ability of controlling the number of used processors makes
it possible for users to select the optimal number which can
achieve the maximum speedup. In addition, users can flexibly
allocate the computational resources in a computer for a spe-
cific parallel procedure by setting the number of processors.
The execution times and speedups of the six algorithms on
the computers with Windows OS and Linux OS are indi-
cated in Table 4 and Table 5, respectively. In the two tables,
the shortest times and the maximum speedups are bold. The
speedups achieved by different numbers of processors are
also shown in Figure 8 and Figure 9, which correspond to the
Windows computer and the Linux computer, respectively.
These results show that the maximum speedups vary for the
different algorithms and computing platforms. The character-
istics of parallel performance attained by different numbers of
processors are summarized in the following.
T
able
4. E
xecution
T
imes
(I
ncluding
D
isk
I/O
and
C
omputation
,
in
S
econds
)
of
D
ifferent
N
umbers
of
P
rocessors
on
the
W
indows
C
omputer
and
S
peedup
Number of Processors
1
2
4
6
8
10 12 14 16 18 20 22 24
Filtering
Times
317 274 113 78 65
59
61 68 80 81 83 83 93
Speedup 1.0 1.2 2.8 4.1 4.9
5.4
5.2 4.7 4.0 3.9 3.8 3.8 3.4
De-correlation
Times
277 328 118 75 56 46 42 56 53 49 45 43
39
Speedup 1.0 0.8 2.3 3.7 4.9 6.0 6.6 4.9 5.2 5.7 6.2 6.4
7.1
RPC-based correction
Times
48 42 18 13 11 11
10
11 12 12 11 12 12
Speedup 1.0 1.1 2.7 3.7 4.4 4.4
4.8
4.4 4.0 4.0 4.4 4.0 4.0
BR fusion
Times
683 660 419 191 119 136 116 173 153 119 126 118
112
Speedup 1.0 1.0 1.6 3.6 5.7 5.0 5.9 3.9 4.5 5.7 5.4 5.8
6.1
Mosaics
Times
228 210 78 51 41 35
31
40 37 35 32 31 31
Speedup 1.0 1.1 2.9 4.5 5.6 6.5
7.4
5.7 6.2 6.5 7.1 7.4 7.4
DEM extractions
Times 5434 5410 1843 1153 822 637 519 657 614 527 475 440
401
Speedup 1.0 1.0 2.9 4.7 6.6 8.5 10.5 8.3 8.9 10.3 11.4 12.4
13.6
T
able
5. E
xecution
T
imes
(I
ncluding
D
isk
I/O
and
C
omputation
,
in
S
econds
)
of
D
ifferent
N
umbers
of
P
rocessors
on
the
L
inux
C
omputer
and
S
peedup
Number of Processors
1
2
4
6
8
10 12 14 16
Filtering
Times
408 360 134 142 110 95 81 76
67
Speedup 1.0 1.1 3.0 2.9 3.7 4.3 5.0 5.4
6.1
De-correlation
Times
964 882 304 189 197 185 152 137
122
Speedup 1.0 1.1 3.2 5.1 4.9 5.2 6.3 7.0
7.9
RPC-based correction
Times
58 55 20 12 13 11
9
11
8
Speedup 1.0 1.1 2.9 4.8 4.5 5.3 6.4 5.3
7.3
BR fusion
Times 1608 1558 799 376 359 354 288 280
241
Speedup 1.0 1.0 2.0 4.3 4.5 4.5 5.6 5.8
6.7
Mosaics
Times
213 189 68 52
38
54 43 44 47
Speedup 1.0 1.1 3.1 4.1
5.6
3.9 5.0 4.8 4.5
DEM extractions
Times 14081 14066 4767 4863 3643 2804 2347 2024
1752
Speedup 1.0 1.0 3.0 2.9 3.9 5.0 6.0 7.0
8.0
Figure 8. Speedups of different numbers of processors on the
Windows computer.
Figure 9. Speedups of different numbers of processors on the Linux
computer.
382
May 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
