Visualizing Hierarchical Performance Profiles of Parallel Codes Using CallFlow

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 performan...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on visualization and computer graphics 2019-11, Vol.27 (4)
Main Authors: Nguyen, Huu Tan, Bhatele, Abhinav, Jain, Nikhil, Kesavan, Suraj P., Bhatia, Harsh, Gamblin, Todd, Ma, Kwan-Liu, Bremer, Peer-Timo
Format: Article
Language:English
Subjects:
CallFlow
calling context trees
CCT
coordinated and multiple views
data visualization
degraded performance
execution time
hierarchical performance profiles
high-level overview
histograms
interactive visual analysis tool
layout
MATHEMATICS AND COMPUTING
measurement
parallel codes
parallel programming
performance analysis
production simulation codes
query processing
semantic refinement operations
software visualization
trees (mathematics)
vegetation
very large databases
visual analytics
visual representations
visualization
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary: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. In this work, 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.
ISSN:1077-2626
1941-0506