Performance Debugging and Tuning of Flash-X with Data Analysis Tools

Kevin Huck; Xingfu Wu; Anshu Dubey; Antigoni Georgiadou; J. Austin Harris; Tom Klosterman; Matthew Trappett; Klaus Weide

doi:10.1109/ProTools56701.2022.00009

Performance Debugging and Tuning of Flash-X with Data Analysis Tools

Kevin Huck, Xingfu Wu, Anshu Dubey, Antigoni Georgiadou, J. Austin Harris, Tom Klosterman, Matthew Trappett, Klaus Weide

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

Abstract

State-of-the-art multiphysics simulations running on large scale leadership computing platforms have many variables contributing to their performance and scaling behavior. We recently encountered an interesting performance anomaly in Flash-X, a multiphysics multicomponent simulation software, when characterizing its performance behavior on several large-scale HPC platforms. The anomaly was tracked down to the interaction between the use of dynamic allocation of scratch data and data locality in the cache hierarchy. In this paper we present the details of unexpected performance variability of Flash-X, its extensive analysis using the performance measurement tool TAU to collect the data and Python data analysis libraries to explore the data, and our insights from this experience. In this process, we discovered and removed or mitigated two additional performance limiting bottlenecks for performance tuning.

Original language	English
Title of host publication	Proceedings of ProTools 2022
Subtitle of host publication	Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	21-30
Number of pages	10
ISBN (Electronic)	9781665475648
DOIs	https://doi.org/10.1109/ProTools56701.2022.00009
State	Published - 2022
Event	4th IEEE/ACM Workshop on Programming and Performance Visualization Tools, ProTools 2022 - Dallas, United States Duration: Nov 13 2022 → Nov 18 2022

Publication series

Name	Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis

Conference

Conference	4th IEEE/ACM Workshop on Programming and Performance Visualization Tools, ProTools 2022
Country/Territory	United States
City	Dallas
Period	11/13/22 → 11/18/22

Funding

of Science laboratory, is operated under Contract No. DE-AC02-06CH11357. The U.S. Government retains for itself, and others acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies publicly and display publicly, by or on behalf of the Government. 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. This work was also supported by the Scientific Discovery through Advanced Computing (SciDAC) program funded by U.S. Department of Energy, Office of Science, Advanced Scientific Computing Research (ASCR) under contract DE-SC0021299.

Keywords

Multiphysics application, adaptive mesh refinement, performance, data analysis

Access to Document

10.1109/ProTools56701.2022.00009

Cite this

Huck, K., Wu, X., Dubey, A., Georgiadou, A., Harris, J. A., Klosterman, T., Trappett, M., & Weide, K. (2022). Performance Debugging and Tuning of Flash-X with Data Analysis Tools. In Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis (pp. 21-30). (Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ProTools56701.2022.00009

Huck, Kevin ; Wu, Xingfu ; Dubey, Anshu et al. / Performance Debugging and Tuning of Flash-X with Data Analysis Tools. Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc., 2022. pp. 21-30 (Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis).

@inproceedings{92e97ff2272243aba202189a8d9bef04,

title = "Performance Debugging and Tuning of Flash-X with Data Analysis Tools",

abstract = "State-of-the-art multiphysics simulations running on large scale leadership computing platforms have many variables contributing to their performance and scaling behavior. We recently encountered an interesting performance anomaly in Flash-X, a multiphysics multicomponent simulation software, when characterizing its performance behavior on several large-scale HPC platforms. The anomaly was tracked down to the interaction between the use of dynamic allocation of scratch data and data locality in the cache hierarchy. In this paper we present the details of unexpected performance variability of Flash-X, its extensive analysis using the performance measurement tool TAU to collect the data and Python data analysis libraries to explore the data, and our insights from this experience. In this process, we discovered and removed or mitigated two additional performance limiting bottlenecks for performance tuning.",

keywords = "Multiphysics application, adaptive mesh refinement, performance, data analysis",

