Bench report : P1OptV2vsHanJunvsRahValvsChen (2013-01-23-17-35)
Thursday 24 th January, 2013
Contents
1 benchs : MassAssembling ( P 1 -Lagrange) 2
1.1 mesh : carre4-1 . . . . 2 1.2 mesh : disque4-1 . . . . 3
2 benchs : StiAssembling ( P 1 -Lagrange) 5
2.1 mesh : carre4-1 . . . . 5
2.2 mesh : disque4-1 . . . . 6
1 benchs : MassAssembling ( P 1 -Lagrange)
1.1 mesh : carre4-1
n q P1OptV2 P1HanJun P1RahValT P1ChenT P1iFEMT
3043 0.018 (s)
x 1.00 0.020 (s)
x 0.90 0.012 (s)
x 1.44 0.004 (s)
x 4.99 0.009 (s) x 1.93 26910 0.047 (s)
x 1.00 0.057 (s)
x 0.82 0.035 (s)
x 1.32 0.034 (s)
x 1.38 0.043 (s) x 1.09 74326 0.136 (s)
x 1.00 0.170 (s)
x 0.80 0.107 (s)
x 1.27 0.100 (s)
x 1.36 0.124 (s) x 1.09 145828 0.279 (s)
x 1.00 0.397 (s)
x 0.70 0.247 (s)
x 1.13 0.255 (s)
x 1.09 0.256 (s) x 1.09 241024 0.455 (s)
x 1.00 0.709 (s)
x 0.64 0.457 (s)
x 0.99 0.450 (s)
x 1.01 0.418 (s) x 1.09 360040 0.682 (s)
x 1.00 1.089 (s)
x 0.63 0.690 (s)
x 0.99 0.679 (s)
x 1.00 0.636 (s) x 1.07 505379 0.957 (s)
x 1.00 1.532 (s)
x 0.63 0.978 (s)
x 0.98 0.957 (s)
x 1.00 0.924 (s) x 1.04 668176 1.348 (s)
x 1.00 2.142 (s)
x 0.63 1.399 (s)
x 0.96 1.392 (s)
x 0.97 1.338 (s) x 1.01 866481 1.795 (s)
x 1.00 2.820 (s)
x 0.64 1.819 (s)
x 0.99 1.815 (s)
x 0.99 1.792 (s) x 1.00 1067083 2.204 (s)
x 1.00 3.489 (s)
x 0.63 2.262 (s)
x 0.97 2.248 (s)
x 0.98 2.312 (s) x 0.95 1197591 2.441 (s)
x 1.00 3.882 (s)
x 0.63 2.531 (s)
x 0.96 2.520 (s)
x 0.97 2.491 (s) x 0.98 1443408 3.021 (s)
x 1.00 4.881 (s)
x 0.62 3.108 (s)
x 0.97 3.110 (s)
x 0.97 3.254 (s) x 0.93 1707938 3.604 (s)
x 1.00 5.787 (s)
x 0.62 3.679 (s)
x 0.98 3.674 (s)
x 0.98 3.818 (s) x 0.94 2005296 4.247 (s)
x 1.00 6.861 (s)
x 0.62 4.334 (s)
x 0.98 4.321 (s)
x 0.98 4.605 (s) x 0.92 2339961 4.969 (s)
x 1.00 8.155 (s)
x 0.61 5.059 (s)
x 0.98 5.061 (s)
x 0.98 5.402 (s) x 0.92 2666201 5.674 (s)
x 1.00 9.368 (s)
x 0.61 5.772 (s)
x 0.98 5.775 (s)
x 0.98 6.100 (s) x 0.93
Table 1: Functions MassAssembling computation times in seconds (top value) and speedup (bottom value) on unit square mesh. The speedup reference is P1OptV2 version. Compute with Octave (3.6.3)
2
10
310
410
510
610
710
−310
−210
−110
010
110
2Sparse Matrix size (n
q)
time (s)
P1OptV2 P1HanJun P1RahValT P1ChenT P1iFEMT O(n
q) O(n
qlog(n
q
))
0 0.5 1 1.5 2 2.5 3
x 10
60.5
1 1.5 2 2.5 3 3.5 4 4.5 5
Sparse Matrix size (n
q)
Speed Up X
P1HanJun P1RahValT P1ChenT P1iFEMT
Figure 1: Functions MassAssembling computation times in seconds (left) and speedup (right) on unit square mesh. The speedup reference is P1OptV2 version.
