Hierarchical scalable photonic architectures for high-performance processor interconnection

Dowd, P.W.; Bogineni, Kalyani; Aly, K.A.; Perreault, J.A.

doi:10.1109/12.241599

Cited by 59 publications

(56 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As expected, the value of u increases with r ′. The fact that for a given r ′, the value of u increases with K is consistent with previous results which show that, in general, multiplexing can improve both bandwidth utilization and network throughput [14], [15], [16]. Our results have also shown that for a given r ′ and K, u decreases with H. This is due to the fact that a longer path has a smaller chance of being established successfully.…”

Section: Probability Analysissupporting

confidence: 78%

“…Having virtual channels increases bandwidth utilization and facilitates adaptive routing algorithms as well as the static mapping of the communication requirements of various applications [6], [7], [8]. TDM techniques are also useful for tolerating the propagation latency in optically interconnected multiprocessor systems [9], [10], [11], [12], and for reducing the control complexity of channel allocation in TDM systems [13], [14], and in Wavelength Division Multiplexed (WDM) systems [15], [16].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Reducing communication latency with path multiplexing: in optically interconnected multiprocessor systems

Qiao

Melhem

Proceedings of 1995 1st IEEE Symposium on High Performance Computer Architecture

View full text Add to dashboard Cite

Abstract-Reducing communication latency, which is a performance bottleneck in optically interconnected multiprocessor systems, is of prominent importance. A conventional approach for establishing connections in multiplexed networks uses a set of independent time slots (or virtual channels) along a path for each connection. This approach requires the use of switching devices capable of interchanging time slots, and thus introduces latency in addition to hardware and control complexity. In this paper, we propose an approach to all-optical Time Division Multiplexed (TDM) communications in multiprocessor systems. The idea is to establish a connection along a path using a set of time slots (or virtual channels) that are dependent on each other, so that no time-slot interchanging is required. We compare the proposed approach with the conventional one in terms of the overall communication latency. We found that, despite the possibility that establishing a connection may take a longer time, the proposed approach will result in lower overall communication latency as it eliminates the delays introduced by the time-slot interchanging switching devices.

show abstract

Section: Probability Analysissupporting

confidence: 78%

Section: Introductionmentioning

confidence: 99%

Reducing communication latency with path multiplexing: in optically interconnected multiprocessor systems

Qiao

Melhem

Proceedings of 1995 1st IEEE Symposium on High Performance Computer Architecture

View full text Add to dashboard Cite

show abstract

“…Point-To-Point Network (PTP): A point-to-point network assigns each node with at least one unique receiving channel, referred to as home channel [3,12] (see Figure 4(c)). One node is able to make a point-to-point connection to any other node by transmitting on the unique home channel of that destination node.…”

Section: Network Configurationsmentioning

confidence: 99%

A New Token-Based Channel Access Protocol for Wavelength Division Multiplexed Multiprocessor Interconnects

Pinkston

2000

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

“…Moreover using the hierarchical networks to decrease the diameters and average distances is worth to study topics, especially the networks exceed the thousand nodes. The hierarchical ring network architectures can decrease the network diameters [2][3][4].…”

mentioning

confidence: 99%

Delay Differentiated IP over Hierarchical WDM Ring Networks

Ho¹

2016

JCC

View full text Add to dashboard Cite

WDM (wavelength division multiplexing) appears to be the solution of choice for providing a faster networking infrastructure that can meet the explosive growth of the Internet. It has the scalability problem with increasing the access node numbers in WDM ring architecture, so we further propose the alternatives of hierarchical WDM ring architecture. In this architecture the number of routing nodes and user nodes shown as formulas and propose the routing rules. Finally, the proportional delay differentiated mechanism is added for this system to support DiffServ (differentiated services).

show abstract

Hierarchical scalable photonic architectures for high-performance processor interconnection

Cited by 59 publications

References 27 publications

Reducing communication latency with path multiplexing: in optically interconnected multiprocessor systems

Reducing communication latency with path multiplexing: in optically interconnected multiprocessor systems

A New Token-Based Channel Access Protocol for Wavelength Division Multiplexed Multiprocessor Interconnects

Delay Differentiated IP over Hierarchical WDM Ring Networks

Contact Info

Product

Resources

About