[SciPy-user] Building Scipy on Dell 64bit Xeon

David Noriega neruocomp at yahoo.com
Thu Jul 14 17:44:47 CDT 2005

I would like to report the partial working of Scipy on
a 64bit Xeon.  I figured out that it was LAPACK that
was causing problems.  Luckly there is an rpm for my
distro and arch.  I used that to compleate ATLAS,
after recompiling it with -fPIC. 

I then ran the scipy test and got one fail, four
errors.  Something about arrays.  I installed Numeric
with an rpm.  So I uninstalled that one and installed
the latest version from the source.  That got rid of
the errors, but the test still had one failure.

FAIL: check_round
Traceback (most recent call last):
line 1789, in check_round
line 715, in assert_array_equal
    assert cond,\
Arrays are not equal (mismatch 25.0%):
        Array 1: [10 10 11 11]
        Array 2: [10 10 10 11]

Ran 690 tests in 1.340s

This is using the latest ATLAS with the cvs source of
Scipy.  The normal scipy wouldn't even compile.  I get
declaration errors when it compiles ranlib_all.c

error: Command "gcc -pthread -fno-strict-aliasing
-fexceptions -m64 -mtune=nocona -D_GNU_SOURCE -fPIC
-fPIC -I/usr/include/python2.4 -c
Lib/stats/ranlib_all.c -o
failed with exit status 1

I guess this should get to the developers, but I'm not
part of that mailing list.  If someone could pass it
on, that would be great.

Don't fear that philosophy's an impious way
--superstition's more likely to lead folk astray.

~Lucretius, De rerum natura, Book One


Start your day with Yahoo! - make it your home page 

More information about the SciPy-user mailing list