Numarray, numeric, NumPy, scpy_core ??!!

[Robert Kern]

Re: Numarray, numeric, NumPy, scpy_core ??!!

Sébastien Boisgérault wrote:
[color=blue]
> By the way, I tried numpy 0.9.4 10 minutes ago and guess
> what ? 'eigenvalue' is broken too ... (hangs forever)[/color]

On what platform? Are you linking against an optimized BLAS? We can't fix
anything without details. I'll be happy to work with you on this bug over on the
numpy-discussion list.

--
Robert Kern
robert.kern@gma il.com

"In the fields of hell where the grass grows high
Are the graves of dreams allowed to die."
-- Richard Harter

[Szabolcs Nagy]

Re: Numarray, numeric, NumPy, scpy_core ??!!

[color=blue]
> Basically all I need is vectors and 3x3 matrices.[/color]

hmm
is numpy really efficient for 3x3 (or 4x4) matrices and vectors?

IMHO an optimized matrix4x4 class can be much faster (i'm just guessing
here)

eg cgtypes is a simple c++ implementation with boost-python wrapper:

http://cgkit.sourceforge.net/

[Tim Hochberg]

Re: Numarray, numeric, NumPy, scpy_core ??!!

Szabolcs Nagy wrote:[color=blue][color=green]
>>Basically all I need is vectors and 3x3 matrices.[/color]
>
>
> hmm
> is numpy really efficient for 3x3 (or 4x4) matrices and vectors?
>
> IMHO an optimized matrix4x4 class can be much faster (i'm just guessing
> here)
>
> eg cgtypes is a simple c++ implementation with boost-python wrapper:
> http://cgkit.sourceforge.net/
>[/color]

If you're operating on a single 3x3 or 4x4 matrix, nothing you do in
Python will be "efficient" , by which I assume you mean not all that far
from C speed; the interpreter overhead is simply too high. If you can
operate on large blocks of small arrays and vectors at the same time,
which one can usually do after consuming sufficient coffee, then yes
numpy can be fast if you're careful.

-tim

[greg.landrum@gmail.com]

Re: Numarray, numeric, NumPy, scpy_core ??!!


Robert Kern wrote:[color=blue]
> Sébastien Boisgérault wrote:
>[color=green]
> > By the way, I tried numpy 0.9.4 10 minutes ago and guess
> > what ? 'eigenvalue' is broken too ... (hangs forever)[/color]
>
> On what platform? Are you linking against an optimized BLAS? We can't fix
> anything without details. I'll be happy to work with you on this bug overon the
> numpy-discussion list.[/color]

This is a guess, but the original poster is probably using one of the
newer (3.4.x or 4.0.x) versions of gcc. This is a known problem:

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=138791

(that's one of the applicable bug reports).

A workaround to this problem is to add the option '-ffloat-store' to
your CFLAGS.

<insert "gcc 4.0 is such a disaster" rant here>

-greg

[Sébastien Boisgérault]

Re: Numarray, numeric, NumPy, scpy_core ??!!

Robert Kern wrote:[color=blue]
> Sébastien Boisgérault wrote:
>[color=green]
> > By the way, I tried numpy 0.9.4 10 minutes ago and guess
> > what ? 'eigenvalue' is broken too ... (hangs forever)[/color]
>
> On what platform?[/color]
Linux, Mandriva 2006 (gcc 4.0.1, etc.)[color=blue]
> Are you linking against an optimized BLAS?[/color]
Nope -- I tried the basic install.
[color=blue]
> We can't fix
> anything without details.[/color]
Obviously, sorry. I was not really expecting a fix,
just wanted to point out that Numpy (or Numeric)
may well not work "out of the box" today on some
platforms.

