Visualizing the Topology and Data Traffic of Multi-Dimensional Torus Interconnect Networks

Shenghui Cheng, Wen Zhong, Katherine E. Isaacs, Klaus Mueller

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Torus networks are an attractive topology in supercomputing, balancing the tradeoff between network diameter and hardware costs. The nodes in a torus network are connected in a k -dimensional wrap-around mesh where each node has 2k neighbors. Effectively utilizing these networks can significantly decrease parallel communication overhead and in turn the time necessary to run large parallel scientific and data analysis applications. The potential gains are considerable - 5-D torus networks are used in the majority of the top 10 machines in the November 2017 Graph 500 list. However, the multi-dimensionality of these networks makes it difficult for analysts to diagnose ill-formed communication patterns and poor network utilization since human spatial understanding is by and large limited to 3-Ds. We propose a method based on a space-filling Hilbert curve to linearize and embed the network into a ring structure, visualizing the data traffic as flowlines in the ring interior. We compare our method with traditional 2-D embedding techniques designed for high-dimensional data, such as MDS and RadViz, and show that they are inferior to ours in this application. As a demonstration of our approach, we visualize the data flow of a massively parallel scientific code on a 5-D torus network.

Original languageEnglish (US)
Article number8473686
Pages (from-to)57191-57204
Number of pages14
JournalIEEE Access
Volume6
DOIs
StatePublished - Jan 1 2018

Fingerprint

Economic and social effects
Clustering algorithms
Topology
Flowlines
Networks (circuits)
Communication
Demonstrations
Hardware
Costs
flowable hybrid composite

Keywords

  • multi-dimensional data
  • networks
  • supercomputing
  • Torus

ASJC Scopus subject areas

  • Computer Science(all)
  • Materials Science(all)
  • Engineering(all)

Cite this

Visualizing the Topology and Data Traffic of Multi-Dimensional Torus Interconnect Networks. / Cheng, Shenghui; Zhong, Wen; Isaacs, Katherine E.; Mueller, Klaus.

In: IEEE Access, Vol. 6, 8473686, 01.01.2018, p. 57191-57204.

Research output: Contribution to journalArticle

Cheng, Shenghui ; Zhong, Wen ; Isaacs, Katherine E. ; Mueller, Klaus. / Visualizing the Topology and Data Traffic of Multi-Dimensional Torus Interconnect Networks. In: IEEE Access. 2018 ; Vol. 6. pp. 57191-57204.
@article{844efb04706246ff9f6acdb886247c50,
title = "Visualizing the Topology and Data Traffic of Multi-Dimensional Torus Interconnect Networks",
abstract = "Torus networks are an attractive topology in supercomputing, balancing the tradeoff between network diameter and hardware costs. The nodes in a torus network are connected in a k -dimensional wrap-around mesh where each node has 2k neighbors. Effectively utilizing these networks can significantly decrease parallel communication overhead and in turn the time necessary to run large parallel scientific and data analysis applications. The potential gains are considerable - 5-D torus networks are used in the majority of the top 10 machines in the November 2017 Graph 500 list. However, the multi-dimensionality of these networks makes it difficult for analysts to diagnose ill-formed communication patterns and poor network utilization since human spatial understanding is by and large limited to 3-Ds. We propose a method based on a space-filling Hilbert curve to linearize and embed the network into a ring structure, visualizing the data traffic as flowlines in the ring interior. We compare our method with traditional 2-D embedding techniques designed for high-dimensional data, such as MDS and RadViz, and show that they are inferior to ours in this application. As a demonstration of our approach, we visualize the data flow of a massively parallel scientific code on a 5-D torus network.",
keywords = "multi-dimensional data, networks, supercomputing, Torus",
author = "Shenghui Cheng and Wen Zhong and Isaacs, {Katherine E.} and Klaus Mueller",
year = "2018",
month = "1",
day = "1",
doi = "10.1109/ACCESS.2018.2872344",
language = "English (US)",
volume = "6",
pages = "57191--57204",
journal = "IEEE Access",
issn = "2169-3536",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Visualizing the Topology and Data Traffic of Multi-Dimensional Torus Interconnect Networks

AU - Cheng, Shenghui

AU - Zhong, Wen

AU - Isaacs, Katherine E.

AU - Mueller, Klaus

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Torus networks are an attractive topology in supercomputing, balancing the tradeoff between network diameter and hardware costs. The nodes in a torus network are connected in a k -dimensional wrap-around mesh where each node has 2k neighbors. Effectively utilizing these networks can significantly decrease parallel communication overhead and in turn the time necessary to run large parallel scientific and data analysis applications. The potential gains are considerable - 5-D torus networks are used in the majority of the top 10 machines in the November 2017 Graph 500 list. However, the multi-dimensionality of these networks makes it difficult for analysts to diagnose ill-formed communication patterns and poor network utilization since human spatial understanding is by and large limited to 3-Ds. We propose a method based on a space-filling Hilbert curve to linearize and embed the network into a ring structure, visualizing the data traffic as flowlines in the ring interior. We compare our method with traditional 2-D embedding techniques designed for high-dimensional data, such as MDS and RadViz, and show that they are inferior to ours in this application. As a demonstration of our approach, we visualize the data flow of a massively parallel scientific code on a 5-D torus network.

AB - Torus networks are an attractive topology in supercomputing, balancing the tradeoff between network diameter and hardware costs. The nodes in a torus network are connected in a k -dimensional wrap-around mesh where each node has 2k neighbors. Effectively utilizing these networks can significantly decrease parallel communication overhead and in turn the time necessary to run large parallel scientific and data analysis applications. The potential gains are considerable - 5-D torus networks are used in the majority of the top 10 machines in the November 2017 Graph 500 list. However, the multi-dimensionality of these networks makes it difficult for analysts to diagnose ill-formed communication patterns and poor network utilization since human spatial understanding is by and large limited to 3-Ds. We propose a method based on a space-filling Hilbert curve to linearize and embed the network into a ring structure, visualizing the data traffic as flowlines in the ring interior. We compare our method with traditional 2-D embedding techniques designed for high-dimensional data, such as MDS and RadViz, and show that they are inferior to ours in this application. As a demonstration of our approach, we visualize the data flow of a massively parallel scientific code on a 5-D torus network.

KW - multi-dimensional data

KW - networks

KW - supercomputing

KW - Torus

UR - http://www.scopus.com/inward/record.url?scp=85054256447&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054256447&partnerID=8YFLogxK

U2 - 10.1109/ACCESS.2018.2872344

DO - 10.1109/ACCESS.2018.2872344

M3 - Article

AN - SCOPUS:85054256447

VL - 6

SP - 57191

EP - 57204

JO - IEEE Access

JF - IEEE Access

SN - 2169-3536

M1 - 8473686

ER -