12.
If p is non-preemptive then
13.
Find a free I/O pair;
14.
If no such I/O pair is available then
15.
Update delivery time of p; Repeat
from 11;
16.
else if p is preemptive or p is BIST then
17.
Find a free access path to/from core i;
18.
If no such path is available then
19.
Update delivery time for p; Repeat
from 11;
20.
Calculate duration of packet transmission;
21.
Calculate power consumption for packet
transmission;
22.
If power constraint is met then
23.
Assign packet to the chosen path;
24.
Update schedule and time tags;
25.
If p is non-preemptive then
26.
Assign response packet to chosen I/O
pair;
27.
Update schedule and time tags;
28.
else if p is preemptive or p is BIST then
29.
Define delivery time of next packet of
core i;
30.
Update L t ;
31.
else
32.
Update packet delivery time; Repeat from 11;
33.
If all packets have been attempted then
34.
Update current time; Repeat from 9;
35. End.
258
Network-on-Chip
8.3.4 PSO-Based Strategy
Apart from the ILP and heuristic approaches, discussed so far for testing
cores, the meta-search techniques can also be employed to determine a test
schedule. Farah and Harmanani (2010) has suggested a simulated annealing
based strategy to generate a schedule for testing the cores in the NoC. In the
following, a discrete PSO (DPSO)-based formulation will be discussed for
the NoC core test scheduling problem to minimize the test application time.
A similar DPSO formulation was discussed in Section 5.6. Hence, only the
particle structure and evolution mechanism is discussed in Section 8.3.4.1.
8.3.4.1 Particle Structure and Fitness
A particle corresponds to possible test scheduling order of the cores. It has
two components—the core order part (core part) and the I/O assignment