Visualizing Hierarchical Performance Profiles of Parallel Codes Using CallFlow

Huu Tan Nguyen; Abhinav Bhatele; Nikhil Jain; Suraj P. Kesavan; Harsh Bhatia; Todd Gamblin; Kwan Liu Ma; Peer Timo Bremer

doi:10.1109/TVCG.2019.2953746

Visualizing Hierarchical Performance Profiles of Parallel Codes Using CallFlow

Huu Tan Nguyen, Abhinav Bhatele, Nikhil Jain, Suraj P. Kesavan, Harsh Bhatia, Todd Gamblin, Kwan Liu Ma, Peer Timo Bremer

Computer Science

Research output: Contribution to journal › Article › peer-review

2 Scopus citations

Abstract

Calling context trees (CCTs) couple performance metrics with call paths, helping understand the execution and performance of parallel programs. To identify performance bottlenecks, programmers and performance analysts visually explore CCTs to form and validate hypotheses regarding degraded performance. However, due to the complexity of parallel programs, existing visual representations do not scale to applications running on a large number of processors. We present CallFlow, an interactive visual analysis tool that provides a high-level overview of CCTs together with semantic refinement operations to progressively explore CCTs. Using a flow-based metaphor, we visualize a CCT by treating execution time as a resource spent during the call chain, and demonstrate the effectiveness of our design with case studies on large-scale, production simulation codes.

Original language	English (US)
Article number	8901998
Pages (from-to)	2455-2468
Number of pages	14
Journal	IEEE Transactions on Visualization and Computer Graphics
Volume	27
Issue number	4
DOIs	https://doi.org/10.1109/TVCG.2019.2953746
State	Published - Apr 1 2021

Keywords

coordinated and multiple views
hierarchical data
Performance analysis
software visualization
visual analytics

ASJC Scopus subject areas

Software
Signal Processing
Computer Vision and Pattern Recognition
Computer Graphics and Computer-Aided Design

Access to Document

10.1109/TVCG.2019.2953746

Fingerprint Dive into the research topics of 'Visualizing Hierarchical Performance Profiles of Parallel Codes Using CallFlow'. Together they form a unique fingerprint.

Cite this

Visualizing Hierarchical Performance Profiles of Parallel Codes Using CallFlow. / Nguyen, Huu Tan; Bhatele, Abhinav; Jain, Nikhil; Kesavan, Suraj P.; Bhatia, Harsh; Gamblin, Todd; Ma, Kwan Liu; Bremer, Peer Timo.

In: IEEE Transactions on Visualization and Computer Graphics, Vol. 27, No. 4, 8901998, 01.04.2021, p. 2455-2468.

Research output: Contribution to journal › Article › peer-review

@article{a02dbe8552a147cf89b15083a5abcbe9,

title = "Visualizing Hierarchical Performance Profiles of Parallel Codes Using CallFlow",

abstract = "Calling context trees (CCTs) couple performance metrics with call paths, helping understand the execution and performance of parallel programs. To identify performance bottlenecks, programmers and performance analysts visually explore CCTs to form and validate hypotheses regarding degraded performance. However, due to the complexity of parallel programs, existing visual representations do not scale to applications running on a large number of processors. We present CallFlow, an interactive visual analysis tool that provides a high-level overview of CCTs together with semantic refinement operations to progressively explore CCTs. Using a flow-based metaphor, we visualize a CCT by treating execution time as a resource spent during the call chain, and demonstrate the effectiveness of our design with case studies on large-scale, production simulation codes.",

keywords = "coordinated and multiple views, hierarchical data, Performance analysis, software visualization, visual analytics",

author = "Nguyen, {Huu Tan} and Abhinav Bhatele and Nikhil Jain and Kesavan, {Suraj P.} and Harsh Bhatia and Todd Gamblin and Ma, {Kwan Liu} and Bremer, {Peer Timo}",

year = "2021",

month = apr,

day = "1",

doi = "10.1109/TVCG.2019.2953746",

language = "English (US)",

volume = "27",

pages = "2455--2468",

journal = "IEEE Transactions on Visualization and Computer Graphics",

issn = "1077-2626",

publisher = "IEEE Computer Society",

number = "4",

}

TY - JOUR

T1 - Visualizing Hierarchical Performance Profiles of Parallel Codes Using CallFlow

AU - Nguyen, Huu Tan

AU - Bhatele, Abhinav

AU - Jain, Nikhil

AU - Kesavan, Suraj P.

AU - Bhatia, Harsh

AU - Gamblin, Todd

AU - Ma, Kwan Liu

AU - Bremer, Peer Timo

PY - 2021/4/1

Y1 - 2021/4/1

N2 - Calling context trees (CCTs) couple performance metrics with call paths, helping understand the execution and performance of parallel programs. To identify performance bottlenecks, programmers and performance analysts visually explore CCTs to form and validate hypotheses regarding degraded performance. However, due to the complexity of parallel programs, existing visual representations do not scale to applications running on a large number of processors. We present CallFlow, an interactive visual analysis tool that provides a high-level overview of CCTs together with semantic refinement operations to progressively explore CCTs. Using a flow-based metaphor, we visualize a CCT by treating execution time as a resource spent during the call chain, and demonstrate the effectiveness of our design with case studies on large-scale, production simulation codes.

AB - Calling context trees (CCTs) couple performance metrics with call paths, helping understand the execution and performance of parallel programs. To identify performance bottlenecks, programmers and performance analysts visually explore CCTs to form and validate hypotheses regarding degraded performance. However, due to the complexity of parallel programs, existing visual representations do not scale to applications running on a large number of processors. We present CallFlow, an interactive visual analysis tool that provides a high-level overview of CCTs together with semantic refinement operations to progressively explore CCTs. Using a flow-based metaphor, we visualize a CCT by treating execution time as a resource spent during the call chain, and demonstrate the effectiveness of our design with case studies on large-scale, production simulation codes.

KW - coordinated and multiple views

KW - hierarchical data

KW - Performance analysis

KW - software visualization

KW - visual analytics

UR - http://www.scopus.com/inward/record.url?scp=85102021809&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85102021809&partnerID=8YFLogxK

U2 - 10.1109/TVCG.2019.2953746

DO - 10.1109/TVCG.2019.2953746

M3 - Article

C2 - 31751276

AN - SCOPUS:85102021809

VL - 27

SP - 2455

EP - 2468

JO - IEEE Transactions on Visualization and Computer Graphics

JF - IEEE Transactions on Visualization and Computer Graphics

SN - 1077-2626

IS - 4

M1 - 8901998

ER -

Visualizing Hierarchical Performance Profiles of Parallel Codes Using CallFlow

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Cite this