Re: Mac-Intel math performance libraries

[Jaroslav Hajek]

On Mon, Sep 29, 2008 at 8:23 PM, Samir Sharshar <address@hidden> wrote:
> Hi octave users,
>
> I've just bought a fair new MacPro Intel-based processor.
> Searching the internet I've found that the IntelMath Kernel Libraries
> for mac os x could increase by a 30% factor the performance of octave
> (as said by an Ubuntu user).
>
> I'm running under 10.5.5, with XCode 3.1 (aka gcc 4.0)
>
> I'm in trouble in compiling octave with :
>
> 1. UMFPACK : everything goes fine but the .a libraries have no symbols
> (ranlib errors),
> 2. FFTW 3.0 : cannot build shared libraries
>
> Secondly, I imagine I have to pass some compilation arguments to
> compile UMFPACK and FFTW against the Intel libraries but can't figure
> how... sorry I'm not a C/C++/Fortran developper :)
>
> Have anyone test or does the job ?

I'm successfully building Octave using Intel C++/Fortran and Intel MKL
(multithreaded), including FFTW and UFSparse (UMFPACK, CHOLMOD, etc.).
Your information is, however, too vague for me to provide any useful
feedback. Do you use the Intel compilers?

> Help would be greatly appreciated,
>
> Cheers,
>
> Samir Sharshar MD
>
> _______________________________________________
> Help-octave mailing list
> address@hidden
> https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
>



-- 
RNDr. Jaroslav Hajek
computing expert
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz