C from Matlab 0.019 0.076 0.17 0.302 0.472 0.681 0.926 1.209
Orio (Seq.) 0.017 0.069 0.154 0.28 0.431 0.623 0.844 1.215
Orio (Par.) 0.011 0.035 0.078 0.14 0.21 0.303 0.414 0.533
BLAS 0.0637 0.264 0.594 1.08 1.7 2.43 3.31 4.32
Intel MKL 0.0321 0.0686 0.154 0.277 0.433 0.62 0.845 1.11
ATLAS ##### 0.000166 0.000338 0.00057 0.000865 0.00121 0.00163 0.0021
1.53 1.888
1.383 1.744
0.678
5.45 6.78
1.39 1.73
0.00263 0.00292
C from Matlab 0.038 0.156 0.351 0.625 0.974 1.404 1.907 2.488 3.136 3.869
Orio (Seq.) 0.012 0.047 0.105 0.189 0.294 0.426 0.58 0.782 0.948 1.178
Orio (Par.) 0.009 0.024 0.053 0.088 0.134 0.193 0.272 0.411 0.582 0.789
BLAS 0.0209 0.0862 0.194 0.357 0.564 0.807 1.1 1.43 1.8 2.25
Intel MKL 0.0518 0.0684 0.154 0.276 0.431 0.619 0.845 1.1 1.39 1.73
ATLAS 0.00935 0.0418 0.125 0.316 0.556 0.835 1.14 1.51 1.9 2.35
GEMVER Sequential
2000 4000 6000 8000 10000 12000 14000
C from Matlab 0.038 0.157 0.352 0.626 0.976 1.406 1.909
Orio (S) 0.029 0.118 0.263 0.48 0.759 1.075 1.447
Orio (P) 0.037 0.112 0.239 0.42 0.646 0.923 1.25
BLAS 0.064 0.265 0.598 1.091 1.711 2.456 3.34
Intel MKL 0.037 0.149 0.326 0.581 0.908 1.3 1.78
ATLAS 0.039 0.159 0.395 0.787 1.29 1.9 2.582
C from Matlab Orio (S)
BiCGkernel Orio (P) BLAS
2000 2.5 4000 6000 8000 10000 12000 14000
C from Matlab 0.01 0.046 0.104 0.185 0.29 0.417 0.569
Orio (S) 0.009 0.036 0.081 0.144 0.226 0.325 0.445
Orio (P) 0.094 2 0.033 0.048 0.083 0.138 0.183 0.245
BLAS 0.0209 0.0863 0.194 0.358 0.563 0.807 1.1
Time (Sec.)
Intel MKL 0.0246 0.0688 0.154 0.277 0.432 0.62 0.846
ATLAS 0.00927 0.0415 0.129 0.317 0.557 0.838 1.14
1.5
1
0.5
0
2000 4000 6000 8000 10000 12000 14000 16000
GESUMMV
Matrix Size
2000 4000 6000 8000 10000 12000 14000
C from Matlab 0.019 0.076 0.17 0.302 0.472 0.681 0.926
Orio (S) 0.017 0.069 0.154 0.28 0.431 0.623 0.844
Orio (P) 0.011 0.035 0.078 0.14 0.21 0.303 0.414
BLAS 0.0637 0.264 0.594 1.08 1.7 2.43 3.31
Intel MKL 0.0321 0.0686 0.154 0.277 0.433 0.62 0.845
ATLAS 5.36E-05 0.000152 0.00031 0.00052 0.00086 0.0011 0.00163
ATAX
2000 4000 6000 8000 10000 12000 14000
C from Matlab 0.038 0.156 0.351 0.625 0.974 1.404 1.907
Orio (S) 0.012 0.047 0.105 0.189 0.294 0.426 0.58
Orio (P) 0.009 0.024 0.053 0.088 0.134 0.193 0.272
BLAS 0.0209 0.0862 0.194 0.357 0.564 0.807 1.1
Intel MKL 0.0518 0.0684 0.154 0.276 0.431 0.619 0.845
ATLAS 0.00935 0.0418 0.125 0.316 0.556 0.835 1.14
VADD
2000 4000 6000 8000 10000 12000 14000
C from Matlab
Orio (S)
Orio (P)
BLAS 1.36E-05 2.90E-05 ##### ##### ##### 8.61E-05 0.0001
Intel MKL 2.08E-05 1.21E-05 ##### ##### ##### 3.32E-05 3.81E-05
ATLAS 1.27E-05 2.78E-05 ##### ##### ##### 8.27E-05 9.67E-05
BG/P (MFLOPS)
0.01 0.1 1 10 50 100 500
Base (S) 69.2047 221.72 364.54 247.19 254.94 255.88 289.18
Base (P) 15.14 122.53 695.77 834.75 984.1 1003.73 1151.7
ESSL (S)
ESSL (P)
Goto (S)
Goto (P)
Orio (S) 6744.61 1910.63 1902.67 781.26 781.26 781.27 544.49
Orio (P) 6739.19 1910.63 842.6 1285.72 1719.88 1739.13 1547.58
BG/P (seconds)
10 100 1000 10000 50000 100000 500000
