I am working on Debian. Although Ubuntu is Debian-based, there are
differences in NVIDIA drivers, so not everything will apply cleanly,
but I hope I will be able to point you in right direction.

Dnia 2012-04-19, czw o godzinie 08:37 -0700, maxrider11 pisze:
> pycuda 2011.2.2, ubuntu 11.10, gcc 4.4. python 2.7. I am tired because every
> next step in it bring some error, and I am trying it for 2-3 days.

Which CUDA version have you installed and into which directory?

Can you give us PyCUDA build configuration, or how are you calling
configure.py?
E.g. I am calling it following way on Debian:
./configure.py --boost-python-libname=boost_python-py27 \
 --boost-thread-libname=boost_thread \
 --cuda-root=/usr --cuda-enable-gl

In your case --cuda-root will probably be different (Debian
has packages with CUDA, Ubuntu does not).

> 
>  i am getting this error on running "..pycuda-2011.2.2$ make -j 4"
> /usr/bin/ld: cannot find -lboost_python-mt
> 

Why are you trying to use boost_python-mt? Try boost_python-py27.
In Debian and Ubuntu one can have more than one Python version
installed, so you need to point PyCUDA to the proper library file
using --boost-python-libname in ./configure.py


> /usr/bin/ld: cannot find -lcuda

It cannot find cuda. Give ./configure.py directory in which you have
CUDA installed through --cuda-root

> 
> /usr/bin/ld: skipping incompatible /usr/local/cuda/lib/libcurand.so when
> searching for -lcurand

This is strange. You might need to check compiler versions - but let's
leave it until you solve previous problems.

> 
>  I tried this but its not helping:
> $ sudo ln -s /usr/lib/libboost_python-mt-py26 /usr/lib/libboost_python-mt 

Library files also need to have .so extension. Instead of symlinking
library give its proper name to ./configure.py using
--boost-python-libname

> 
>  Additional information: 
> I don't see ld directory in /usr/bin. This file "libboost_python-mt-py26" is
> present in usr/lib. Also, I installed pycuda even after the error above,
> then I got error in ">>> import pycuda.driver as cuda" : 
>  from pycuda._driver import *
> ImportError: No module named _driver

This is result of previous problem. PyCUDA was not able to build
binary wrapper, so Python code is lost. This will be fixed when
you solve previous problems.

I hope it helps.
Write to list if you need more help.

Regards.

-- 
Tomasz Rybak  GPG/PGP key ID: 2AD5 9860
Fingerprint A481 824E 7DD3 9C0E C40A  488E C654 FB33 2AD5 9860
http://member.acm.org/~tomaszrybak

Attachment: signature.asc
Description: This is a digitally signed message part

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

Reply via email to