[SciPy-user] SciPy with gcc4 and gfortran on OS X

Pearu Peterson pearu at scipy.org
Sat Apr 8 14:40:12 CDT 2006



On Sat, 8 Apr 2006, Pearu Peterson wrote:

>
>
> On Sat, 8 Apr 2006, Robert Kern wrote:
>
>> Pearu Peterson wrote:
>>> On Sat, 8 Apr 2006, Zachary Pincus wrote:
>>>> Finally, I would note that maybe the patched setup.py files should only turn
>>>> off optimization for the mach files if gfortran is being used. (If this is
>>>> something that can easily be done.)
>>>
>>> Nope. Actually mach files should be compiled without optimization for all
>>> compilers. So the patched setup.py files are ok.

Now I realized that mach files should be compiled with optimization on (I 
confused them with LAPACK ?lamch.f files that should be always compiled 
without optimization)

> Btw, now that I have compiled the whole scipy with gfortran, I get
>
> ======================================================================
> ERROR: check_dot (scipy.lib.tests.test_blas.test_fblas1_simple)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File
> "/usr/local/lib/python2.3/site-packages/scipy/lib/blas/tests/test_blas.py",
> line 76, in check_dot
>     assert_almost_equal(f([3j,-4,3-4j],[2,3,1]),-9+2j)
> SystemError: NULL result without error in PyObject_Call

When I compiled blas sources with gfortran and disabled ATLAS (export 
ATLAS=None; export BLAS=None; export BLAS_OPT=None; export 
BLAS_SRC=/path/to/blas/sources/)
then all tests are passed. So it could be also an ATLAS issue. Try 
building scipy against BLAS sources and disable optimized BLAS libraries.

Pearu



More information about the SciPy-user mailing list