Internet-router buffered crossbars based on networks on chip

K.G.W. Goossens, L. Mhamdi, I.V. Senín

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

20 Citations (Scopus)

Abstract

The scalability and performance of the Internet depends critically on the performance of its packet switches. Current packet switches are based on single-hop crossbar fabrics, with line cards that use virtual output-queueing to reduce head-of-line blocking. In this paper we propose to use a multi-hop network on a chip (NOC) as the crossbar fabric, with FIFO-queued line cards. The use of a multi-hop crossbar fabric has several advantages. 1) Speed-up, i.e. the crossbar fabric can operate faster because NOC inter-router wires are shorter than those in a single-hop crossbar, and because arbitration is distributed instead of centralised. 2) Load balancing because paths from different input-output port pairs share the same router buffers, unlike the internal buffers of buffered crossbar fabric that are dedicated to a single input-output pair. 3) Path diversity allows traffic from an input port to follow different paths to its destination output port. This results in further load balancing, especially for non-uniform traffic patterns. 4) Simpler line-card design: the use of FIFOs on the line cards simplifies both the line cards and the (inter-chip) flow control between the crossbar fabric and line cards, reducing the number of (expensive) chip pins required for flow control. 5) Scalability, in the sense that the crossbar speed is independent of the number of ports, which is not the case for single-hop crossbar fabrics. We analyzed the performance of our architecture both analytically and by simulation, and show that it performs well for a wide range of traffic conditions and switch sizes. Additionally we prototyped a 32 x 32 NOC-based crossbar fabric in a 65nm CMOS technology. The unoptimised implementation operates at 413 MHz, achieving an aggregate throughput in excess of 1010 ATM cells per second. © 2009 IEEE.
Original languageEnglish
Title of host publication12th Euromicro Conference on Digital System Design: Architectures, Methods and Tools, DSD 2009, 27 August - 29 August 2009, Patras
Pages365-374
DOIs
Publication statusPublished - 2009

Fingerprint

Routers
Internet
Switches
Flow control
Resource allocation
Scalability
Network-on-chip
Automatic teller machines
Throughput
Wire

Cite this

Goossens, K. G. W., Mhamdi, L., & Senín, I. V. (2009). Internet-router buffered crossbars based on networks on chip. In 12th Euromicro Conference on Digital System Design: Architectures, Methods and Tools, DSD 2009, 27 August - 29 August 2009, Patras (pp. 365-374) https://doi.org/10.1109/DSD.2009.211
Goossens, K.G.W. ; Mhamdi, L. ; Senín, I.V. / Internet-router buffered crossbars based on networks on chip. 12th Euromicro Conference on Digital System Design: Architectures, Methods and Tools, DSD 2009, 27 August - 29 August 2009, Patras. 2009. pp. 365-374
@inproceedings{5c6a822d2b5b4ed2bf84b0867049f80a,
title = "Internet-router buffered crossbars based on networks on chip",
abstract = "The scalability and performance of the Internet depends critically on the performance of its packet switches. Current packet switches are based on single-hop crossbar fabrics, with line cards that use virtual output-queueing to reduce head-of-line blocking. In this paper we propose to use a multi-hop network on a chip (NOC) as the crossbar fabric, with FIFO-queued line cards. The use of a multi-hop crossbar fabric has several advantages. 1) Speed-up, i.e. the crossbar fabric can operate faster because NOC inter-router wires are shorter than those in a single-hop crossbar, and because arbitration is distributed instead of centralised. 2) Load balancing because paths from different input-output port pairs share the same router buffers, unlike the internal buffers of buffered crossbar fabric that are dedicated to a single input-output pair. 3) Path diversity allows traffic from an input port to follow different paths to its destination output port. This results in further load balancing, especially for non-uniform traffic patterns. 4) Simpler line-card design: the use of FIFOs on the line cards simplifies both the line cards and the (inter-chip) flow control between the crossbar fabric and line cards, reducing the number of (expensive) chip pins required for flow control. 5) Scalability, in the sense that the crossbar speed is independent of the number of ports, which is not the case for single-hop crossbar fabrics. We analyzed the performance of our architecture both analytically and by simulation, and show that it performs well for a wide range of traffic conditions and switch sizes. Additionally we prototyped a 32 x 32 NOC-based crossbar fabric in a 65nm CMOS technology. The unoptimised implementation operates at 413 MHz, achieving an aggregate throughput in excess of 1010 ATM cells per second. {\circledC} 2009 IEEE.",
author = "K.G.W. Goossens and L. Mhamdi and I.V. Sen{\'i}n",
year = "2009",
doi = "10.1109/DSD.2009.211",
language = "English",
pages = "365--374",
booktitle = "12th Euromicro Conference on Digital System Design: Architectures, Methods and Tools, DSD 2009, 27 August - 29 August 2009, Patras",

}

Goossens, KGW, Mhamdi, L & Senín, IV 2009, Internet-router buffered crossbars based on networks on chip. in 12th Euromicro Conference on Digital System Design: Architectures, Methods and Tools, DSD 2009, 27 August - 29 August 2009, Patras. pp. 365-374. https://doi.org/10.1109/DSD.2009.211

Internet-router buffered crossbars based on networks on chip. / Goossens, K.G.W.; Mhamdi, L.; Senín, I.V.

