All-to-all broadcast in torus with wormhole-like routing

Calvin, Christophe; Pérennès, Stéphane; Trystram, Denis

doi:10.1109/spdp.1995.530675

Cited by 20 publications

(9 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There have been several studies on all-to-all broadcast for different interconnection off-chip networks such as the torus interconnection networks [24], [25], [26], [27], [28], [29], and [30]. With the emergence of NoC architectures these studies are more focused now in developing collective communication primitives for NoC interconnects.…”

Section: Related Workmentioning

confidence: 99%

On All-to-All Broadcast in Dense Gaussian Network On-Chip

Touzene

2015

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

Gaussian networks are gaining popularity as good candidates Network On-Chip (NoC) for interconnecting Multiprocessor System-on-Chips (MPSoCs). They showed better topological properties compared to the 2D torus networks with the same number of nodes N and the same degree 4. All-to-all broadcast is a collective communication algorithm used frequently in many parallel applications. Recently, Z. Zhang et al. [1] have proposed an all-to-all broadcast algorithm for Gaussian on-chip networks that achieves the minimum delay time but requires 4k extra buffers per router, where k is the network diameter. In this paper, we propose a new all-toall broadcast algorithm for dense Gaussian on-chip networks that achieves the minimum delay time without requiring any extra buffers per router. In this paper, we propose a new all-to-all broadcast algorithm for dense Gaussian on-chip networks that achieves the minimum delay time without requiring any extra buffers per router. Along with low latency, reducing the amount of buffer space and power consumption are very important issues in NoCs architectures.Index Terms-Network-on-Chip (NoC), Gaussian networks, all-to-all broadcasting, spanning trees Ç 1 INTRODUCTION T HE scale down of transistor technology allows microelectronics manufacturers to embed more sophisticated systems on a single micro-chip (SoC) [2], [3], [4], [5], [6], [7].On chip integration is becoming popular not only in the Chip Multi Processor (CMP) research area but also in the wider range of nowadays devices such as cell-phones, smart-phones, smart-houses and vehicle embedded systems, etc.As the number of cores increase in CMP, the system performance begins to be affected by on-chip interconnects. The widely used bus structures (shared bus, hierarchal bus, and bus matrix) are becoming a limiting factor for performance, space and energy consumption [5]. In order to overcome the disadvantages of bus structures, they are being replaced by Network-on-Chip (NoC). NoC architectures are an attempt to scale down the concepts of large-scale networks, and apply them in the embedded system-on-chipSoC) domain [8], [9], [10], [11], [12], [13]. A NoC is an onchip communication network used to route data from a source Processing Element (PE) to a destination PE via a network fabric that consists of switches (routers) and interconnection links (wires) [8]. NoC interconnects address scalability issues and provide a low latency communication layer, higher bandwidth, and communication parallelism. Among the most popular NoC topologies developed for CMPs such as n-dimensional mesh, torus, folded torus, hypercube, and octagon, low-dimensional networks, like 2D-mesh and 2D-torus, offer better performance in terms of higher throughput and lower latency than high-dimensional networks (high dimensional k-aryn-cubes and meshes) [13], [14]. Recently, two other wrap-around networks have been proposed as suitable alternatives to the 2D torus network: the degree 4 Gaussian networks [15], [16], [17] and the degree 6 Eisenstein-Jacobi (EJ) networks [...

show abstract

Section: Related Workmentioning

confidence: 99%

On All-to-All Broadcast in Dense Gaussian Network On-Chip

Touzene

2015

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

show abstract

“…When G has nice symmetry properties there exists a better constructive approach. As example, when G is a toroidal mesh one can use a sequence <S made up of linear codes over a vector space Z^ (for broadcasting in the ^-dimensional torus [7], for gossiping in respectively the 2 and 3-dimensional tori [8], [5]). Here, in the case of the cube, the sequence will be built from linear binary codes.…”

Section: Theorem 5 (Ford-fulkerson) -In a Flow Network N The Maximummentioning

confidence: 99%

“…Hence to improve it one would need to have informed at round t a set having a structure more complex than a linear one, the analysis would certainly be complicated. The example given for H (5) in Figure 1 uses non-linear set of vertices and two rounds, and any scheme using linear sets takes at least three rounds.…”

Section: Comparison Of the Different Algorithmsmentioning

confidence: 99%

Symmetric flows and broadcasting in hypercubes

Bermond¹,

Bonnecaze²,

Kodate³

et al. 1999

Annales De L’institut Fourier

View full text Add to dashboard Cite

“…As example, when G is a toroidal mesh one can use a sequence S made up of linear codes over a vector space Z n k (for broadcasting in the k-dimensional torus [7], for gossiping in respectively the 2 and 3-dimensional tori [8,5]). Here, in the case of the cube, the sequence will be built from linear binary codes.…”

Section: Multi-broadcast Flows and Symmetriesmentioning

confidence: 99%

Broadcasting in hypercubes in the circuit switched model

Bermond¹,

Kodate²,

Pérennès³

et al.

Proceedings 14th International Parallel and Distributed Processing Symposium. IPDPS 2000

Self Cite

View full text Add to dashboard Cite

In this paper, we propose a method which enables us to construct almost optimal broadcast schemes on an n-dimensional hypercube in the circuit switched, -port model. In this model, an initiator must inform all the nodes of the network in a sequence of rounds. During a round, vertices communicate along arc-disjoint dipaths. Our construction is based on particular sequences of nested binary codes having the property that each code can inform the next one in a single round. This last property is insured by a flow technique and results about symmetric flow networks.We apply the method to design new schemes improving and generalizing the previous results. Our schemes are the best possible algebraic schemes, and they are optimal in the case n = 2 p , 1.

show abstract

All-to-all broadcast in torus with wormhole-like routing

Cited by 20 publications

References 7 publications

On All-to-All Broadcast in Dense Gaussian Network On-Chip

On All-to-All Broadcast in Dense Gaussian Network On-Chip

Symmetric flows and broadcasting in hypercubes

Broadcasting in hypercubes in the circuit switched model

Contact Info

Product

Resources

About