasks on Summit. / Turilli, Matteo; Merzky, Andre; Naughton, Thomas et al.
Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc., 2019. p. 18-25 8956483 (Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis).

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

TY - GEN

T1 - Characterizing the Performance of Executing Many-tasks on Summit

AU - Turilli, Matteo

AU - Merzky, Andre

AU - Naughton, Thomas

AU - Elwasif, Wael

AU - Jha, Shantenu

PY - 2019/11

Y1 - 2019/11

N2 - Many scientific workloads are comprised of many tasks, where each task is an independent simulation or analysis of data. The execution of millions of tasks on heterogeneous HPC platforms requires scalable dynamic resource management and multi-level scheduling. RADICAL-Pilot (RP) - an implementation of the Pilot abstraction, addresses these challenges and serves as an effective runtime system to execute workloads comprised of many tasks. In this paper, we characterize the performance of executing many tasks using RP when interfaced with JSM and PRRTE on Summit: RP is responsible for resource management and task scheduling on acquired resource; JSM or PRRTE enact the placement of launching of scheduled tasks. Our experiments provide lower bounds on the performance of RP when integrated with JSM and PRRTE. Specifically, for workloads comprised of homogeneous single-core, 15 minutes-long tasks we find that: PRRTE scales better than JSM for > O(1000) tasks; PRRTE overheads are negligible; and PRRTE supports optimizations that lower the impact of overheads and enable resource utilization of 63% when executing O(16K), 1-core tasks over 404 compute nodes.

AB - Many scientific workloads are comprised of many tasks, where each task is an independent simulation or analysis of data. The execution of millions of tasks on heterogeneous HPC platforms requires scalable dynamic resource management and multi-level scheduling. RADICAL-Pilot (RP) - an implementation of the Pilot abstraction, addresses these challenges and serves as an effective runtime system to execute workloads comprised of many tasks. In this paper, we characterize the performance of executing many tasks using RP when interfaced with JSM and PRRTE on Summit: RP is responsible for resource management and task scheduling on acquired resource; JSM or PRRTE enact the placement of launching of scheduled tasks. Our experiments provide lower bounds on the performance of RP when integrated with JSM and PRRTE. Specifically, for workloads comprised of homogeneous single-core, 15 minutes-long tasks we find that: PRRTE scales better than JSM for > O(1000) tasks; PRRTE overheads are negligible; and PRRTE supports optimizations that lower the impact of overheads and enable resource utilization of 63% when executing O(16K), 1-core tasks over 404 compute nodes.

KW - Data-Vortex;-irregular-application;-high-performance-computing

UR - http://www.scopus.com/inward/record.url?scp=85078874027&partnerID=8YFLogxK

U2 - 10.1109/IPDRM49579.2019.00007

DO - 10.1109/IPDRM49579.2019.00007

M3 - Conference contribution

AN - SCOPUS:85078874027

T3 - Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis

SP - 18

EP - 25

BT - Proceedings of IPDRM 2019

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 3rd IEEE/ACM Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware, IPDRM 2019

Y2 - 22 November 2019

ER -

Turilli M, Merzky A, Naughton T , Elwasif W, Jha S. Characterizing the Performance of Executing Many-tasks on Summit. In Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc. 2019. p. 18-25. 8956483. (Proceedings of IPDRM 2019: 3rd Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware - Held in conjunction with SC 2019: The International Conference for High Performance Computing, Networking, Storage and Analysis). doi: 10.1109/IPDRM49579.2019.00007

Characterizing the Performance of Executing Many-tasks on Summit

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this