Contention-based congestion management in large-scale networks

Kim, Gwangsun; Kim, Changhyun; Jeong, Jiyun; Parker, Michael; Kim, John

doi:10.1109/micro.2016.7783733

Cited by 22 publications

(6 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[9] propagates downstream load information from routerto-router at high frequency to ensure that routing decisions are made on the basis of up-to-date information. In terms of congestion management, [22] identified how global adaptive routing without proper congestion management can actually be harmful or degrade network performance if an endpoint destination is oversubscribed.…”

Section: Related Workmentioning

confidence: 99%

Dynamic global adaptive routing in high-radix networks

Kasan

Kim

et al. 2022

Proceedings of the 49th Annual International Symposium on Computer Architecture

Self Cite

View full text Add to dashboard Cite

Global adaptive routing is a critical component of high-radix networks in large-scale systems and is necessary to fully exploit the path diversity of a high-radix topology. The routing decision in global adaptive routing is made between minimal and non-minimal paths, often based on local information (e.g., queue occupancy) and rely on "approximate" congestion information through backpressure. Different heuristic-based adaptive routing algorithms have been proposed for high-radix topologies; however, heuristic-based routing has performance trade-off for different traffic patterns and leads to inefficient routing decisions. In addition, previously proposed global adaptive routing algorithms are static as the same routing decision algorithm is used, even if the congestion information changes. In this work, we propose a novel global adaptive routing that we refer to as dynamic global adaptive routing that adjusts the routing decision algorithm through a dynamic bias based on the network traffic and congestion to maximize performance. In particular, we propose DGB -Decoupled, Gradient descent-based Bias global adaptive routing algorithm. DGB introduces a dynamic bias to the global adaptive routing decision by leveraging gradient descent to dynamically adjust the adaptive routing bias based on the network congestion. In addition, both the local and global congestion information are decoupled in the routing decision -global information is used for the dynamic bias while local information is used in the routing decision to more accurately estimate the network congestion. Our evaluations show that DGB consistently outperforms previously proposed routing algorithms across diverse range of traffic patterns and workloads. For asymmetric traffic pattern, DGB improves throughput by 65% compared to the stateof-the-art global adaptive routing algorithm while matching the performance for symmetric traffic patterns. For trace workloads, DGB provides average performance improvement of 26%.

show abstract

Section: Related Workmentioning

confidence: 99%

Dynamic global adaptive routing in high-radix networks

Kasan

Kim

et al. 2022

Proceedings of the 49th Annual International Symposium on Computer Architecture

Self Cite

View full text Add to dashboard Cite

show abstract

“…It should be noted that the private part of the CIB is organized as fixed-length VCs. Such design allows congestion information to be transmitted to the source node as soon as possible, reducing the impact of endpoint congestion on the overall network performance [17]. To maintain the order of flits in the packet, data is written to the CIB at the granularity of the packet.…”

Section: Centralized Input Buffer Architecturementioning

confidence: 99%

Cib-Hier

Dong

Yang

et al. 2021

ACM Trans. Archit. Code Optim.

View full text Add to dashboard Cite

Hierarchical organization is widely used in high-radix routers to enable efficient scaling to higher switch port count. A general-purpose hierarchical router must be symmetrically designed with the same input buffer depth, resulting in a large amount of unused input buffers due to the different link lengths. Sharing input buffers between different input ports can improve buffer utilization, but the implementation overhead also increases with the number of shared ports. Previous work allowed input buffers to be shared among all router ports, which maximizes the buffer utilization but also introduces higher implementation complexity. Moreover, such design can impair performance when faced with long packets, due to the head-of-line blocking in intermediate buffers. In this work, we explain that sharing unused buffers between a subset of router ports is a more efficient design. Based on this observation, we propose Centralized Input Buffer Design in Hierarchical High-radix Routers (CIB-HIER), a novel centralized input buffer design for hierarchical high-radix routers. CIB-HIER integrates multiple input ports onto a single tile and organizes all unused input buffers in the tile as a centralized input buffer. CIB-HIER only allows the centralized input buffer to be shared between ports on the same tile, without introducing additional intermediate virtual channels or global scheduling circuits. Going beyond the basic design of CIB-HIER, the centralized input buffer can be used to relieve the head-of-line blocking caused by shallow intermediate buffers, by stashing long packets in the centralized input buffer. Experimental results show that CIB-HIER is highly effective and can significantly increase the throughput of high-radix routers.

show abstract

“…As an example, ERCA (Xu et al 2017), RCA (Gratz et al 2008) collect the congestion information from a specified Region. The CBCM (Kim et al 2016) and Footprint (Fu and Kim 2017) are both focused on adaptive routing and congestion management to differentiate between endpoint and network congestion based on local information of virtual channel usage. Thus, each node decides the routing path only based on the congestion information of its neighbors in the region specified.…”

Section: Previous Workmentioning

confidence: 99%

CACBR: Congestion Aware Cluster Buffer base routing algorithm with minimal cost on NOC

et al. 2020

View full text Add to dashboard Cite

Congestion aware routing algorithm enhanced the Network on Chip (NoC) performance and power consumption. The main assignment to reach this goal is to select the most proper congestion parameters along with the accuracy of collecting the congestion data. Congestion detection can be perceived while monitoring the network resources (for example monitoring the free buffers of nodes) or predicting the network traffic injection rate. The network monitoring has communication overheads as the number of nodes extend, and predicting the traffic injection rate has communication overhead along with processing overhead. In a congestion aware routing algorithm, after determining possible paths, the selection function chooses a path among the suggested ones, using the congestion detection function. In this paper, congestion detection investigates the free resources of the next cluster (neighbor). This method is distinguished in two phases. In the first phase, candidate paths will be selected by a minimal adaptive algorithm, and in the next phase, the cluster's congestion information will be sent to the selected node in order to select the best path among the suggested paths. There has been no extra signal added to the network to transfer the congestion information. In order to ensure deadlock freedom of the routing algorithm, turn restriction and virtual channels are used. Experimental results can clearly prove the improvements of average delay and network performance, whereas the average delay is reduced by 13% and 2.5% based on the traffic patterns used, and network performance gained 9% and 57% direct improvement on the algorithm output.

show abstract

Contention-based congestion management in large-scale networks

Cited by 22 publications

References 30 publications

Dynamic global adaptive routing in high-radix networks

Dynamic global adaptive routing in high-radix networks

Cib-Hier

CACBR: Congestion Aware Cluster Buffer base routing algorithm with minimal cost on NOC

Contact Info

Product

Resources

About