Information Technology Reference
In-Depth Information
Table 2.
DNA instance results for CPU-DFA and GPU-DFA
Instances
CPU-DFA
GPU-DFA
ST
pm
Best
Avg. Time (sec.)
Best
Avg. Time (sec.)
16 11478.00
11084
.
93
±
81
.
12
4
.
13 11478
.
00
11420
.
23
±
45
.
77
2
.
38
•
16
32 11478
.
00
11390
.
93
±
41
.
48
2
.
33 11478
.
00
11469
.
60
±
25
.
63
1
.
51
•
48 11478
.
00
11469
.
83
±
25
.
95
1
.
76 11478
.
00
11473
.
73
±
11
.
29
1
.
72
•
16 11478
.
00
11464
.
93
±
58
.
19
7
.
73 11478
.
00
11462
.
00
±
35
.
16
2
.
72
•
x
60189
4
32
32 11478
.
00
11574
.
73
±
31
.
43
4
.
06 11478
.
00
11466
.
80
±
28
.
80
1
.
59
•
48 11478
.
00
11467
.
93
±
16
.
75
2
.
90 11478
.
00
11475
.
20
±
15
.
21
1
.
19
•
16 11478
.
00
11608
.
03
±
47
.
49
11
.
26
11478
.
00
11470
.
23
±
23
.
87
1
.
25
•
48
32 11478
.
00
11444
.
47
±
39
.
73
5
.
79 11478
.
00
11462
.
40
±
31
.
36
0
.
86
•
48 11478
.
00
11466
.
00
±
23
.
06
4
.
06 11478
.
00
11463
.
93
±
27
.
71
0
.
77
•
16 47963
.
00
45230
.
90
±
312
.
77
52
.
37 47803
.
00
46219
.
96
±
184
.
61
5
.
37
•
16
32 47550
.
00
47357
.
63
±
180
.
33
27
.
21 48050
.
00
47935
.
93
±
90
.
40
12
.
62
•
48 47811
.
00
47593
.
13
±
112
.
79
18
.
88 47919
.
00
47696
.
47
±
136
.
71
3
.
32
•
16 47521
.
00
46255
.
50
±
297
.
18
104
.
34 47750
.
00
47511
.
07
±
154
.
07
16
.
64
•
m
154216
6
32
32 48052
.
00
47416
.
60
±
130
.
49
53
.
41
48052
.
00
47673
.
57
±
114
.
99
8
.
96
•
48 47881
.
00
47651
.
07
±
117
.
43
36
.
51 47917
.
00
47724
.
60
±
132
.
16
6
.
49
•
16 46931
.
00
46348
.
87
±
277
.
81
157
.
36 46817
.
00
46223
.
69
±
296
.
13
11
.
85
•
48
32 47569
.
00
47204
.
63
±
183
.
40
80
.
54 47612
.
00
47174
.
77
±
201
.
70
7
.
82
•
48 47918
.
00
47429
.
10
±
206
.
12
54
.
76 47961
.
00
47688
.
36
±
100
.
61
6
.
98
•
16 212802
.
00 211408
.
97
±
863
.
62
393
.
18 213512
.
00 211709
.
77
±
977
.
70
139
.
73
•
32 225918
.
00 224496
.
94
±
916
.
74
199
.
19 226165
.
00 225110
.
91
±
1017
.
12
101
.
67
•
16
48 218523
.
00 216496
.
00
±
1054
.
03
134
.
93 217649
.
00 216800
.
73
±
1100
.
46
72
.
70
•
16 211857
.
00 210783
.
79
±
1163
.
03
787
.
43 212381
.
00 210599
.
81
±
1711
.
13
128
.
78
•
bx
842596
4
32
32 213821
.
00 211529
.
03
±
1887
.
81
395
.
64 212479
.
00 210181
.
33
±
1281
.
12
66
.
22
•
48 212659
.
00 211233
.
76
±
1291
.
36
266
.
27 212177
.
00 211918
.
10
±
1151
.
91
45
.
03
•
16 209954
.
00 208333
.
23
±
990
.
41
1185
.
84 206025
.
00 203691
.
00
±
1280
.
78
70
.
62
•
48
32 207797
.
00 204743
.
23
±
1293
.
12
596
.
09 207522
.
00 206787
.
00
±
1484
.
26
46
.
31
•
48 193276
.
00 189610
.
83
±
1641
.
84
399
.
42 192587
.
00 191152
.
00
±
1679
.
52
31
.
93
•
Table 2 indicates that the GPU-DFA algorithm obtains lower times in all
the instances for all the configurations. These good times might be due to the
FA operations that are translated into a parallel model that can maximize the
eciency of each thread and thus, the simplicity of each kernel is maintained.
Concerning the amount of gain time obtained, we have computed this metric
by dividing the time of the CPU-DFA with the GPU-DFA. The GPU-DFA gain
time ranges from 1.02 to 16.79. In fact, the execution times in Table 2 confirm
this fact, since the execution time when using GPU-DFA is much lower than the
CPU's. The results clearly indicate that executing CPU-DFA is more expensive,
when compared with the GPU version in all the DNA-FAP instances.
Additionally, in order to make a better comparison between both versions
developed in this work, namely GPU-DFA and CPU-DFA, Fig. 3, Fig. 4 and
Fig. 5 display the gain time factor obtained for each configuration among the
three selected DNA-FAP instances, respectively. As an initial observation, we
can say that the time gain values are above the value 1.00, which indicates that
the CPU has always spent more execution time than GPU. In the same way,
we see that the biggest gain factors of the time appeared in smaller parameter
settings, especially those with
m
= 16. These figures clearly indicate that the
GPU-DFA is the faster algorithm for all the instances. With respect to the time
gain factor values, they range from 1.02 to 16.79.