Base (S) 1.00E-06 4.00E-06 ##### ##### ##### 3.13E-03 1.38E-02
Base (P) 5.00E-06 7.00E-06 ##### ##### ##### 7.97E-04 3.47E-03
ESSL (S)
ESSL (P)
Goto (S)
Goto (P)
Orio (S) 3E-09 1.04678E-07 ##### ##### ##### 1.02E-03 7.35E-03
Orio (P) 3E-09 1.04678E-07 ##### ##### ##### 4.60E-04 2.59E-03
SPLIT into two parts:
BG/P (seconds)
10 100 1000 10000 50000 100000
Base (S) 1.00E-06 4.00E-06 ##### ##### ##### 3.13E-03
Base (P) 5.00E-06 7.00E-06 ##### ##### ##### 7.97E-04
Orio (S) 3E-09 1.04678E-07 ##### ##### ##### 1.02E-03
Orio (P) 3E-09 1.04678E-07 ##### ##### ##### 4.60E-04
BG/P (seconds)
100000 500000 1000000 5000000
Base (S) 3.13E-03 1.38E-02 ##### #####
Base (P) 7.97E-04 3.47E-03 ##### #####
ESSL (S) ##### #####
ESSL (P) ##### #####
Goto (S&P) ##### #####
Orio (S) 1.02E-03 7.35E-03 ##### #####
Orio (P) 4.60E-04 2.59E-03 ##### #####
Base (S) Base (P) ESSL (S)
ESSL (P) Goto (S&P) Orio (S)
1.60E-01
1.40E-01
1.20E-01
Time (Sec.)
1.00E-01
8.00E-02
6.00E-02
4.00E-02
2.00E-02
0.00E+00
100000 500000 1000000 5000000
Array Size
16000 18000 20000
2.494 3.143 3.886 C from Matlab
1.899 2.338 2.978
1.636 2.063 2.546 Orio (P)
4.345 5.48 6.84 8
2.32 2.93 3.63
3.42 4.29 5.3 7
6
5
Time (Sec.)
4
3
2
1
0
Orio (S) 2000 4000 6000 8000 10000 12000
BLAS
16000 18000 20000 Matrix
0.743 0.947 1.171
0.589 0.734 0.923
0.33 0.401 0.509
1.43 1.82 2.28
1.11 1.39 1.76
1.53 1.65 1.95
16000 18000 20000 C from Matlab
Orio (P)
16000 18000 20000 8
1.209 1.53 1.888
1.215 1.383 1.744
7
0.533 0.678 0.833
6
Time (Sec.)
4.32 5.45 6.78
6
Time (Sec.)
1.11 1.39 1.73
0.0019 0.00238 0.00322
5
4
3
2
1
0
2000 4000 6000 8000 10000 12000
Matrix Size
16000 18000 20000
2.488 3.136 3.869 C from Matlab
0.782 0.948 1.178
0.411 0.582 0.789 Orio (P)
1.43 1.8 2.25 4.5
1.1 1.39 1.73
1.51 1.9 2.35 4
3.5
3
Time (Sec.)
2.5
2
1.5
1
0.5
0
2000 4000 6000 8000 10000 12000
16000 18000 20000
Matrix Size
1.E+04
0.000115 0.000131 0.000146
4.35E-05 4.95E-05 5.50E-05
0.000111 0.000127 0.000142
1.E+03
MFLOP/s
1000 5000 10000
289 289.2 289.18
1150 1154.57 1153.5
473.21 487.49 488.23 Base (S)
831.31 856.25 858.71 1.E+02
408.91 411.55 411.78 ESSL (P)
408.91 411.54 411.77
534.09 537.54 533.59
Goto (P)
1482.57 1678.11 1511.44 Orio (S)
1.E+01
0.01 0.1 1 10 50 100
Array Size (x
1000000 5000000 10000000
Base (S) Base (P)
2.77E-02 1.38E-01 2.77E-01 Goto (S) Goto (P)
6.96E-03 3.46E-02 6.94E-02
1.69E-02 8.21E-02 1.64E-013.00E-01
9.62E-03 4.67E-02 9.32E-02
1.96E-02 9.72E-02 1.94E-012.50E-01
1.96E-02 9.72E-02 1.94E-01
1.50E-02 7.44E-02 1.50E-01
5.40E-03 2.38E-02 5.29E-022.00E-01
Time (Sec.)
1.50E-01
1.00E-01
5.00E-02
0.00E+00
Matrix Size
Matrix Size
ESSL (S)
Orio (S)
5000000
Orio (S)
BLAS
10000 12000 14000 16000 18000 20000
Matrix Size
Orio (S)
BLAS
12000 14000 16000 18000 20000
Matrix Size
Orio (S)
Intel MKL
12000 14000 16000 18000 20000
Matrix Size
Base (S) Base (P)
ESSL (P) Goto (S)
Goto (P) ESSL (S)
Orio (S) Orio (P)
100 500 1000 5000 10000
Size (x 1,000)
ase (P) ESSL (S) ESSL (P)
oto (P) Orio (S) Orio (P)
Matrix Size
Matrix Size