On-chip networks from a networking perspective

NychisGeorge, P; FallinChris,; MoscibrodaThomas,; MutluOnur,; Seshan, Srinivasan

doi:10.1145/2377677.2377757

Cited by 35 publications

(18 citation statements)

References 76 publications

(93 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The status of network congestion can be detected in different ways such as the number of a local router's busy buffers or neighbor routers' busy buffers [14], [15], time-out value [16], applications' cache miss rate or its variants [17], [18] and so on. According to the range of being aware of network congestion, there are local-knowledge based mechanisms and global-knowledge based mechanisms.…”

Section: Related Workmentioning

confidence: 99%

“…Considering the drawback of local-knowledge based mechanism, many source-throttling algorithms are aware of the congestion status of the whole network, i.e., globalknowledge based mechanisms [2], [3], [17]. The simplest available approach of achieving the global congestion status is to evaluate the time-out value which originates from computer networks [12], [16].…”

Section: Related Workmentioning

confidence: 99%

“…For example, Nychis, Fallin, etc. [17] proposed a source-throttling method with application-level awareness for bufferless NoC. They noticed that starvation rate grew super-linearly with network utilization and was a more accurate indicator of the level of congestion than network latency in a bufferless NoC.…”

Section: Related Workmentioning

confidence: 99%

“…At the same time, anchor routers change with the number and location of all congested RSD routers. Only quantity of all congested routers which is proposed in some related work such as [17] is not sufficient to evaluate congestion status of a network accurately.…”

Section: ) Figuring Out Every Rsd Router's Ccr Valuementioning

confidence: 99%

See 3 more Smart Citations

A Fine-Grained Source-Throttling Method for Mesh Architectures

et al. 2020

View full text Add to dashboard Cite

Source-throttling algorithms aim at adjusting network load appropriately so that high network throughput can be maintained and latency can be decreased efficiently. One challenge of designing a source-throttling algorithm is how to precisely evaluate network congestion status. State-of-the-art source-throttling algorithms for mesh-connected network-on-chip usually evaluate congestion status of the whole network or only that of neighbor routers. They usually appear empirical or locally-greedy. To avoid such problems, we propose a fine-grained source-throttling algorithm. Our main contributions are: 1, we can quantify the impact of throttling action on network routers; 2, we can monitor network traffic more intelligently compared with previous work, only the routers that are most affected by the source's throttling action are used to evaluate congestion status of the RSD network. Based on such measures, most throttling parameters shall not be empirically set but be precisely figured out by a smarter searching algorithm of finding anchor routers as the throttling baseline. Thus, it can solve over-throttling and under-throttling problems more precisely. Time complexity of proposed method is only O(M * N) (M*N is the scale of 2D mesh network). Simulation results show that it has higher throughput, lower fluctuation and lower latency than source-throttling algorithms with empirical parameters such as INC and self-tuned technologies under different network scenarios. INDEX TERMS Source-throttling algorithm, fine-grained, accurate throttling parameters, over-throttling, under-throttling.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: ) Figuring Out Every Rsd Router's Ccr Valuementioning

confidence: 99%

See 2 more Smart Citations

A Fine-Grained Source-Throttling Method for Mesh Architectures

et al. 2020

View full text Add to dashboard Cite

show abstract

“…On the one hand, the contentions can be relieved, thus reducing the probability of deflection to some extent. On the other hand, it is a self-throttling congestion control that allows the network to scale more effectively [6]. When the network is under congestion, the injection volume of the network can be lowered.…”

Section: Optical Router Architecturementioning

confidence: 99%

A bufferless optical network-on-chip router

Zhang

Yang

et al. 2013

IEICE Electron. Express

View full text Add to dashboard Cite

Abstract:In optical Network-on-Chip (NoC), packet switching remains popular owing to its scalability and reliability. Considering no mature optical buffering technology, deflection routing is preferable to resolve the output port contention. Reduction of deflection probability is essential to bufferless deflection optical NoC. This letter proposes a new 5*5 router architecture, and especially a deflection-supported switching fabric. Moreover, an ejection unit and an injection unit are designed to reduce the deflection. Additionally, priority-based routing computation and port allocation algorithms are designed based on the new switching fabric. The simulation results show that our proposal can improve performance at acceptable insertion loss.

show abstract

PKCA: A priori‐knowledge and congestion‐awareness method for adaptive routing algorithms in mesh architectures

Wang,

Zhao,

Wang

et al. 2024

Trans Emerging Tel Tech

View full text Add to dashboard Cite

Adaptive routing algorithms have been widely utilized in Network‐on‐Chip (NoC) architectures and have shown to enhance overall throughput in numerous studies. The adaptive routing algorithms can effectively detect network congestion. On the one hand, one‐hop awareness or local awareness can easily detect network congestion but may also result in local greed. On the other hand, global awareness is better for load balancing, but it is difficult to be aware of the network congestion status. This article proposes a lightweight adaptive on‐chip routing algorithm based on the concentric circles theory and prior knowledge derived from real‐life observations. The algorithm, named the Priori‐Knowledge and Congestion‐Awareness method (PKCA), aims to optimize the routing efficiency within the chip. PKCA is designed to be not only simple but also to have low time complexity, allowing it to calculate paths to destinations without local greed. We performed evaluations, and the results demonstrated that our design surpasses one‐hop awareness, two‐hop awareness, and global awareness by 31%, 25%, and 20%, respectively, in terms of latency, and by 22%, 14%, and 22%, respectively, in terms of throughput. Furthermore, the time complexity is only in an 2D mesh.

show abstract

On-chip networks from a networking perspective

Cited by 35 publications

References 76 publications

A Fine-Grained Source-Throttling Method for Mesh Architectures

A Fine-Grained Source-Throttling Method for Mesh Architectures

A bufferless optical network-on-chip router

PKCA: A priori‐knowledge and congestion‐awareness method for adaptive routing algorithms in mesh architectures

Contact Info

Product

Resources

About