[boost] Re: uBLAS vs MTL for sparse matrices

From: Paul C. Leopardi (leopardi_at_[hidden])
Date: 2002-10-26 05:59:45

Michael,
I've now tried using column ordering for one crucial matrix multiplication in 
the function basis_element(). The speed has vastly improved, but it is still 
not as fast as MTL. Waiting for the row compressed type is getting more 
attactive.
Best regards

       MTL | using column |  std::map    |   map_array
Size  Time | Time Slowdown| Time Slowdown| Time Slowdown
  32    10 |    30   3    |    40   4    |    30   3  
  64    20 |    50   2.5  |    70   3.5  |    70   3.5
 128    80 |   390   4.9  |   640   8    |   830  10.4
 256   160 |   890   5.6  |  1440   9    |  1940  12.1
 512   700 |  7180  10.3  | 20380  29.1  | 31770  45.4