Escolar Documentos
Profissional Documentos
Cultura Documentos
170
2013 IEEE Asia Pacific Conference on Postgraduate Research in Microelectronics and Electronics (PrimeAsia)
171
2013 IEEE Asia Pacific Conference on Postgraduate Research in Microelectronics and Electronics (PrimeAsia)
172
2013 IEEE Asia Pacific Conference on Postgraduate Research in Microelectronics and Electronics (PrimeAsia)
173
2013 IEEE Asia Pacific Conference on Postgraduate Research in Microelectronics and Electronics (PrimeAsia)
R EFERENCES
[1] J. Nurmi, Network-on-Chip: A New Paradigm for System-on-Chip
Design, in International Symposium on System-on-Chip, pp. 26, 2005.
[2] V. Rantala, T. Lehtonen, and J. Plosila, Network on Chip Routing
Algorithms, tech. rep., TUCS Technical Reports 779, Turku Centre
for Computer Science, 2006.
[3] Tilera Announces the worlds first 100-core processor. Online, Octo-
ber 2009. Available: http://goo.gl/K9c85.
[4] U. of Glasgow, Scientists Squeeze More Than 1,000 cores on to
Computer Chip. Online. Available: http://goo.gl/KdBbW.
[5] S. Kundu, R. P. Dasari, S. Chattopadhyay, and K. Manna, Mesh-of-
Tree Based Scalable Network-on-Chip Architecture, in IEEE Region
10 Colloquium and the Third ICIIS, 2008.
[6] J. Chen, P. Gillard, and C. Li, Performance evaluation of three
Network-on-Chip (NoC) architectures (Invited), in 1st IEEE Interna-
Fig. 7: Packet loss rate as a function of increasing network tional Conference on Communications in China (ICCC), pp. 9196,
load for 4 4 sized topology 2012.
[7] C. J. Glass and L. M. Ni, The Turn Model for Adaptive Routing,
Journal of the Association for Computing Machinery, vol. 41, pp. 874
902, September 1994.
transmission time. Packet delay for proposed topology (i.e. [8] Z. Song, G. Ma, and D. Song, Hierarchical Star: An Optimal NoC
HLMGS topology) reaches to thresholds value comparatively Topology for High-Performance SoC Design, in International Multi-
at higher load compared to others. An increase of 25% and symposiums on Computer and Computational Sciences (IMSCCS 08),
pp. 158163, 2008.
27% in maximum throughput compared to sd2d and l2star
[9] K.-J. Chen, C.-H. Peng, and F. Lai, Star-type architecture with low
and increase of 48% compared to simple 2d mesh has been transmission latency for a 2D mesh NOC, in IEEE Asia Pacific
observed for this proposed topology (see Fig.6). Packet loss Conference on Circuits and Systems (APCCAS), pp. 919922, 2010.
rate is also is negligible or very low for this proposed topology [10] P. Ghosal and T. S. Das, L2STAR: A Star Type level-2 2D Mesh
(as shown in Fig.7). Low packet drop rate signifies a low architecture for NoC, in Asia Pacific Conference on Postgraduate
channel contention problem also. This signifies a properly Research in Microelectronics and Electronics (PrimeAsia), pp. 155
distributed traffic throughout the network. A comparison on 159, 2012.
required area has been calculated followed by a method [11] M. Saneei, A. Afzali-Kusha, and Z. Navabi, Low-Latency Multi-Level
Mesh Topology for NoCs, in The 18th International Confernece on
proposed by S Suboh et. al in [20]. Where average required Microelectronics (ICM), pp. 3639, 2006.
area (Av ) has been calculated as follows
[12] P. Ghosal and T. S. Das, Network-on-chip routing using Structural
Diametrical 2D mesh architecture, in Third International Conference
Av = Ns (Rs + as dg Sf Bs ) + Nc Ac + al Nl Ll (1) on Emerging Applications of Information Technology (EAIT), pp. 471
474, 2012.
Number of switches (Ns ) is considered 16 for mesh and [13] Y. Fukushima, M. Fukushi, I. E. Yairi, and T. Hattori, A Hardware-
Oriented Fault-Tolerant Routing Algorithm for Irregular 2D-Mesh
21 for HLMGS, whereas average node degree for mesh and Network-on-Chip without Virtual Channels, in IEEE 25th International
proposed HLMGS are consider as 4 and 5.4 respectively. Link Symposium on Defect and Fault Tolerance in VLSI Systems (DFT),
length for leaf level connection has been kept 1, while link pp. 5259, 2010.
length for non-leaf connection is kept 2 units. A 24% to [14] Y. M. Boura and C. R. Das, Fault-tolerant routing in mesh networks,
12% area overhead has been observed by taking other variable in International Conference on Parallel Processing, pp. I.106I.109,
1995.
values as proposed in [20] and varying number of connected IP
core (N) to each router from 1 to 4 for this proposed topology [15] D. H. Linder and J. C. Harden, An adaptive and fault-tolerant
wormhole routing strategies for k-ary n-cubes, IEEE Transactions on
over 2dmesh. So require area overhead decreases with the Computer, vol. 40, pp. 212, 1991.
increasing IP core number. [16] A. A. Chien, A cost and speed model for k-ary n-cube wormhole
routers, in Hot Interconnects 93, 1993.
[17] The Network Simulator-NS-2. Online. Available:
VII. C ONCLUSION http://www.isi.edu/nsnam/ns/.
Though experimental result is quite convincing but lot [18] M. Ali, M. Welzl, A. Adnan, and F. Nadeem, Using the NS-2 Network
Simulator for Evaluating Network on Chips (NoC), in International
more things are still left to experiment in future. For example Conference on Emerging Technologies (ICET 06), pp. 506512, 2006.
result is observed under static routing and assumption of [19] Y.-R. Sun, S. Kumar, and A. Jain, Simulation and Evaluation for a net-
a single resource connected to each leaf router. So how it work on chip architecture using NS-2, in 20th NORCHIP conference,
will perform under dynamic routing and what will be the 2002.
requisite channel bandwidth with the increasing number of [20] S. Suboh, M. Bakhouya, J. Gaber, and T. El-Ghazawi, Analytical mod-
resource connected to each leaf router. Those things need to eling and evaluation of network-on-chip architectures, in International
be addressed in future work. A result under different kind Conference on High Performance Computing and Simulation (HPCS),
pp. 615622, 2010.
of traffic pattern is also a subject to observe. However, the
observed experimental result shows the superiority of proposed
topology over other compared topologies with respect to some
performance centric measurement parameter like throughput,
latency, load balancing and packet loss rate in cost of additional
area overhead.
174