[Numpy-discussion] Ashigabou Repository atlas vs yum blas/lapack

David Cournapeau david at ar.media.kyoto-u.ac.jp
Mon Jul 28 22:39:35 EDT 2008


jb wrote:
> I also tried removing the yum blas/lapack libs and installing atlas
> via the instructions given on the scipy site for Ashigabou Repository.
>   

Did you install blas/lapack from ashigabou repository as well ? When I
developed those packages, FC packages for blas/lapack were unusable. But
maybe with recent versions it is ok. I don't have very much time to
spend on this unfortunately (I called for people using fedora to take
care of this and pushing it to official FC repositories as I am not
using FC myself, but nothing happened).

>  Atlas built fine from source and after installing the new rpm there
> were two files in the lib64/atlas/sse2 folder: libblas.so.3.0 and
> liblapack.so.3.0.  However, when I try to install numpy, it cannot
> find any blas, lapack, or atlas, even though my site.cfg file has:
>
> [DEFAULT]
> library_dirs = /usr/lib64:/usr/lib64/atlas/sse2
>
> [blas_opt]
> libraries = f77blas, cblas, atlas
>
> [lapack_opt]
> libraries = lapack, f77blas, cblas, atlas
>
> [atlas]
> library_dirs = /usr/lib64/atlas/sse2
> atlas_libs = lapack, blas
>
>   

Could you paste the configuration log (when it says whether it finds the
packages or not). I believe that you put too much information, the
following site.cfg should work:

[DEFAULT]
library_dirs = /usr/lib64/atlas/sse2

Should work, but I can't be sure without seeing the log (Installing in
sse2 is strange BTW; a quadcore certainly means you have more than sse2.
That's something else to fix).
> Using LD_LIBRARY_PATH=/usr/lib64/atlas/sse2 before installing numpy
> does not make a difference.
>   

LD_LIBRARY_PATH does not change how numpy looks for libraries. It only
changes how the OS looks for libraries when you launch programs, so this
is expected.

> My questions are, are the yum versions of lapack/blas just as good as
> the one built from Ashigabou source, and if not, why would numpy not
> be able to find the Ashigabou blas and lapack files (even though its
> looking in the right directory)?
>   

In the old times (FC 5), the yum ones did not work. If they do now, I
would say just use them. ATLAS is faster than blas/lapack, but it is
more work, and it is useful mainly for large problems anyway.

cheers,

David



More information about the NumPy-Discussion mailing list