Speculative Flow Control for High-Radix Datacenter Interconnect Routers

Minkenberg, Cyriel; Gusat, Mitch

doi:10.1109/ipdps.2007.370260

Cited by 12 publications

(3 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An early use for credit-based flow control was to manage available buffer space across transmission links within switched ATM networks [8]. Credit-based flow control has also been enhanced to incorporate speculation [11] in order to improve buffer usage efficiency. Both of these efforts focused on managing credits between a single sender and a single receiver whereas we focus on managing credits between many senders and a single receiver.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

High-performance ethernet-based communications for future multi-core processors

Schlansker

Chitlur

Oertli

et al. 2007

Proceedings of the 2007 ACM/IEEE Conference on Supercomputing

View full text Add to dashboard Cite

Data centers and HPC clusters often incorporate specialized networking fabrics to satisfy system requirements. However, Ethernet's low cost and high performance are causing a shift from specialized fabrics toward standard Ethernet. Although Ethernet's low-level performance approaches that of specialized fabrics, the features that these fabrics provide such as reliable in-order delivery and flow control are implemented, in the case of Ethernet, by endpoint hardware and software. Unfortunately, current Ethernet endpoints are either slow (commodity NICs with generic TCP/IP stacks) or costly (offload engines). To address these issues, the JNIC project developed a novel Ethernet endpoint. JNIC's hardware and software were specifically designed for the requirements of high-performance communications within future data-centers and compute clusters. The architecture combines capabilities already seen in advanced network architectures with new innovations to create a comprehensive solution for scalable and high-performance Ethernet. We envision a JNIC architecture that is suitable for most in-data-center communication needs.

show abstract

Section: Related Workmentioning

confidence: 99%

“…To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. SC07 November [10][11][12][13][14][15][16]2007, Reno, Nevada, USA (c) 2007 ACM 978-1-59593-764-3/07/0011…$5.00…”

Section: Introductionmentioning

confidence: 99%

High-performance ethernet-based communications for future multi-core processors

Schlansker

Chitlur

Oertli

et al. 2007

Proceedings of the 2007 ACM/IEEE Conference on Supercomputing

View full text Add to dashboard Cite

show abstract

“…The bandwidth of network is not fully utilized. The concept of speculative flow control composed of creditbased and speculation has been proposed in [1]. It is used to reduce the effect of latency between adjacent nodes.…”

Section: Introductionmentioning

confidence: 99%

Implementation and analysis of speculative flow control for on-chip interconnection network

Sun

Hsu

2011

2011 IEEE 2nd International Conference on Networked Embedded Systems for Enterprise Applications

View full text Add to dashboard Cite

We have built a model for credit-based flow control with speculation and discuss the performance evaluation in this paper. Speculative flow control has been proposed in [1]. First, we design and implement this flow control mechanic in Verilog HDL. In order to make performance evaluation, we set up a 4-by-4 mesh network model using credit-based flow control with and without speculation by Verilog HDL in RTL level. We tune some parameters of the network model, such as buffer size, delay time between adjacent nodes and speculated packet size for performance evaluation. We also devise two types of traffic model on mesh network to show the influence of traffic on the network. Hardware cost in FPGA or ASIC is also presented.I.

show abstract

Applying search algorithms to obtain the optimal configuration of nDT torus nodes

Andújar‐Muñoz

Villar-Ortiz

Sánchez

et al. 2017

Concurrency and Computation

View full text Add to dashboard Cite

SummaryAn nDT torus is a topology where each node comprises 2 identical (n+1)‐port communication cards interconnected by 1 port. By using the current switches or communication cards, this node architecture allows to build torus networks having a greater number of dimensions than networks including only 1 card per node. There are multiple ways to use the ports of the 2 cards to connect a node to other nodes on the nDT torus, and therefore, checking all the configurations is only an affordable problem for small values of n. In this paper, we use artificial intelligence and data mining techniques to obtain the optimal port configuration of all the nodes in the network. We include a performance evaluation that shows nDT torus effectively increases the performance compared with the equivalent torus in resources, with synthetic and application trace–based workloads. We also apply these techniques to 3DT and 5DT tori to confirm the increase in the number of dimensions that does not affect to performance of the nDT torus.

show abstract

Speculative Flow Control for High-Radix Datacenter Interconnect Routers

Abstract: Abstract

Cited by 12 publications

References 23 publications

High-performance ethernet-based communications for future multi-core processors

High-performance ethernet-based communications for future multi-core processors

Implementation and analysis of speculative flow control for on-chip interconnection network

Applying search algorithms to obtain the optimal configuration of nDT torus nodes

Contact Info

Product

Resources

About