author = "Kevin Huck and Xingfu Wu and Anshu Dubey and Antigoni Georgiadou and Harris, {J. Austin} and Tom Klosterman and Matthew Trappett and Klaus Weide",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 4th IEEE/ACM Workshop on Programming and Performance Visualization Tools, ProTools 2022 ; Conference date: 13-11-2022 Through 18-11-2022",

year = "2022",

doi = "10.1109/ProTools56701.2022.00009",

language = "English",

series = "Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis",

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

pages = "21--30",

booktitle = "Proceedings of ProTools 2022",

}

Huck, K, Wu, X, Dubey, A, Georgiadou, A , Harris, JA, Klosterman, T, Trappett, M & Weide, K 2022, Performance Debugging and Tuning of Flash-X with Data Analysis Tools. in Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis. Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis, Institute of Electrical and Electronics Engineers Inc., pp. 21-30, 4th IEEE/ACM Workshop on Programming and Performance Visualization Tools, ProTools 2022, Dallas, United States, 11/13/22. https://doi.org/10.1109/ProTools56701.2022.00009

Performance Debugging and Tuning of Flash-X with Data Analysis Tools. / Huck, Kevin; Wu, Xingfu; Dubey, Anshu et al.
Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc., 2022. p. 21-30 (Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: 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 - Performance Debugging and Tuning of Flash-X with Data Analysis Tools

AU - Huck, Kevin

AU - Wu, Xingfu

AU - Dubey, Anshu

AU - Georgiadou, Antigoni

AU - Harris, J. Austin

AU - Klosterman, Tom

AU - Trappett, Matthew

AU - Weide, Klaus

PY - 2022

Y1 - 2022

N2 - State-of-the-art multiphysics simulations running on large scale leadership computing platforms have many variables contributing to their performance and scaling behavior. We recently encountered an interesting performance anomaly in Flash-X, a multiphysics multicomponent simulation software, when characterizing its performance behavior on several large-scale HPC platforms. The anomaly was tracked down to the interaction between the use of dynamic allocation of scratch data and data locality in the cache hierarchy. In this paper we present the details of unexpected performance variability of Flash-X, its extensive analysis using the performance measurement tool TAU to collect the data and Python data analysis libraries to explore the data, and our insights from this experience. In this process, we discovered and removed or mitigated two additional performance limiting bottlenecks for performance tuning.

AB - State-of-the-art multiphysics simulations running on large scale leadership computing platforms have many variables contributing to their performance and scaling behavior. We recently encountered an interesting performance anomaly in Flash-X, a multiphysics multicomponent simulation software, when characterizing its performance behavior on several large-scale HPC platforms. The anomaly was tracked down to the interaction between the use of dynamic allocation of scratch data and data locality in the cache hierarchy. In this paper we present the details of unexpected performance variability of Flash-X, its extensive analysis using the performance measurement tool TAU to collect the data and Python data analysis libraries to explore the data, and our insights from this experience. In this process, we discovered and removed or mitigated two additional performance limiting bottlenecks for performance tuning.

KW - Multiphysics application, adaptive mesh refinement, performance, data analysis

UR - http://www.scopus.com/inward/record.url?scp=85147900120&partnerID=8YFLogxK

U2 - 10.1109/ProTools56701.2022.00009

DO - 10.1109/ProTools56701.2022.00009

M3 - Conference contribution

AN - SCOPUS:85147900120

T3 - Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis

SP - 21

EP - 30

BT - Proceedings of ProTools 2022

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 4th IEEE/ACM Workshop on Programming and Performance Visualization Tools, ProTools 2022

Y2 - 13 November 2022 through 18 November 2022

ER -

Huck K, Wu X, Dubey A, Georgiadou A , Harris JA, Klosterman T et al. Performance Debugging and Tuning of Flash-X with Data Analysis Tools. In Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc. 2022. p. 21-30. (Proceedings of ProTools 2022: Workshop on Programming and Performance Visualization Tools, Held in conjunction with SC 2022: The International Conference for High Performance Computing, Networking, Storage and Analysis). doi: 10.1109/ProTools56701.2022.00009

Performance Debugging and Tuning of Flash-X with Data Analysis Tools

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this