Hi,

I have a Core i7 920 Desktop with an NVIDIA GTX 780 graphics card, running
Ubuntu 13.04 (Raring) 64-bit. I am unable to install PyCUDA but I have been
able to install the CUDA Toolkit and the SDK examples compile properly.

The NVIDIA GTX 780 GPU works only with nvidia-325 drivers. So I installed
them first, and then installed the CUDA 5.5 Toolkit from the nvidia website
(the .run file). All the dependencies have been installed (linux-headers,
linux-source, etc.)

I have numpy for both python 2.7 as well as python 3.3. I followed the
advice on http://wiki.tiker.net/PyCuda/Installation/Linux/Ubuntu#previewwith
the following commands:

vivek@farpoint:~/pycuda-2013.1.1$ ./configure.py
--python-exe=/usr/bin/python3 --cuda-root=/usr/local/cuda
--cudadrv-lib-dir=/usr/lib --boost-inc-dir=/usr/include
--boost-lib-dir=/usr/lib --boost-python-libname=boost_python-mt-py33
--boost-thread-libname=boost_thread-mt --no-use-shipped-boost

# note that I use py33

vivek@farpoint:~/pycuda-2013.1.1$ sudo make -j 4 # works only with sudo

I get the following error message from this command

/usr/bin/ld: cannot find -lcuda
/usr/bin/ld: skipping incompatible /usr/local/cuda/lib/libcurand.so when
searching for -lcurand
collect2: error: ld returned 1 exit status
error: command 'x86_64-linux-gnu-g++' failed with exit status 1
make: *** [all] Error 1
vivek@farpoint:~/pycuda-2013.1.1$

I referred to
http://askubuntu.com/questions/135494/how-to-resolve-pycuda-installation-error-usr-bin-ld-cannot-find-lcudabut
this did not help me either.

I have a functional PyCUDA installation on another Ubuntu box (a laptop)
but there I used the default python-pycuda package from the repositories,
which ends up installing nvidia-current and nvidia-304, which does not
support my graphics card.

I'd really appreciate some help in fixing this because I'm due to submit my
Master's thesis and the code I have runs too slowly on my laptop, freezing
its resources. I really want to get PyCUDA to work on this desktop.

Thanks in advance!

Cheers,
Vivek.
_______________________________________________
PyCUDA mailing list
[email protected]
http://lists.tiker.net/listinfo/pycuda

Reply via email to