[SciPy-dev] Progress with linalg2

[Pearu Peterson]

On Sun, 3 Mar 2002, Travis Oliphant wrote:

> >       Solving system of linear equations
> >       ==================================
> >
> >       |    contiguous     |   non-contiguous
> >
> > ----------------------------------------------
> >  size |  scipy  | Numeric |  scipy  | Numeric
> >    20 |   1.17  |   3.87  |   1.18  |   4.07     (secs for 2000 calls)
> >   100 |   1.60  |   3.11  |   1.61  |   3.99     (secs for 300 calls)
> >   500 |   1.64  |   2.12  |   1.64  |   2.38     (secs for 4 calls)
> >  1000 |   5.64  |   6.65  |   5.62  |   7.19     (secs for 2 calls)
> > .
> >            Finding matrix inverse
> >       ==================================
> >
> >       |    contiguous     |   non-contiguous
> >
> > ----------------------------------------------
> >  size |  scipy  | Numeric |  scipy  | Numeric
> >    20 |   1.50  |   5.94  |   1.49  |   6.15     (secs for 2000 calls)
> >   100 |   4.37  |   9.23  |   4.42  |  10.14     (secs for 300 calls)
> >   500 |   4.78  |   7.36  |   4.77  |   7.60     (secs for 4 calls)
> >  1000 |  17.55  |  24.59  |  17.53  |  24.49     (secs for 2 calls)
> > .
> 
> Now, just to be clear these are only testing interfaces, correct?  ATLAS is 
> being used in both scipy and Numeric for these tests, right.

Correct. Right.

>  It would be interesting to see what how a non-atlased Numeric
> compared.

Here are the results for downgraded Numeric that uses the C lite versions
of LAPACK from Numeric/Src:

      Solving system of linear equations
      ==================================
      |    contiguous     |   non-contiguous 
----------------------------------------------
 size |  scipy  | Numeric |  scipy  | Numeric
   20 |   1.36  |   1.66  |   1.16  |   1.89     (secs for 2000 calls)
  100 |   1.66  |   5.64  |   1.65  |   6.86     (secs for 300 calls)
  500 |   1.92  |  10.02  |   1.67  |  10.10     (secs for 4 calls)
 1000 |   5.69  |  49.10  |   5.68  |  49.89     (secs for 2 calls)
.
           Finding matrix inverse
      ==================================
      |    contiguous     |   non-contiguous 
----------------------------------------------
 size |  scipy  | Numeric |  scipy  | Numeric
   20 |   1.44  |   3.22  |   1.43  |   3.48     (secs for 2000 calls)
  100 |   4.25  |  17.63  |   4.17  |  18.43     (secs for 300 calls)
  500 |   4.75  |  36.82  |   4.75  |  37.08     (secs for 4 calls)
 1000 |  17.55  | 156.70  |  17.48  | 156.88     (secs for 2 calls)
.
----------------------------------------------------------------------
Ran 13 tests in 644.039s

Pearu