Recursive diagonal torus

An interconnection network for massively parallel computers

Yulu Yang, Akira Funahashi, Akiya Jouraku, Hiroaki Nishi, Hideharu Amano, Toshinori Sueyoshi

Research output: Contribution to journalArticle

49 Citations (Scopus)

Abstract

Recursive Diagonal Torus (RDT), a class of interconnection network is proposed for massively parallel computers with up to 2 16 nodes. By making the best use of a recursively structured diagonal mesh (torus) connection, the RDT has a smaller diameter (e.g., it is 11 for 2 16 nodes) with a smaller number of links per node (i.e., 8 links per node) than those of the hypercube, A simple routing algorithm, called vector routing, which is near-optimal and easy to implement is also proposed. Although the congestion on upper rank tori sometimes degrades the performance under the random traffic, the RDT provides much better performance than that of a 2D/3D torus in most cases and, under hot spot traffic, the RDT provides much better performance than that of a 2D/3D/4D torus. The RDT router chip which provides a message multicast for maintaining cache consistency is available. Using the 0.5μm BiCMOS SOG technology, versatile functions, including hierarchical multicasting, combining acknowledge packets, shooting down/restart mechanism, and time-out/setup mechanisms, work at a 60MHz clock rate.

Original languageEnglish
Pages (from-to)701-715
Number of pages15
JournalIEEE Transactions on Parallel and Distributed Systems
Volume12
Issue number7
DOIs
Publication statusPublished - 2001 Jul

Fingerprint

BiCMOS technology
Multicasting
Interconnection Networks
Routing algorithms
Parallel Computers
Routers
Clocks
Torus
Vertex of a graph
Traffic
Shooting
Restart
Routing Algorithm
Hot Spot
Router
Multicast
Hypercube
Congestion
Cache
Routing

Keywords

  • Interconnection network
  • Massively parallel computer
  • Mesh network
  • Message multicast
  • Router chip
  • Routing algorithm
  • Torus network

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Theoretical Computer Science
  • Computational Theory and Mathematics

Cite this

Recursive diagonal torus : An interconnection network for massively parallel computers. / Yang, Yulu; Funahashi, Akira; Jouraku, Akiya; Nishi, Hiroaki; Amano, Hideharu; Sueyoshi, Toshinori.

In: IEEE Transactions on Parallel and Distributed Systems, Vol. 12, No. 7, 07.2001, p. 701-715.

Research output: Contribution to journalArticle

@article{08effc3558e6413d920d2dbe89ab340a,
title = "Recursive diagonal torus: An interconnection network for massively parallel computers",
abstract = "Recursive Diagonal Torus (RDT), a class of interconnection network is proposed for massively parallel computers with up to 2 16 nodes. By making the best use of a recursively structured diagonal mesh (torus) connection, the RDT has a smaller diameter (e.g., it is 11 for 2 16 nodes) with a smaller number of links per node (i.e., 8 links per node) than those of the hypercube, A simple routing algorithm, called vector routing, which is near-optimal and easy to implement is also proposed. Although the congestion on upper rank tori sometimes degrades the performance under the random traffic, the RDT provides much better performance than that of a 2D/3D torus in most cases and, under hot spot traffic, the RDT provides much better performance than that of a 2D/3D/4D torus. The RDT router chip which provides a message multicast for maintaining cache consistency is available. Using the 0.5μm BiCMOS SOG technology, versatile functions, including hierarchical multicasting, combining acknowledge packets, shooting down/restart mechanism, and time-out/setup mechanisms, work at a 60MHz clock rate.",
keywords = "Interconnection network, Massively parallel computer, Mesh network, Message multicast, Router chip, Routing algorithm, Torus network",
author = "Yulu Yang and Akira Funahashi and Akiya Jouraku and Hiroaki Nishi and Hideharu Amano and Toshinori Sueyoshi",
year = "2001",
month = "7",
doi = "10.1109/71.940745",
language = "English",
volume = "12",
pages = "701--715",
journal = "IEEE Transactions on Parallel and Distributed Systems",
issn = "1045-9219",
publisher = "IEEE Computer Society",
number = "7",

}

TY - JOUR

T1 - Recursive diagonal torus

T2 - An interconnection network for massively parallel computers

AU - Yang, Yulu

AU - Funahashi, Akira

AU - Jouraku, Akiya

AU - Nishi, Hiroaki

AU - Amano, Hideharu

AU - Sueyoshi, Toshinori

PY - 2001/7

Y1 - 2001/7

N2 - Recursive Diagonal Torus (RDT), a class of interconnection network is proposed for massively parallel computers with up to 2 16 nodes. By making the best use of a recursively structured diagonal mesh (torus) connection, the RDT has a smaller diameter (e.g., it is 11 for 2 16 nodes) with a smaller number of links per node (i.e., 8 links per node) than those of the hypercube, A simple routing algorithm, called vector routing, which is near-optimal and easy to implement is also proposed. Although the congestion on upper rank tori sometimes degrades the performance under the random traffic, the RDT provides much better performance than that of a 2D/3D torus in most cases and, under hot spot traffic, the RDT provides much better performance than that of a 2D/3D/4D torus. The RDT router chip which provides a message multicast for maintaining cache consistency is available. Using the 0.5μm BiCMOS SOG technology, versatile functions, including hierarchical multicasting, combining acknowledge packets, shooting down/restart mechanism, and time-out/setup mechanisms, work at a 60MHz clock rate.

AB - Recursive Diagonal Torus (RDT), a class of interconnection network is proposed for massively parallel computers with up to 2 16 nodes. By making the best use of a recursively structured diagonal mesh (torus) connection, the RDT has a smaller diameter (e.g., it is 11 for 2 16 nodes) with a smaller number of links per node (i.e., 8 links per node) than those of the hypercube, A simple routing algorithm, called vector routing, which is near-optimal and easy to implement is also proposed. Although the congestion on upper rank tori sometimes degrades the performance under the random traffic, the RDT provides much better performance than that of a 2D/3D torus in most cases and, under hot spot traffic, the RDT provides much better performance than that of a 2D/3D/4D torus. The RDT router chip which provides a message multicast for maintaining cache consistency is available. Using the 0.5μm BiCMOS SOG technology, versatile functions, including hierarchical multicasting, combining acknowledge packets, shooting down/restart mechanism, and time-out/setup mechanisms, work at a 60MHz clock rate.

KW - Interconnection network

KW - Massively parallel computer

KW - Mesh network

KW - Message multicast

KW - Router chip

KW - Routing algorithm

KW - Torus network

UR - http://www.scopus.com/inward/record.url?scp=0035392176&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0035392176&partnerID=8YFLogxK

U2 - 10.1109/71.940745

DO - 10.1109/71.940745

M3 - Article

VL - 12

SP - 701

EP - 715

JO - IEEE Transactions on Parallel and Distributed Systems

JF - IEEE Transactions on Parallel and Distributed Systems

SN - 1045-9219

IS - 7

ER -