Re: Performance of subsasgn and subsref

[mavram]

On Wed, Aug 11, 2004 at 09:15:57AM -0500, address@hidden wrote:
> Hi 
> 
> I did very simple experiments in both matlab and octave. 
> In octave:
> octave:1>  m = rand(3000);
> octave:2> tic, a=m(1:2999, 1:2999); toc
> ans = 2.2935
> 
> octave:4> tic,m(2:3000, 2:3000)=a; toc
> ans = 1.8978
> 
> -----------
> In matlab
> >> m=rand(3000);
> >> tic, a=m(1:2999, 1:2999); toc
> 
> elapsed_time =
> 
>     0.5277
> 
> >> tic,m(2:3000, 2:3000)=a; toc
> 
> elapsed_time =
> 
>     0.4800
> 
> 
> I was wondering what causes the difference in performance. 
> I am using octave 2.1.57
> 
> Thanks,
> Jia
> 
> 
I have a similar experience with the file emd.m adapted from G. Rilling,
P. Flandrin and P. Gonחalvטs. The file (either in the original form or
with the slight modifications required by minor incompatibilities
between matlab and octave runs about three times slower on the later.
(I cannot be more precise: I either compare octave on a debian linux
installed on my computer with matlab on windows XP on the same machine,
or with matlab on solaris from a remote server)
As the program spends a lot of time doing cubic spline interpolations, I
assume that the problem is there. But is there a way to check this guess
? (something like a profiler or some way to emulate its function).

Thanks for any pointer, Avraham



-------------------------------------------------------------
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
-------------------------------------------------------------