Re: unexpected behavior when multiplying large matrices

[Mike Miller]

On Sun, 1 Dec 2002, John W. Eaton wrote:

>   octave2.1:9> x = eye(1000);
>   octave2.1:10> y = [1:1000]' * [1:1000];
>   octave2.1:11> z = [1:1000]';
>   octave2.1:12> tic; a = 2 * (x - ((x * z * z' * x)/((1/2) + z' * x * z))); 
> toc
>   ans = 2.1231
>   octave2.1:13> tic; b = 2 * (y - ((y * z * z' * y)/((1/2) + z' * y * z))); 
> toc
>   ans = 2.0958
>
> The above results were obtained on a 1.4GHz Athlon system running Debian
> and Octave is linked with ATLAS:


I'm running on a Sun Ultra 2 2200 under Solaris and I get results like
those of Joshua, but worse! (see my results below).  My octave is not
linked to any BLAS, so maybe that has something to do with it.

octave:1> x = eye(1000);
octave:2> y = [1:1000]' * [1:1000];
octave:3> z = [1:1000]';
octave:4> tic; a = 2 * (x - ((x * z * z' * x)/((1/2) + z' * x * z))); toc
ans = 2.9595
octave:5> tic; b = 2 * (y - ((y * z * z' * y)/((1/2) + z' * y * z))); toc
ans = 464.83


Mike



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------