12th Euromicro Conference on Digital System Design: Architectures, Methods and Tools, DSD 2009, 27 August - 29 August 2009, Patras. 2009. p. 365-374.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Internet-router buffered crossbars based on networks on chip

AU - Goossens, K.G.W.

AU - Mhamdi, L.

AU - Senín, I.V.

PY - 2009

Y1 - 2009

N2 - The scalability and performance of the Internet depends critically on the performance of its packet switches. Current packet switches are based on single-hop crossbar fabrics, with line cards that use virtual output-queueing to reduce head-of-line blocking. In this paper we propose to use a multi-hop network on a chip (NOC) as the crossbar fabric, with FIFO-queued line cards. The use of a multi-hop crossbar fabric has several advantages. 1) Speed-up, i.e. the crossbar fabric can operate faster because NOC inter-router wires are shorter than those in a single-hop crossbar, and because arbitration is distributed instead of centralised. 2) Load balancing because paths from different input-output port pairs share the same router buffers, unlike the internal buffers of buffered crossbar fabric that are dedicated to a single input-output pair. 3) Path diversity allows traffic from an input port to follow different paths to its destination output port. This results in further load balancing, especially for non-uniform traffic patterns. 4) Simpler line-card design: the use of FIFOs on the line cards simplifies both the line cards and the (inter-chip) flow control between the crossbar fabric and line cards, reducing the number of (expensive) chip pins required for flow control. 5) Scalability, in the sense that the crossbar speed is independent of the number of ports, which is not the case for single-hop crossbar fabrics. We analyzed the performance of our architecture both analytically and by simulation, and show that it performs well for a wide range of traffic conditions and switch sizes. Additionally we prototyped a 32 x 32 NOC-based crossbar fabric in a 65nm CMOS technology. The unoptimised implementation operates at 413 MHz, achieving an aggregate throughput in excess of 1010 ATM cells per second. © 2009 IEEE.

AB - The scalability and performance of the Internet depends critically on the performance of its packet switches. Current packet switches are based on single-hop crossbar fabrics, with line cards that use virtual output-queueing to reduce head-of-line blocking. In this paper we propose to use a multi-hop network on a chip (NOC) as the crossbar fabric, with FIFO-queued line cards. The use of a multi-hop crossbar fabric has several advantages. 1) Speed-up, i.e. the crossbar fabric can operate faster because NOC inter-router wires are shorter than those in a single-hop crossbar, and because arbitration is distributed instead of centralised. 2) Load balancing because paths from different input-output port pairs share the same router buffers, unlike the internal buffers of buffered crossbar fabric that are dedicated to a single input-output pair. 3) Path diversity allows traffic from an input port to follow different paths to its destination output port. This results in further load balancing, especially for non-uniform traffic patterns. 4) Simpler line-card design: the use of FIFOs on the line cards simplifies both the line cards and the (inter-chip) flow control between the crossbar fabric and line cards, reducing the number of (expensive) chip pins required for flow control. 5) Scalability, in the sense that the crossbar speed is independent of the number of ports, which is not the case for single-hop crossbar fabrics. We analyzed the performance of our architecture both analytically and by simulation, and show that it performs well for a wide range of traffic conditions and switch sizes. Additionally we prototyped a 32 x 32 NOC-based crossbar fabric in a 65nm CMOS technology. The unoptimised implementation operates at 413 MHz, achieving an aggregate throughput in excess of 1010 ATM cells per second. © 2009 IEEE.

U2 - 10.1109/DSD.2009.211

DO - 10.1109/DSD.2009.211

M3 - Conference contribution

SP - 365

EP - 374

BT - 12th Euromicro Conference on Digital System Design: Architectures, Methods and Tools, DSD 2009, 27 August - 29 August 2009, Patras

ER -

Goossens KGW, Mhamdi L, Senín IV. Internet-router buffered crossbars based on networks on chip. In 12th Euromicro Conference on Digital System Design: Architectures, Methods and Tools, DSD 2009, 27 August - 29 August 2009, Patras. 2009. p. 365-374 https://doi.org/10.1109/DSD.2009.211