Visual analysis of inter-process communication for large-scale parallel computing

Chris Muelder, Francois Gygi, Kwan-Liu Ma

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

In serial computation, program profiling is often helpful for optimization of key sections of code. When moving to parallel computation, not only does the code execution need to be considered but also communication between the different processes which can induce delays that are detrimental to performance. As the number of processes increases, so does the impact of the communication delays on performance. For large-scale parallel applications, it is critical to understand how the communication impacts performance in order to make the code more efficient. There are several tools available for visualizing program execution and communications on parallel systems. These tools generally provide either views which statistically summarize the entire program execution or process-centric views. However, process-centric visualizations do not scale well as the number of processes gets very large. In particular, the most common representation of parallel processes is a Gantt chart with a row for each process. As the number of processes increases, these charts can become difficult to work with and can even exceed screen resolution. We propose a new visualization approach that affords more scalability and then demonstrate it on systems running with up to 16,384 processes.

Original languageEnglish (US)
Article number5290721
Pages (from-to)1129-1136
Number of pages8
JournalIEEE Transactions on Visualization and Computer Graphics
Volume15
Issue number6
DOIs
StatePublished - Nov 1 2009

Fingerprint

Parallel processing systems
Communication
Visualization
Scalability

Keywords

  • Information Visualization
  • MPI Profiling
  • Scalability

ASJC Scopus subject areas

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

Cite this

Visual analysis of inter-process communication for large-scale parallel computing. / Muelder, Chris; Gygi, Francois; Ma, Kwan-Liu.

In: IEEE Transactions on Visualization and Computer Graphics, Vol. 15, No. 6, 5290721, 01.11.2009, p. 1129-1136.

Research output: Contribution to journalArticle

@article{bd2901a6ad3a4daea357ef54f1d6797a,
title = "Visual analysis of inter-process communication for large-scale parallel computing",
abstract = "In serial computation, program profiling is often helpful for optimization of key sections of code. When moving to parallel computation, not only does the code execution need to be considered but also communication between the different processes which can induce delays that are detrimental to performance. As the number of processes increases, so does the impact of the communication delays on performance. For large-scale parallel applications, it is critical to understand how the communication impacts performance in order to make the code more efficient. There are several tools available for visualizing program execution and communications on parallel systems. These tools generally provide either views which statistically summarize the entire program execution or process-centric views. However, process-centric visualizations do not scale well as the number of processes gets very large. In particular, the most common representation of parallel processes is a Gantt chart with a row for each process. As the number of processes increases, these charts can become difficult to work with and can even exceed screen resolution. We propose a new visualization approach that affords more scalability and then demonstrate it on systems running with up to 16,384 processes.",
keywords = "Information Visualization, MPI Profiling, Scalability",
author = "Chris Muelder and Francois Gygi and Kwan-Liu Ma",
year = "2009",
month = "11",
day = "1",
doi = "10.1109/TVCG.2009.196",
language = "English (US)",
volume = "15",
pages = "1129--1136",
journal = "IEEE Transactions on Visualization and Computer Graphics",
issn = "1077-2626",
publisher = "IEEE Computer Society",
number = "6",

}

TY - JOUR

T1 - Visual analysis of inter-process communication for large-scale parallel computing

AU - Muelder, Chris

AU - Gygi, Francois

AU - Ma, Kwan-Liu

PY - 2009/11/1

Y1 - 2009/11/1

N2 - In serial computation, program profiling is often helpful for optimization of key sections of code. When moving to parallel computation, not only does the code execution need to be considered but also communication between the different processes which can induce delays that are detrimental to performance. As the number of processes increases, so does the impact of the communication delays on performance. For large-scale parallel applications, it is critical to understand how the communication impacts performance in order to make the code more efficient. There are several tools available for visualizing program execution and communications on parallel systems. These tools generally provide either views which statistically summarize the entire program execution or process-centric views. However, process-centric visualizations do not scale well as the number of processes gets very large. In particular, the most common representation of parallel processes is a Gantt chart with a row for each process. As the number of processes increases, these charts can become difficult to work with and can even exceed screen resolution. We propose a new visualization approach that affords more scalability and then demonstrate it on systems running with up to 16,384 processes.

AB - In serial computation, program profiling is often helpful for optimization of key sections of code. When moving to parallel computation, not only does the code execution need to be considered but also communication between the different processes which can induce delays that are detrimental to performance. As the number of processes increases, so does the impact of the communication delays on performance. For large-scale parallel applications, it is critical to understand how the communication impacts performance in order to make the code more efficient. There are several tools available for visualizing program execution and communications on parallel systems. These tools generally provide either views which statistically summarize the entire program execution or process-centric views. However, process-centric visualizations do not scale well as the number of processes gets very large. In particular, the most common representation of parallel processes is a Gantt chart with a row for each process. As the number of processes increases, these charts can become difficult to work with and can even exceed screen resolution. We propose a new visualization approach that affords more scalability and then demonstrate it on systems running with up to 16,384 processes.

KW - Information Visualization

KW - MPI Profiling

KW - Scalability

UR - http://www.scopus.com/inward/record.url?scp=77957914829&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77957914829&partnerID=8YFLogxK

U2 - 10.1109/TVCG.2009.196

DO - 10.1109/TVCG.2009.196

M3 - Article

VL - 15

SP - 1129

EP - 1136

JO - IEEE Transactions on Visualization and Computer Graphics

JF - IEEE Transactions on Visualization and Computer Graphics

SN - 1077-2626

IS - 6

M1 - 5290721

ER -