[PyCUDA] pycuda installation problem

[Andreas Klöckner]

On Freitag 10 Juli 2009, Gökhan SEVER wrote:
> On Thu, Jul 9, 2009 at 10:04 PM, Andreas Klöckner
>
> <lists at informa.tiker.net>wrote:
> > On Donnerstag 09 Juli 2009, Gökhan SEVER wrote:
> > > I should have bitten my tongue before saying all working fine.
> > >
> > > examples complain this time :(
> > >
> > > [gsever at ccn examples]$ python demo.py
> > > RuntimeError: FATAL: module compiled as big endian, but detected
> >
> > different
> >
> > > endianness at runtime
> > > RuntimeError: FATAL: module compiled as big endian, but detected
> >
> > different
> >
> > > endianness at runtime
> > > Traceback (most recent call last):
> > >   File "demo.py", line 3, in <module>
> > >     import pycuda.driver as cuda
> > >   File
> >
> > "/usr/lib/python2.5/site-packages/pycuda-0.94beta-py2.5-linux-i686.egg/py
> >cu
> >
> > >da/driver.py", line 1, in <module>
> > >     from _driver import *
> > > ImportError: numpy.core.multiarray failed to import
> >
> > Something fishy is going on here. i686 is a little endian architecture,
> > but somehow (I assume) numpy is complaining that it got compiled as big
> > endian. Would you mind trying with a release numpy?
> >
> > Andreas
> >
> > _______________________________________________
> > PyCUDA mailing list
> > PyCUDA at tiker.net
> > http://tiker.net/mailman/listinfo/pycuda_tiker.net
>
> Good morning,
>
> Rolled back to numpy.i386 0:1.2.0-3.fc10 via yum.
>
> However now I have even a prettier error:
>
> [gsever at ccn examples]$ python demo.py
> Segmentation fault
>
> [gsever at ccn examples]$ python hello_gpu.py
> Segmentation fault

Have you recompiled PyCUDA against the new numpy? If yes, send the output of

gdb --args python hello_gpu.py
...stuff happening...
r<Enter>
...more stuff happening...
Crashed with SIGSEGV.
bt<Enter>

Andreas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.tiker.net/pipermail/pycuda/attachments/20090710/995fa412/attachment-0001.pgp>