Wiki
Clone wikiConjugateGradient / P100 vs KNL
n | P100 time [ms] | KNL time [ms] | ratio | P100 speed [elements/ms] | KNL speed [elements/ms] |
---|---|---|---|---|---|
100 | 4.003 | 7.702 | 1.92 | 25.0 | 13.0 |
200 | 4.016 | 6.263 | 1.56 | 49.8 | 31.9 |
400 | 3.982 | 6.942 | 1.74 | 100.5 | 57.6 |
800 | 3.98 | 6.544 | 1.64 | 201.0 | 122.2 |
1600 | 4.006 | 6.513 | 1.63 | 399.4 | 245.7 |
3200 | 3.991 | 6.557 | 1.64 | 801.8 | 488.0 |
6400 | 4.006 | 6.912 | 1.73 | 1597.6 | 925.9 |
12800 | 4.054 | 7.375 | 1.82 | 3157.4 | 1735.6 |
25600 | 5.043 | 8.917 | 1.77 | 5076.3 | 2870.9 |
51200 | 5.662 | 11.432 | 2.02 | 9042.7 | 4478.7 |
102400 | 7.041 | 17.57 | 2.50 | 14543.4 | 5828.1 |
204800 | 11.08 | 26.262 | 2.37 | 18483.8 | 7798.3 |
409600 | 19.556 | 45.488 | 2.33 | 20945.0 | 9004.6 |
819200 | 35.727 | 81.674 | 2.29 | 22929.4 | 10030.1 |
1000000 | 42.077 | 96.312 | 2.29 | 23766.0 | 10382.9 |
1638400 | 66.377 | 154.229 | 2.32 | 24683.2 | 10623.2 |
3276800 | 126.625 | 297.931 | 2.35 | 25878.0 | 10998.5 |
6553600 | 249.148 | 585.783 | 2.35 | 26304.0 | 11187.8 |
10000000 | 376.651 | 866.736 | 2.30 | 26549.8 | 11537.5 |
13107200 | 492.447 | 1147.08 | 2.33 | 26616.5 | 11426.6 |
20000000 | 747.573 | 1722.9 | 2.30 | 26753.2 | 11608.3 |
26214400 | 979.213 | 2270.56 | 2.32 | 26770.9 | 11545.3 |
figure
Updated