inv() optimization flags

[Giorgos Bontzios]

Hi

I am building a software tool in C++ .

At a specific part I need the inverse of matrix.

For the inversion I use Seldon C++ library which is essentially a LAPACK wrapper.

While experimenting with the results between my code and Octave's

I've noticed that the inv() function of Octave is quite faster than Seldon's which is weird

since, to the best of my knowledge, Octave calls also the same LAPACK functions for inversion.

 

So I would like to ask you what function specifically you call for the inversion and what optimization flags you use to achieve this speedup.

 

In my project I use the Seldon's GetInverse() function

 

http://seldon.sourceforge.net/doc-5.2/functions_lapack.php#getinverse

Implementation

http://seldon.sourceforge.net/doc-5.2/_lapack___linear_equations_8cxx_source.php

 

and the following optimization flags

-m32 -mfpmath=sse -Ofast -flto -march=native -funroll-loops

 

on a Linux Ubuntu 12.04 32bit, Intel i3 machine.

 

Thank you in advance for your time

 

Best regards

Giorgos