Information Technology Reference
In-Depth Information
Table 9.5:
Results of test cases for benchmark suite 1 (20 runs of each
test case) and suite 2 (3 runs of each test case).
Test case
Nr of Tiles
Runtime in seconds
Runtime
agents
Avg
Min
Max
(normalized)
TC P 11
10
3
19.90
19.52
20.30
1.00
TC P 12
20
13
40.65
39.78
41.17
2.04
TC P 13
30
18
66.09
36.96
67.00
3.32
TC P 14
40
22
90.28
89.61
91.29
4.54
TC P 15
50
27
107.45
105.46
108.89
5.40
TC P 16
100
60
215.61
211.82
219.29
10.84
TC P 17
200
111
439.96
434.47
446.10
22.11
TC P 11 b
10
3
19.80
19.61
20.03
1.00
TC P 12 b
20
13
40.65
40.18
41.13
2.05
TC P 13 b
30
18
65.98
65.33
66.68
3.33
TC P 14 b
40
22
90.15
89.05
91.13
4.55
TC P 15 b
50
27
108.02
106.72
109.93
5.45
TC P 16 b
100
60
217.51
215.41
220.25
10.98
TC P 17 b
200
111
451.64
445.46
456.23
22.81
TC P 21
1 000
2 150
4587.92
4548.60
4616.50
1.00
TC P 22
2 000
3 730
9598.96
9405.94
9725.28
2.09
TC P 23
3 000
4 909
14548.35
14350.38
14714.94
3.17
TC P 24
4 000
5 648
19791.52
19395.85
20048.88
4.31
TC P 25
5 000
6 212
24803.27
24603.33
25094.80
5.41
TC P 26
10 000
7 265
54764.76
54206.57
55252.98
11.94
TC P 21 b
1 000
2 150
4946.49
4886.69
4984.89
1.00
TC P 22 b
2 000
3 730
10692.24
10487.35
10795.47
2.16
TC P 23 b
3 000
4 909
17360.08
17329.65
17379.97
3.51
TC P 24 b
4 000
5 648
24502.99
24450.30
24543.21
4.95
TC P 25 b
5 000
6 212
32164.10
31826.07
32601.67
6.50
TC P 26 b
10 000
7 265
84363.13
84131.88
84591.67
17.06
Suite 1 suggests an almost linear scaling of runtime as a function of
the number of agents (at least for small numbers of agents up to 200).
In contrast, suite 2 illustrates clearly the effect of constraints within
a model (see also Figure 9.14). Whereas the runtime growths almost
linearly for test cases without constraint, the presence of a constraint
