[PyCUDA] nvcc errors

Tomasz Rybak bogomips at post.pl
Sun Oct 3 10:31:55 PDT 2010


In the future - please be more specific with which programs are you
trying to run. Saying "a simple example like on
http://documen.tician.de/pycuda/"  is not helpful - there
are many examples on that page.

Your problems with missing nvcc are probably caused by installing
CUDA toolkit in /usr/local/cuda
This means that neither nvcc, nor libraries, nor header files,
are in directories that are searched by compiler.
You solved problem with nvcc by linking, but still your compiler
cannot find libraries.

Add this to your  ~/.bashrc
PATH=$PATH:/usr/local/cuda/bin

Add those lines to your /etc/ld.so.conf
/usr/local/cuda/lib64
/usr/local/cuda/lib
and then run ldconfig (as root).
If you have 32-bit system, add only second line.

To have all necessary header files, install package
nvidia-current-dev.
In Debian NVIDIA header files are in separate packages,
in Ubuntu all files necessary to compile OpenCL, OpenGL, CUDA,
are in one package.

Hope it helps.

-- 
Tomasz Rybak <bogomips at post.pl> GPG/PGP key ID: 2AD5 9860
Fingerprint A481 824E 7DD3 9C0E C40A  488E C654 FB33 2AD5 9860
http://member.acm.org/~tomaszrybak
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://lists.tiker.net/pipermail/pycuda/attachments/20101003/205537a4/attachment.pgp>


More information about the PyCUDA mailing list