I hope you had a pleasant summer. I apologize for not keeping touch since
our last meeting. I have been at the Oak Ridge National Lab in Tennessee on
a summer Internship. I am now back in NYC. Unfortunately, my computer
crashed over the summer and I had to reformat my hard drive. That said, I
need to re-install hedge. I am going to follow the installation
instructions on the HEDGE website. I have been getting familiar with using
a UNIX shell so I hope the installation will NOT be like pulling teeth - as
you witnessed last time.
Speak to you soon.
On Thu, Apr 14, 2011 at 3:35 PM, Andreas Kloeckner
> On Thu, 14 Apr 2011 15:03:02 -0400, Saumil Patel <
> saumil.patel134(a)gmail.com> wrote:
> > Hi Andreas,
> > I am following the instructions to acquire HEDGE and I am having trouble
> > understanding the lines that follow "Then, proceed to install hedge
> > In line 1., it seems like I should be updating something via "sudo
> > aptitude...." but I am not quite sure what that should be.
> > When I type "apt-cache search libboost-python.*dev | sort " I get the
> > following:
> > libboost-python1.40-dev - Boost.Python Library development files
> > libboost-python1.42-dev - Boost.Python Library development files
> > libboost-python-dev - Boost.Python Library development files (default
> > version)
> Simply type this command:
> BOOST_VER=1.42; sudo aptitude install build-essential python-dev
> libboost-serialization$BOOST_VER-dev libboost-python$BOOST_VER-dev
> libboost-thread$BOOST_VER-dev libboost-math$BOOST_VER-dev python-numpy
> libopenmpi-dev openmpi-bin git-core
> (all on one line)
> Andreas Kloeckner
> Room 1311 (Warren Weaver Hall), Courant Institute, NYU
I've installed hedge with pycuda. I got the Install OK message after ./repotool start-over. I can run examples on CPU, but when I try to run on CUDA with "-f cuda". I get the follow error:
/src/hedge/examples/advection$ python advection.py -f cuda
Traceback (most recent call last):
File "advection.py", line 164, in <module>
File "advection.py", line 77, in main
discr = rcon.make_discretization(mesh_data, order=4)
File "/home/ktubbs/HEDGE/src/hedge/hedge/backends/__init__.py", line 125, in make_discretization
return self.discr_class(mesh_data, *args, **kwargs)
File "/home/ktubbs/HEDGE/src/hedge/hedge/backends/__init__.py", line 144, in discr_class
from hedge.backends.cuda import Discretization
File "/home/ktubbs/HEDGE/src/hedge/hedge/backends/cuda/__init__.py", line 30, in <module>
import pycuda.gpuarray as gpuarray
File "/home/ktubbs/HEDGE/src/pycuda/pycuda/gpuarray.py", line 6, in <module>
from pycuda.compyte.array import (
ImportError: No module named compyte.array
On Wed, 20 Jul 2011 17:46:23 +0200, Peter17 <peter017(a)gmail.com> wrote:
> 2011/7/11 Andreas Kloeckner <lists(a)informa.tiker.net>:
> > This is rather odd--your valgrind log looks like indeed the memory is being
> > lost from somewhere where numpy is being called directly from Python. Where
> > did you get you numpy from? Did you build it yourself? Can you try and isolate
> > the source of the leak by bisecting the code? Try a smaller operator? (e.g.
> > advection?) No operator at all? No visualization?
> I tried with different versions of numpy, from Ubuntu's packages and
> installed via easy_install. Then, I totally uninstalled Atlas, Lapack,
> Blas and numpy, recompiled Lapack and Blas from the latest sources and
> installed a fresh version of numpy from easy_install but I still had
> the same problem...
> After several days of analysis, I finally abandoned: I installed
> Ubuntu 10.10, then Hedge and its dependencies and now, it is working
> perfectly. My conclusion is that it is due to a specific version of
> one (or more) of the packages of Ubuntu 11.04 but unfortunately, I
> could not find which one. I had almost nothing else than the default
> packages installed on my system.
Paul and I tracked down a leak in PyCUDA. If you were using GPU support,
then that might've been the source of the leak.