1.2 mesh : disque4-1
10
310
410
510
610
710
−310
−210
−110
010
110
2Sparse Matrix size (n
q)
time (s)
P1OptV2 P1HanJun P1RahValT P1ChenT P1iFEMT O(n
q) O(n
qlog(n
q
))
0 0.5 1 1.5 2 2.5 3 3.5
x 10
60.5
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
Sparse Matrix size (n
q)
Speed Up X
P1HanJun P1RahValT P1ChenT P1iFEMT
Figure 2: Functions MassAssembling computation times in seconds (left) and speedup (right) on unit disk
mesh. The speedup reference is P1OptV2 version.
n q P1OptV2 P1HanJun P1RahValT P1ChenT P1iFEMT 3576 0.006 (s)
x 1.00 0.008 (s)
x 0.70 0.004 (s)
x 1.40 0.004 (s)
x 1.43 0.006 (s) x 0.97 31575 0.051 (s)
x 1.00 0.065 (s)
x 0.80 0.040 (s)
x 1.29 0.039 (s)
x 1.33 0.051 (s) x 1.02 86488 0.152 (s)
x 1.00 0.199 (s)
x 0.76 0.125 (s)
x 1.22 0.123 (s)
x 1.24 0.148 (s) x 1.03 170355 0.309 (s)
x 1.00 0.462 (s)
x 0.67 0.284 (s)
x 1.09 0.282 (s)
x 1.10 0.294 (s) x 1.05 281769 0.515 (s)
x 1.00 0.828 (s)
x 0.62 0.523 (s)
x 0.99 0.518 (s)
x 1.00 0.497 (s) x 1.04 424178 0.799 (s)
x 1.00 1.297 (s)
x 0.62 0.820 (s)
x 0.97 0.800 (s)
x 1.00 0.769 (s) x 1.04 582024 1.101 (s)
x 1.00 1.801 (s)
x 0.61 1.145 (s)
x 0.96 1.127 (s)
x 0.98 1.091 (s) x 1.01 778415 1.549 (s)
x 1.00 2.530 (s)
x 0.61 1.633 (s)
x 0.95 1.617 (s)
x 0.96 1.570 (s) x 0.99 992675 2.020 (s)
x 1.00 3.237 (s)
x 0.62 2.095 (s)
x 0.96 2.075 (s)
x 0.97 2.049 (s) x 0.99 1251480 2.697 (s)
x 1.00 4.190 (s)
x 0.64 2.684 (s)
x 1.01 2.682 (s)
x 1.01 2.666 (s) x 1.01 1401129 2.887 (s)
x 1.00 4.874 (s)
x 0.59 3.161 (s)
x 0.91 2.989 (s)
x 0.97 3.025 (s) x 0.95 1671052 3.622 (s)
x 1.00 5.750 (s)
x 0.63 3.646 (s)
x 0.99 3.630 (s)
x 1.00 3.829 (s) x 0.95 1978602 4.176 (s)
x 1.00 6.766 (s)
x 0.62 4.293 (s)
x 0.97 4.277 (s)
x 0.98 4.478 (s) x 0.93 2349573 4.966 (s)
x 1.00 8.267 (s)
x 0.60 5.155 (s)
x 0.96 5.125 (s)
x 0.97 5.499 (s) x 0.90 2732448 5.862 (s)
x 1.00 10.556 (s)
x 0.56 6.080 (s)
x 0.96 6.078 (s)
x 0.96 6.575 (s) x 0.89 3085628 6.634 (s)
x 1.00 11.109 (s)
x 0.60 6.833 (s)
x 0.97 6.793 (s)
x 0.98 7.500 (s) x 0.88
Table 2: Functions MassAssembling computation times in seconds (top value) and speedup (bottom value) on unit disk mesh. The speedup reference is P1OptV2 version. Compute with Octave (3.6.3)
4
2 benchs : StiAssembling ( P 1 -Lagrange)
2.1 mesh : carre4-1
n q P1OptV2 P1HanJun P1RahValT P1ChenT P1iFEMT
