Thank you, I did not know, or had forgotten, that property of sudo. It seems, however, that even after following your suggestion the behavior and error messages are the same.

//nvcc exists in PATH

mrule <at> computer:~/installers/pycuda$ nvcc

nvcc fatal   : No input files specified; use option --help for more information

//CUDA_ROOT is defined 

mrule <at> computer:~/installers/pycuda$ export CUDA_ROOT='/usr/local/cuda/'

mrule <at> computer:~/installers/pycuda$ sudo -E sh -c 'echo $CUDA_ROOT'

/usr/local/cuda/

//make install can't see either

mrule <at> computer:~/installers/pycuda$ sudo -E sh -c make install

...

*** CUDA_ROOT not set, and nvcc not in path. Giving up.

make: *** [all] Error 1

mrule <at> nathan-tesla1:~/installers/pycuda$ sudo -E sh -c 'make install'

[same thing]

I've got things working well enough to proceed with work today, so, I guess its ok for now. Still, it might be educational to know whats really happening here.

thank you

--mrule

On Fri, Jun 4, 2010 at 12:31 PM, Imran Haque <ihaque-FGKo4X94FMn2fBVCVOL8/A@public.gmane.org> wrote:

Hi Michael,

The problem is that sudo doesn't preserve environment variables. The easiest way to get around this is to pass -E to sudo:

ihaque <at> obsidian:~
$ export CUDA_ROOT=/usr/local/cuda

ihaque <at> obsidian:~
$ echo $CUDA_ROOT
/usr/local/cuda

ihaque <at> obsidian:~
$ sudo sh -c 'echo $CUDA_ROOT'


ihaque <at> obsidian:~
$ sudo -E sh -c 'echo $CUDA_ROOT'
/usr/local/cuda


Cheers,

Imran

On 6/4/2010 7:24 AM, Michael Rule wrote:

Thanks, nvcc is actually on the path, which is why I found the message
confusing. I can type nvcc in bash and it runs and everything. I also
did "CUDA_ROOT='/usr/local/cuda'" in bash such that "echo $CUDA_ROOT"
prints "/usr/local/cuda". Also, usr/local/cuda definitely exists. So, it
would seem that I have taken care of both error conditions, and yet the
message persists.

--mrule

On Thu, Jun 3, 2010 at 10:53 PM, Bogdan Opanchuk <mantihor-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

<mailto:mantihor <at> gmail.com>> wrote:

   Hi Michael,

   The error message is sort of self-explanatory. You need to make 'nvcc'
   (cuda compiler) available to installer. There are two ways to do it:
   either add path to it (usually /usr/local/cuda/bin) to your $PATH
   variable (by modifying bash profile, for example), or pass the path to
   CUDA (usually /usr/local/cuda) as --cuda-path to configure.py of
   PyCuda.

   Best regards,
   Bogdan

   On Fri, Jun 4, 2010 at 8:42 AM, Michael Rule <mrule7404-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

   <mailto:mrule7404-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>> wrote:
    > hello,
    > I recently upgraded CUDA on an Ubuntu machine, and, well,
   everything broke.
    > I thought I had it working, but now scripts that I remember being
   fine are
    > crashing ( actually, most scripts are crashing with 'launch
   failure'). Its
    > likely my own **** fault, but I went ahead and tried to upgrade
   PyCUDA to
    > see if that would fix anything. Well, I went back and tried the
   usual steps
    > and got
    > *** CUDA_ROOT not set, and nvcc not in path. Giving up.
    > when I tryed to do "sudo make install".
    > So, I apologize for the spam but at the moment I'm just too tired
   to be able
    > to figure out what to do myself. Anyone have any links or solved this
    > problem before ?
    > --mrule.
    > _______________________________________________
    > PyCUDA mailing list

    > PyCUDA <at> tiker.net <mailto:PyCUDA-QQNprVDQKGpeoWH0uzbU5w@public.gmane.org>

    > http://lists.tiker.net/listinfo/pycuda
    >
    >




_______________________________________________
PyCUDA mailing list
PyCUDA-QQNprVDQKGpeoWH0uzbU5w@public.gmane.org
http://lists.tiker.net/listinfo/pycuda

_______________________________________________
PyCUDA mailing list
PyCUDA-QQNprVDQKGpeoWH0uzbU5w@public.gmane.org
http://lists.tiker.net/listinfo/pycuda