Optical network technologies for HPC: computer-architects point of view

Koibuchi, Michihiro; Fujiwara, Ikki; Ishii, Kiyo; Namiki, Shu; Chaix, Fabien; Matsutani, Hiroki; Amano, Hideharu; Kudoh, Tomohiro

doi:10.1587/elex.13.20152007

Cited by 18 publications

(10 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Supposing that there are N link links in g k , according to Eq. (1), N link link occupancy rates can be calculated. The average link occupancy rate of g k can be calculated as follows:…”

Section: Network Congestion Factor Computation Modelmentioning

confidence: 99%

“…Progresses in silicon nanophotonics have made optical network-on-chip (ONoC) a promising alternative to electronic network-on-chip (ENoC) owing to its many advantages such as high bandwidth, high signal propagation speed, low energy consumption and electromagnetic interference [1]- [4]. Several ONoC architectures have been proposed for future multicore microprocessors, such as Mesh [5], Corona [6], RONoC [7] and PHENIC [8].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An Optimization Algorithm to Build Low Congestion Multi-Ring Topology for Optical Network-on-Chip

Zhu

Wang

Zhou

et al. 2018

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYRing-based topology is popular for optical network-onchip. However, the network congestion is serious for ring topology, especially when optical circuit-switching is employed. In this paper, we proposed an algorithm to build a low congestion multi-ring architecture for optical network-on-chip without additional wavelength or scheduling overhead. A network congestion model is established with new network congestion factor defined. An algorithm is developed to optimize the low congestion multi-ring topology. Finally, a case study is shown and the simulation results by OPNET verify the superiority over the traditional ONoC architecture.

show abstract

“…Supposing that there are N link links in g k , according to Eq. (1), N link link occupancy rates can be calculated. The average link occupancy rate of g k can be calculated as follows:…”

Section: Network Congestion Factor Computation Modelmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

An Optimization Algorithm to Build Low Congestion Multi-Ring Topology for Optical Network-on-Chip

Zhu

Wang

Zhou

et al. 2018

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

show abstract

“…This is a fundamental problem in graph theory [1,4,11,19]. As a regular graph often models a network topology, which the degree of each node is limited due to some physical constraints, this problem has an important application to design of efficient network topologies in HPC (High Performance Computing) area [8,14,15,20].…”

Section: Introductionmentioning

confidence: 99%

The Diameter of Dense Random Regular Graphs

Shimizu

2018

Proceedings of the Twenty-Ninth Annual ACM-SIAM Symposium on Discrete Algorithms

View full text Add to dashboard Cite

There is a tight upper bound on the order (the number of vertices) of any d-regular graph of diameter D, known as the Moore bound in graph theory. This bound implies a lower bound D 0 (n, d) on the diameter of any d-regular graph of order n. Actually, the diameter diam (G n,d ) of a random dregular graph G n,d of order n is known to be asymptotically "optimal" as n → ∞. Bollobás and de la Vega (1982) holds w.h.p. (with high probability) for fixed d ≥ 3, whereas there exists a gapIn this paper, we investigate the gap diam( (1)) n α where α ∈ (0, 1) and β > 0 are arbitrary constants. We prove that diam(G n,d ) = ⌊α −1 ⌋ + 1 holds w.h.p. for such d. Our result implies that the gap is 1 if α −1 is an integer and d ≥ n α , and is 0 otherwise. One can easily obtain that diam (G n,d ) ≤ ⌊α −1 ⌋ + 1 holds w.h.p. by using the embedding theorem due to Dudek et al. (2017). Our critical contribution is to show that diam (G n,d ) ≥ ⌊α −1 ⌋ + 1 holds w.h.p. by the analysis of the distances of fixed vertex pairs.

show abstract

“…This problem is called the order/degree problem, and has not been studied sufficiently. Recently, the importance of the order/degree problem is pointed out [9].…”

Section: Introductionmentioning

confidence: 99%

“…Some of the authors of [9] and their coworkers held a competition called "Graph Golf" on the order/degree problem [10].…”

Section: Introductionmentioning

confidence: 99%

Average shortest path length of graphs of diameter 3

Mori

2016

2016 Tenth IEEE/ACM International Symposium on Networks-on-Chip (NOCS)

View full text Add to dashboard Cite

Abstract-A network topology with low average shortest path length (ASPL) provides efficient data transmission while the number of nodes and the number of links incident to each node are often limited due to physical constraints. In this paper, we consider the construction of low ASPL graphs under these constraints by using stochastic local search (SLS) algorithms. Since the ASPL cannot be calculated efficiently, the ASPL is not suitable for the evaluation function of SLS algorithms. We first derive an equality and bounds for the ASPL of graphs of diameter 3. On the basis of the simplest upper bound of the ASPL, we propose to use 3 + 22 as the evaluation function for graphs of diameter 3 where and 2 denote the number of triangles and squares in a graph, respectively. We show that the proposed evaluation function can be evaluated in O(1) time as the number of nodes and the maximum degree tend to infinity by using some data tables. By using the simulated annealing with the proposed evaluation function, we construct low ASPL regular graphs of diameter 3 with 10 000 nodes.

show abstract

Optical network technologies for HPC: computer-architects point of view

Cited by 18 publications

References 32 publications

An Optimization Algorithm to Build Low Congestion Multi-Ring Topology for Optical Network-on-Chip

An Optimization Algorithm to Build Low Congestion Multi-Ring Topology for Optical Network-on-Chip

The Diameter of Dense Random Regular Graphs

Average shortest path length of graphs of diameter 3

Contact Info

Product

Resources

About