3043 0.005 (s)
x 1.00 0.017 (s)
x 0.27 0.038 (s)
x 0.12 0.005 (s)
x 1.03 0.006 (s) x 0.77 26910 0.042 (s)
x 1.00 0.093 (s)
x 0.45 0.107 (s)
x 0.39 0.041 (s)
x 1.03 0.050 (s) x 0.85 74326 0.127 (s)
x 1.00 0.302 (s)
x 0.42 0.313 (s)
x 0.41 0.128 (s)
x 0.99 0.146 (s) x 0.88 145828 0.271 (s)
x 1.00 0.626 (s)
x 0.43 0.626 (s)
x 0.43 0.287 (s)
x 0.94 0.299 (s) x 0.91 241024 0.455 (s)
x 1.00 1.088 (s)
x 0.42 1.056 (s)
x 0.43 0.527 (s)
x 0.86 0.512 (s) x 0.89 360040 0.688 (s)
x 1.00 1.638 (s)
x 0.42 1.687 (s)
x 0.41 0.818 (s)
x 0.84 0.783 (s) x 0.88 505379 0.999 (s)
x 1.00 2.507 (s)
x 0.40 2.500 (s)
x 0.40 1.208 (s)
x 0.83 1.130 (s) x 0.88 668176 1.324 (s)
x 1.00 3.242 (s)
x 0.41 3.299 (s)
x 0.40 1.618 (s)
x 0.82 1.562 (s) x 0.85 866481 1.782 (s)
x 1.00 4.388 (s)
x 0.41 4.538 (s)
x 0.39 2.158 (s)
x 0.83 2.097 (s) x 0.85 1067083 2.238 (s)
x 1.00 5.402 (s)
x 0.41 5.684 (s)
x 0.39 2.779 (s)
x 0.81 2.650 (s) x 0.84 1197591 2.599 (s)
x 1.00 6.538 (s)
x 0.40 6.514 (s)
x 0.40 3.143 (s)
x 0.83 3.234 (s) x 0.80 1443408 3.228 (s)
x 1.00 7.942 (s)
x 0.41 7.942 (s)
x 0.41 3.898 (s)
x 0.83 4.013 (s) x 0.80 1707938 3.775 (s)
x 1.00 9.318 (s)
x 0.41 9.408 (s)
x 0.40 4.457 (s)
x 0.85 4.496 (s) x 0.84 2005296 4.297 (s)
x 1.00 10.476 (s)
x 0.41 10.920 (s)
x 0.39 5.512 (s)
x 0.78 5.907 (s) x 0.73 2339961 5.194 (s)
x 1.00 12.929 (s)
x 0.40 13.327 (s)
x 0.39 6.763 (s)
x 0.77 7.234 (s) x 0.72 2666201 5.942 (s)
x 1.00 14.571 (s)
x 0.41 14.632 (s)
x 0.41 7.215 (s)
x 0.82 8.298 (s) x 0.72
Table 3: Functions StiffAssembling computation times in seconds (top value) and speedup (bottom value)
on unit square mesh. The speedup reference is P1OptV2 version. Compute with Octave (3.6.3)
10
310
410
510
610
710
−310
−210
−110
010
110
2Sparse Matrix size (n
q)
time (s)
P1OptV2 P1HanJun P1RahValT P1ChenT P1iFEMT O(n
q) O(n
qlog(n
q
))
0 0.5 1 1.5 2 2.5 3
x 10
60.1
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
Sparse Matrix size (n
q)
Speed Up X
P1HanJun P1RahValT P1ChenT P1iFEMT
Figure 3: Functions StiffAssembling computation times in seconds (left) and speedup (right) on unit square mesh. The speedup reference is P1OptV2 version.
2.2 mesh : disque4-1
10
310
410
510
610
710
−310
−210
−110
010
110
2Sparse Matrix size (n
q)
time (s)
P1OptV2 P1HanJun P1RahValT P1ChenT P1iFEMT O(n
q) O(n
qlog(n
q