It's a problem with the Athlon.  It also happened with a previous version of SciPy this summer.  Look in the archives for:

[SciPy-user] optimize.leastsq crashs my interpreter   Hans Georg Krauthäuser
/Fri Jul 7 05:30:30 CDT 2006/

[SciPy-user] Athlon problems   Ed Schofield
/Sat Jul 29 16:15:58 CDT 2006/

Eventually it was fixed for Python 2.4:

[SciPy-user] Scipy works on AMD now   John Hassler
/Tue Sep 12 09:43:48 CDT 2006/

Now it's back in the new version.  I don't know what the problem was the last time, or who fixed it.  I think it did have something to do with Athlon's version of SSE.
I kept Python 2.4.4 and SciPy 0.5.1, which still works fine.

john

p.s. - sorry if this appears twice.  We've been .. ahem! .. upgraded from Adelphia to Comcast, and my first try bounced.


Jonathan Griffitts wrote:

Hi!

I have installed SciPy 0.5.2 for Python 2.5 on several Windows
computers, using the precompiled binary scipy-0.5.2.win32-py2.5.exe.
This works fine on some, but on others it crashes Python with an Illegal
Instruction exception.  The crash is easy to find by running the test()
suite or by attempting to make any use of scipy.integrate.quad.

Digging into it, I see that the exception comes from _quadpack.pyd, and
it dies at a MOVSD instruction.  I believe MOVSD is an SSE2 instruction
that is only implemented on the more recent CPUs from both Intel and
AMD.

Has anyone out there compiled win32 binaries for Python 2.5 and older
CPUs without SSE2?  If so, could I beg for a copy?  If not, I'll dig
into getting set up to compile SciPy myself.

                        Thanks,