Tree-mesh heterogeneous topology for low-latency NoC

Sungju Han, Jinho Lee, Kiyoung Choi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

In Network-on-Chip (NoC), topology is one of the most important design choices that determine performance and power consumption. Mesh, being the most popular NoC topology for many researches and products, is mainly tailored towards high throughput. However, many researches show that NoCs rarely operate under heavy load and that latency is often much more critical in practice. In this paper, we show that by adding a small tree network to assist the baseline mesh network, the zero-load latency can be greatly reduced while still maintaining the high throughput. For the management of the hybrid network, we propose a novel algorithm to steer each packet to different networks based on hop-count gain and contention monitoring. Experimental results show improvement on not only synthetic traffic but also real application workloads.

Original languageEnglish
Title of host publicationProceedings of the 2014 International Conference on Principles and Practices of Programming on the Java Platform
Subtitle of host publicationVirtual Machines, Languages, and Tools, PPPJ 2014
PublisherAssociation for Computing Machinery
Pages19-24
Number of pages6
ISBN (Electronic)9781450329262
DOIs
Publication statusPublished - 2014 Sep 23
Event11th International Conference on Principles and Practices of Programming on the Java Platform: Virtual Machines, Languages, and Tools, PPPJ 2014 - Cracow, Poland
Duration: 2014 Sep 232014 Sep 26

Publication series

NameACM International Conference Proceeding Series
Volume23-26-September-2014

Conference

Conference11th International Conference on Principles and Practices of Programming on the Java Platform: Virtual Machines, Languages, and Tools, PPPJ 2014
CountryPoland
CityCracow
Period14/9/2314/9/26

Fingerprint

Throughput
Topology
Electric power utilization
Monitoring
Network-on-chip

All Science Journal Classification (ASJC) codes

  • Software
  • Human-Computer Interaction
  • Computer Vision and Pattern Recognition
  • Computer Networks and Communications

Cite this

Han, S., Lee, J., & Choi, K. (2014). Tree-mesh heterogeneous topology for low-latency NoC. In Proceedings of the 2014 International Conference on Principles and Practices of Programming on the Java Platform: Virtual Machines, Languages, and Tools, PPPJ 2014 (pp. 19-24). [2685346] (ACM International Conference Proceeding Series; Vol. 23-26-September-2014). Association for Computing Machinery. https://doi.org/10.1145/2685342.2685346
Han, Sungju ; Lee, Jinho ; Choi, Kiyoung. / Tree-mesh heterogeneous topology for low-latency NoC. Proceedings of the 2014 International Conference on Principles and Practices of Programming on the Java Platform: Virtual Machines, Languages, and Tools, PPPJ 2014. Association for Computing Machinery, 2014. pp. 19-24 (ACM International Conference Proceeding Series).
@inproceedings{050d06449a634c9ea168f4a649ec6c76,
title = "Tree-mesh heterogeneous topology for low-latency NoC",
abstract = "In Network-on-Chip (NoC), topology is one of the most important design choices that determine performance and power consumption. Mesh, being the most popular NoC topology for many researches and products, is mainly tailored towards high throughput. However, many researches show that NoCs rarely operate under heavy load and that latency is often much more critical in practice. In this paper, we show that by adding a small tree network to assist the baseline mesh network, the zero-load latency can be greatly reduced while still maintaining the high throughput. For the management of the hybrid network, we propose a novel algorithm to steer each packet to different networks based on hop-count gain and contention monitoring. Experimental results show improvement on not only synthetic traffic but also real application workloads.",
author = "Sungju Han and Jinho Lee and Kiyoung Choi",
year = "2014",
month = "9",
day = "23",
doi = "10.1145/2685342.2685346",
language = "English",
series = "ACM International Conference Proceeding Series",
publisher = "Association for Computing Machinery",
pages = "19--24",
booktitle = "Proceedings of the 2014 International Conference on Principles and Practices of Programming on the Java Platform",

}

Han, S, Lee, J & Choi, K 2014, Tree-mesh heterogeneous topology for low-latency NoC. in Proceedings of the 2014 International Conference on Principles and Practices of Programming on the Java Platform: Virtual Machines, Languages, and Tools, PPPJ 2014., 2685346, ACM International Conference Proceeding Series, vol. 23-26-September-2014, Association for Computing Machinery, pp. 19-24, 11th International Conference on Principles and Practices of Programming on the Java Platform: Virtual Machines, Languages, and Tools, PPPJ 2014, Cracow, Poland, 14/9/23. https://doi.org/10.1145/2685342.2685346

Tree-mesh heterogeneous topology for low-latency NoC. / Han, Sungju; Lee, Jinho; Choi, Kiyoung.

Proceedings of the 2014 International Conference on Principles and Practices of Programming on the Java Platform: Virtual Machines, Languages, and Tools, PPPJ 2014. Association for Computing Machinery, 2014. p. 19-24 2685346 (ACM International Conference Proceeding Series; Vol. 23-26-September-2014).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Tree-mesh heterogeneous topology for low-latency NoC

AU - Han, Sungju

AU - Lee, Jinho

AU - Choi, Kiyoung

PY - 2014/9/23

Y1 - 2014/9/23

N2 - In Network-on-Chip (NoC), topology is one of the most important design choices that determine performance and power consumption. Mesh, being the most popular NoC topology for many researches and products, is mainly tailored towards high throughput. However, many researches show that NoCs rarely operate under heavy load and that latency is often much more critical in practice. In this paper, we show that by adding a small tree network to assist the baseline mesh network, the zero-load latency can be greatly reduced while still maintaining the high throughput. For the management of the hybrid network, we propose a novel algorithm to steer each packet to different networks based on hop-count gain and contention monitoring. Experimental results show improvement on not only synthetic traffic but also real application workloads.

AB - In Network-on-Chip (NoC), topology is one of the most important design choices that determine performance and power consumption. Mesh, being the most popular NoC topology for many researches and products, is mainly tailored towards high throughput. However, many researches show that NoCs rarely operate under heavy load and that latency is often much more critical in practice. In this paper, we show that by adding a small tree network to assist the baseline mesh network, the zero-load latency can be greatly reduced while still maintaining the high throughput. For the management of the hybrid network, we propose a novel algorithm to steer each packet to different networks based on hop-count gain and contention monitoring. Experimental results show improvement on not only synthetic traffic but also real application workloads.

UR - http://www.scopus.com/inward/record.url?scp=84984674330&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84984674330&partnerID=8YFLogxK

U2 - 10.1145/2685342.2685346

DO - 10.1145/2685342.2685346

M3 - Conference contribution

AN - SCOPUS:84984674330

T3 - ACM International Conference Proceeding Series

SP - 19

EP - 24

BT - Proceedings of the 2014 International Conference on Principles and Practices of Programming on the Java Platform

PB - Association for Computing Machinery

ER -

Han S, Lee J, Choi K. Tree-mesh heterogeneous topology for low-latency NoC. In Proceedings of the 2014 International Conference on Principles and Practices of Programming on the Java Platform: Virtual Machines, Languages, and Tools, PPPJ 2014. Association for Computing Machinery. 2014. p. 19-24. 2685346. (ACM International Conference Proceeding Series). https://doi.org/10.1145/2685342.2685346