[Numpy-discussion] Problem with NumPy 1.10.4 with ATLAS on Python 2.7.8
Davide Vanzo
vanzod at accre.vanderbilt.edu
Thu Jan 28 17:31:23 EST 2016
Nathaniel,
thanks for your reply.
Here is the output you requested (and a little more).
# ldd /usr/local/python2/2.7.8/x86_64/gcc46/nonet/lib/python2.7/site-
packages/numpy/core/multiarray.so
linux-vdso.so.1 => (0x00007fffce3f2000)
libatlas.so =>
/usr/local/atlas/latest/x86_64/gcc46/nonet/lib/libatlas.so
(0x00007f86253df000)
libm.so.6 => /lib64/libm.so.6 (0x00007f8625147000)
libpython2.7.so.1.0 =>
/usr/local/python2/2.7.8/x86_64/gcc46/nonet/lib/libpython2.7.so.1.0
(0x00007f8624d6c000)
libpthread.so.0 => /lib64/libpthread.so.0
(0x00007f8624b4f000)
libc.so.6 => /lib64/libc.so.6 (0x00007f86247ba000)
/lib64/ld-linux-x86-64.so.2 (0x00007f8626409000)
libdl.so.2 => /lib64/libdl.so.2 (0x00007f86245b6000)
libutil.so.1 => /lib64/libutil.so.1
(0x00007f86243b3000)
# nm -D /usr/local/atlas/latest/x86_64/gcc46/nonet/lib/libatlas.so |
grep cblas
# nm -D /usr/local/atlas/latest/x86_64/gcc46/nonet/lib/liblapack.so |
grep cblas
U cblas_cdotc_sub
U cblas_cgemm
[…]
U cblas_sgemm
U cblas_sgemv
[…]
U cblas_ztrsv
I'm pretty familiar with this error and I always solved it by correctly
pointing the installer to the correct library paths with the
BLAS/LAPACK/ATLAS environment variables. However with 1.10.4 no matter
how I tried to define such variables (even by inserting them in the
site.cfg file), there was no way to make it work.
Davide
On Thu, 2016-01-28 at 14:23 -0800, Nathaniel Smith wrote:
> What does
> ldd
> /usr/local/python2/2.7.8/x86_64/gcc46/New_build/lib/python2.7/site-
> packages/numpy/core/multiarray.so
> say?
> (I'm not a numpy build expert but that should at least give a hint at
> which kind of brokenness you're running into... I'm also somewhat
> curious why you're using such an ancient compiler, but that's
> unlikely to be the issue.)
> On Jan 28, 2016 1:43 PM, "Davide Vanzo" <vanzod at accre.vanderbilt.edu>
> wrote:
> > Hi all,
> > I recently upgraded NumPy from 1.9.1 to 1.10.4 on Python 2.7.8 by
> > using pip. As always I specified the paths to Blas, Lapack and
> > Atlas in the respective environment variables. I used the same
> > compiler I used to compile both Python and the libraries (GCC
> > 4.6.1). The problem is that it always tries to get Blas symbols in
> > the wrong library:
> >
> > >>> import numpy
> > Traceback (most recent call last):
> > File "<stdin>", line 1, in <module>
> > File
> > "/usr/local/python2/2.7.8/x86_64/gcc46/New_build/lib/python2.7/site
> > -packages/numpy/__init__.py", line 180, in
> > from . import add_newdocs
> > File
> > "/usr/local/python2/2.7.8/x86_64/gcc46/New_build/lib/python2.7/site
> > -packages/numpy/add_newdocs.py", line 13, in
> > from numpy.lib import add_newdoc
> > File
> > "/usr/local/python2/2.7.8/x86_64/gcc46/New_build/lib/python2.7/site
> > -packages/numpy/lib/__init__.py", line 8, in
> > from .type_check import *
> > File
> > "/usr/local/python2/2.7.8/x86_64/gcc46/New_build/lib/python2.7/site
> > -packages/numpy/lib/type_check.py", line 11, in
> > import numpy.core.numeric as _nx
> > File
> > "/usr/local/python2/2.7.8/x86_64/gcc46/New_build/lib/python2.7/site
> > -packages/numpy/core/__init__.py", line 14, in
> > from . import multiarray
> > ImportError:
> > /usr/local/python2/2.7.8/x86_64/gcc46/New_build/lib/python2.7/site-
> > packages/numpy/core/multiarray.so: undefined symbol: cblas_sgemm
> >
> > I also tried to install from source instead of pip but no luck
> > either.
> > The only way to get it to work is to downgrade to 1.9.1.
> > Any idea why?
> >
> > Thanks.
> >
> > Davide
> >
> > _______________________________________________
> > NumPy-Discussion mailing list
> > NumPy-Discussion at scipy.org
> > https://mail.scipy.org/mailman/listinfo/numpy-discussion
> >
> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion at scipy.org
> https://mail.scipy.org/mailman/listinfo/numpy-discussion
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/numpy-discussion/attachments/20160128/925ef777/attachment.html>
More information about the NumPy-Discussion
mailing list