Re: Congratulations!

[David Bateman]

Richard Hindmarsh wrote:

> I recently got interested in Octave after seeing that sparse matrices 
> were available, and compiled a recent CVS. I tried running a 
> Matlab/mex fluid dynamcis script of probably 2000 lines, and got it to 
> run with little trouble (I had to find the functions "inputname" and 
> "num2cell"). For a particuilar test case on a particular computer, 
> Matlab runs it in 10s. Octave runs it in 15s, but with a dual 
> processing BLAS library I can get the time down to 10s again. his was 
> without the benefit of being able to optimise with a profiler on 
> Octave. The answers are the same (apart from last two figures of 
> double precision numbers). T
>
> I'm hugely impressed, and having Octave as a potential multi-threaded 
> platform on many architectures is a real bonus.
>
> Thanks to all concerned
> Richard Hindmarsh


Richard,

If you are using sparse matrices in the benchmark above I'm curious if 
you used the "\" operator? If so that is probably what took most of the 
time and also I don't see why octave should be slower than matlab on any 
platform for this operator. So then my next question would be what type 
of matrix are you solving? Is it banded? Is it real or complex? Is it 
symmetric/hermitian? Then finally the question would be how was UMFPACK 
installed if the matrix wasn't banded, etc? Is UMFPACK linked against 
atlas or the like?

Regards
David



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