Re: Debian packaging of PyCUDA

Tomasz Rybak <bogomips@...> writes:
> With Ubuntu there can be problem with missing package
> libcuda1 - but I would like for someone to try it
> nonetheless (just download *.deb files and dpkg -i *.deb)
> and report whether packages are working.

How is general progress?

Also, what can be done to further ubuntu compatibility?
You mentioned libcuda1 what is the matter with it?

Re: Debian packaging of PyCUDA

Dnia 2010-06-01, wto o godzinie 12:33 +0000, karyo pisze:
> Tomasz Rybak <bogomips@...> writes:
> > With Ubuntu there can be problem with missing package
> > libcuda1 - but I would like for someone to try it
> > nonetheless (just download *.deb files and dpkg -i *.deb)
> > and report whether packages are working.
> 
> How is general progress?

Currently I am working with DD on python-pytools to allow
it to be included in Debian. python-pycuda depends on python-pytools
and (yet unpackaged) NVIDIA CUDA toolkit, so pytools need to be
in Debian before PyCUDA.

Latest versions of packages can be always found (at least until
they are in Debian) on my web page:
http://bogomips.w.tkb.pl/cuda.html

> 
> Also, what can be done to further ubuntu compatibility?
> You mentioned libcuda1 what is the matter with it?
> 

Ubuntu does not contain CUDA; Debian, together with NVIDIA
drivers provides libcuda1 (and libopencl1) which use
driver to execute code on GPU. As libcuda needs to be
present for PyCUDA to work, there is currently very little
chance for PyCUDA to be included in Ubuntu.

I am not sure what would happen if one would install

Re: Debian packaging of PyCUDA

Tomasz Rybak <bogomips@...> writes
> Ubuntu does not contain CUDA; Debian, together with NVIDIA
> drivers provides libcuda1 (and libopencl1) which use
> driver to execute code on GPU. As libcuda needs to be
> present for PyCUDA to work, there is currently very little
> chance for PyCUDA to be included in Ubuntu.

Keep up the good work. I believe that once every necessary package (at least for 
debian) is included, it would be a matter of time until ubuntu supports it.

I realize that you have currently have working packages, but could you make a 
formal announcement here again when all of the dependencies are included in 
official debian repositories?

Re: PyCuda Launch Timeout under Py2Exe

I was unable to solve the launch timeouts by removing elements of the kernel
code.

This seemed to me to point to a problem with CUDA, after reinstalling
Windows (XP Pro SP3), CUDA, etc., the program executes correctly when
packaged. I am now unable to reproduce the error.

As such it would seem that when Windows/CUDA is playing ball py2exe seems
able to handle pycuda 'out of the box'. The distutils script must explicitly
include the pyCuda header files, and exclude nvcudal.dll.

Many thanks for your time.

Regards,

Sam.

-----Original Message-----
From: Andreas Kloeckner [mailto:lists@...] 
Sent: 21 May 2010 13:56
To: Samuel Powell; pycuda@...
Subject: RE: [PyCUDA] PyCuda Launch Timeout under Py2Exe

On Fri, 21 May 2010 13:24:51 +0100, "Samuel Powell"
<spowell@...> wrote:
> Hi Andreas,
> 
> You are correct in your assertion.
> 
> I followed your advice by packaging some of the pyCuda example scripts 

question from a lazy person : *** CUDA_ROOT not set, and nvcc not in path. Giving up.

Re: question from a lazy person : *** CUDA_ROOT not set, and nvcc not in path. Giving up.

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

Re: question from a lazy person : *** CUDA_ROOT not set, and nvcc not in path. Giving up.

Re: question from a lazy person : *** CUDA_ROOT not set, and nvcc not in path. Giving up.

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@...
> <mailto:mantihor@...>> 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@...
>     <mailto:mrule7404@...>> 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@... <mailto:PyCUDA@...>
>      > http://lists.tiker.net/listinfo/pycuda
>      >
>      >
>
>
>
>
> _______________________________________________
> PyCUDA mailing list
> PyCUDA@...
> http://lists.tiker.net/listinfo/pycuda

Re: question from a lazy person : *** CUDA_ROOT not set, and nvcc not in path. Giving up.

Re: question from a lazy person : *** CUDA_ROOT not set, and nvcc not in path. Giving up.