Matrix Inversion

Presented is the real-time for inverting a h-matrix. The speedup is calculated wrt to the E6000.

An overview of the results is given at the and of this page.

Sun E6000

level real-time speedup accuracy memory
15 163.6   1.0 3.3e-7   314  
16 392.1   1.0 2.0e-6   666  
17 921.3   1.0 1.6e-5   1428  
18 2173.3   1.0 1.2e-4   3074  
19 5095.2   1.0 9.0e-4   6495  
20 11853.9   1.0 4.1e-3   13884  
21 27394.4   1.0 -   29700  

Sun SunFire 6800

level real-time speedup accuracy memory
15 53.3   3.07 2.3e-7   314  
16 129.0   3.04 2.3e-6   667  
17 300.0   3.07 2.7e-5   1435  
18 703.6.0   3.09 2.3e-4   3074  
19 1618.1   3.15 6.7e-4   6512  
20 3734.6   3.17 3.3e-3   13841  
21 8664.7   3.16 7.9e-2   29671  

Sun SunFire 15k

level real-time speedup accuracy memory
15 56.7   2.89 2.1e-7   314  
16 134.0   2.93 1.5e-6   667  
17 314.2   2.93 1.5e-5   1438  
18 741.3   2.93 6.5e-5   3062  
19 1721.3   2.96 2.6e-4   6523  
20 3996.8   2.97 3.2e-3   14002  
21 10280.3   2.64 1.7e-2   29602  

HP 9000 Superdome

level real-time speedup accuracy memory
15 44.6   3.67 1.4e-7   315  
16 108.0   3.63 5.8e-7   672  
17 260.9   3.53 3.5e-6   1435  
18 633.8   3.43 2.7e-5   3079  
19 1562.0   3.26 4.6e-4   6578  
20 3714.2   3.19 3.6e-3   14007  
21 8834.4   3.10 2.5e-2   29701  

IBM eServer p690

level real-time speedup accuracy memory
15 33.0   4.96 3.7e-7   314  
16 78.1   5.02 1.4e-6   664  
17 186.6   4.94 5.2e-5   1437  
18 429.9   5.06 4.2e-5   3067  
19 985.6   5.17 4.3e-4   6537  
20 2364.1   5.01 2.4e-3   14014  
21 5443.4   5.03 1.2e-2   29678  

AMD Athlon MP 2000

level real-time speedup memory
15 35.4   4.62 309  
16 90.5   4.33 652  
17 231.7   3.98 1397  

Dell Xeon 2400

level real-time speedup memory
15 27.5   5.95 309  
16 67.6   5.80 652  
17 163.2   5.65 1397  

Overview

In the following table the speedup of the individual systems is presented in a more compact form. The best on each level is printed in red colour.

level SF6800 SF15k HP9000 P690 Athlon
15 3.07   2.89   3.67   4.96   4.62  
16 3.04   2.93   3.63   5.02   4.33  
17 3.07   2.93   3.53   4.94   3.98  
18 3.09   2.93   3.43   5.06   -  
19 3.15   2.96   3.26   5.17   -  
20 3.17   2.97   3.19   5.01   -  
21 3.16   2.64   3.10   5.03   -