The (similar) problem has been reported (2005-07-11)
is not assigned to anybody and is still open
See: http://sourceforge.net/tracker/?grou...69&atid=101369
[color=blue]
> I'll be happy to work with you on this bug over on the
> numpy-discussion list.[/color]

Great Robert, I will follow your advice then. Thanks
for the proposal.[color=blue]
> --
> Robert Kern
> robert.kern@gma il.com
>
> "In the fields of hell where the grass grows high
> Are the graves of dreams allowed to die."
> -- Richard Harter.[/color]

[Robert Kern]

Re: Numarray, numeric, NumPy, scpy_core ??!!

Sébastien Boisgérault wrote:[color=blue]
> Robert Kern wrote:
>[color=green]
>>Sébastien Boisgérault wrote:
>>[color=darkred]
>>>By the way, I tried numpy 0.9.4 10 minutes ago and guess
>>>what ? 'eigenvalue' is broken too ... (hangs forever)[/color]
>>
>>On what platform?[/color]
>
> Linux, Mandriva 2006 (gcc 4.0.1, etc.)[/color]

Okay, my answer then is, "Don't use gcc 4." gcc 4 broke enough other projects
that I don't feel too bad about saying that. In the meantime, can you try the
workaround that Greg Landrum suggested?

Also, we are using Trac now to manage numpy, so please submit tickets here:

http://projects.scipy.org/scipy/numpy/

Thanks!

--
Robert Kern
robert.kern@gma il.com

"In the fields of hell where the grass grows high
Are the graves of dreams allowed to die."
-- Richard Harter

[Sébastien Boisgérault]

Re: Numarray, numeric, NumPy, scpy_core ??!!


greg.landrum@gm ail.com wrote:[color=blue]
> Robert Kern wrote:[color=green]
> > Sébastien Boisgérault wrote:
> >[color=darkred]
> > > By the way, I tried numpy 0.9.4 10 minutes ago and guess
> > > what ? 'eigenvalue' is broken too ... (hangs forever)[/color]
> >
> > On what platform? Are you linking against an optimized BLAS? We can't fix
> > anything without details. I'll be happy to work with you on this bug over on the
> > numpy-discussion list.[/color]
>
> This is a guess, but the original poster is probably using one of the
> newer (3.4.x or 4.0.x) versions of gcc. This is a known problem:
> https://bugzilla.redhat.com/bugzilla....cgi?id=138791
> (that's one of the applicable bug reports).
>
> A workaround to this problem is to add the option '-ffloat-store' to
> your CFLAGS.[/color]

Great ! It works :)
Thank you greg !

SB
[color=blue]
> <insert "gcc 4.0 is such a disaster" rant here>
>
> -greg[/color]

[Sébastien Boisgérault]

Re: Numarray, numeric, NumPy, scpy_core ??!!


Robert Kern wrote:[color=blue]
> Sébastien Boisgérault wrote:[color=green]
> > Robert Kern wrote:
> >[color=darkred]
> >>Sébastien Boisgérault wrote:
> >>
> >>>By the way, I tried numpy 0.9.4 10 minutes ago and guess
> >>>what ? 'eigenvalue' is broken too ... (hangs forever)
> >>
> >>On what platform?[/color]
> >
> > Linux, Mandriva 2006 (gcc 4.0.1, etc.)[/color]
>
> Okay, my answer then is, "Don't use gcc 4." gcc 4 broke enough other projects
> that I don't feel too bad about saying that.[/color]

You certainly should not feel bad about that. Matlab does not support
anything
more recent than gcc 3.3 -- and specifically Simulink S-Functions do
not work
when compiled with gcc 3.4.x or 4.0.x.
[color=blue]
> In the meantime, can you try the
> workaround that Greg Landrum suggested?[/color]

I tried it, it works :)
[color=blue]
> Also, we are using Trac now to manage numpy, so please submit tickets here:[/color]

ok.
[color=blue]
> http://projects.scipy.org/scipy/numpy/
>
> Thanks!
>
> --
> Robert Kern
> robert.kern@gma il.com
>
> "In the fields of hell where the grass grows high
> Are the graves of dreams allowed to die."
> -- Richard Harter[/color]