Re: Octave C++ performance benchmarks?

[Jaroslav Hajek]

On Mon, Jul 21, 2008 at 9:58 PM, Kyusik Chung <address@hidden> wrote:
> Hello,
> We are looking to potentially use the Octave C++ libraries in our C++
> application.  It sounded like the C++ libraries were the exact same ones
> that power the Octave end-user interface - could someone confirm this?
> In our C++ application, we are concerned with run time performance, esp of
> the following:
>
> Matrix inversions
> LU Decomps (if thats not the method used for the inverse function
> Special improvements for inversions of symmetric and positive-definite
> matrices
>
> Are there any sorts of benchmarks of the Octave C++ libraries?  How would
> Octave C++ compare with other options out there in terms of run time
> performance?  Is it appropriate to use Octave C++ where high performance is
> a key concern?  If not, do people have other suggestions?
> Any help would be appreciated.
> Thanks!
> Kyusik
>

For dense matrices, the speed of Octave essentially derives from the
underlying LAPACK library. With an optimized BLAS, you'll probably
find nothing faster.
The key factorizations (Cholesky, LU, QR, SVD) are wrapped as classes
in liboctave.
Also the MatrixType class allows discovering special types of matrices
(triangular, SPD).
For sparse matrices, SuiteSparse libraries (UMFPACK, CHOLMOD...)
libraries are employed.

> _______________________________________________
> Help-octave mailing list
> address@hidden
> https://www.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