86
Network-on-Chip
Source
d = 0
d = 2
d = 4
d = 2
d = 2
d = 4
d = 4
d = 6
d = 4
d = 4
d = 6
d = 6
d = 6
d = 2
d = 4
d = 4
d = 4
d = 4
d = 6
d = 6
d = 6
d = 6
d = 8
d = 8
d = 8
d = 8
d = 8
d = 8
d = 8
d = 8
d = 4
L
S
S
S
R
R
R
R
R
R
R
R
S
L
L
L
L
S
S
S
S
L
L
L
L
S
S
S
S
S
S
S
S
L
L
L
L
L
L
L
Figure 4.7
Distances of the destination cores from any source in 4 × 4 MoT. L, S, and R represent Leaf,
Stem, and Root routers respectively.
• (6 * 50)/(2 +  4 +  6 +  8) =  15% of the total traffic goes to the cluster
having d  = 4
• (4 * 50)/(2 +  4 +  6 +  8) =  10% of the total traffic goes to the cluster
having d  = 6
• (2  *  50)/(2  +  4  +  6  +  8)  =  5% of the total traffic goes to the cluster
having d  =  8
Now, as there are more than one destination cores in some clusters (d  =  2, 4,
6, and 8), the traffic gets randomly distributed among them.
The effect of traffic spatial localization on the throughput of the MoT-based
network is shown in Figure 4.8. It can be observed that network throughput
increases with increasing locality factor. This is due to the fact that as the
locality factor increases, more traffic is destined for their local clusters, thus
traversing lesser number of hops, which in turn increases throughput.
4.4.3 Average Overall Latency at Different Locality Factors
The average overall latency of any network depends on both the offered load
and the locality factor. Figure   4.9 shows the average overall latency profile
with an offered load under uniformly distributed and localized traffic in a