[gpaw-users] GPAW parallel installation on El Capitan
Varadharajan Srinivasan
varadharajan.srinivasan at gmail.com
Tue Apr 19 21:36:55 CEST 2016
Dear Ask,
I did try both with 2 and 4 cores and indeed I get segmentation faults. I
attach the log file for the tests run with 4 cores.
So far in my trial runs (on the tutorials) I haven't faced any problem but
your insight into this issue would be very helpful.
Best,
Vardha.
On Mon, Apr 18, 2016 at 2:13 PM, Ask Hjorth Larsen <asklarsen at gmail.com>
wrote:
> The test suite is designed for 1, 2, 4, or 8 cores. Definitely there
> will be some tests that do not run on 12. However the segmentation
> fault should never happen. Can you reproduce that with 2, 4 or 8
> cores?
>
> Best regards
> Ask
>
> 2016-04-09 14:42 GMT+02:00 Varadharajan Srinivasan
> <varadharajan.srinivasan at gmail.com>:
> > Dear Jakob,
> >
> > I installed ASE and GPAW without using virtualenvs. gnaw-python $(which
> paw)
> > info gives:
> > python-2.7.11 /usr/local/bin/gpaw-python
> > gpaw-1.0.1b1
> >
> /usr/local/opt/python/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/gpaw/
> > ase-3.11.0b1 /Users/vardha/SWARE/ase/ase/
> > numpy-1.11.0
> >
> /usr/local/opt/python/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/
> > scipy-0.17.0
> >
> /usr/local/opt/python/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/scipy/
> > _gpaw built-in
> > parallel /usr/local/bin/gpaw-python
> > FFTW no
> > scalapack yes
> > libvdwxc no
> > PAW-datasets /usr/local/share/gpaw-setups:
> > /usr/share/gpaw-setups
> >
> >
> > Serial tests run fine. However, on running parallel tests several tests
> fail
> > complaining of errors like :
> > (a) ValueError: Cannot construct interpolator. Grid 12 x 12 x 12 may be
> too
> > small
> > (b) RuntimeError: Cannot distribute 80 bands to 12 processors
> > (c) RuntimeError: use fewer bands or more basis functions
> >
> > I think most of these errors are because I got greedy and used 12 cores
> > which might be too much for the tests. But one error was from
> xc/atomize.py
> > which was a segmentation fault :
> > xc/atomize.py [helios:47447] *** Process
> received
> > signal ***
> > [helios:47447] Signal: Segmentation fault: 11 (11)
> > [helios:47447] Signal code: Address not mapped (1)
> > [helios:47447] Failing at address: 0x1230ea000
> > [helios:47447] [ 0] 0 libsystem_platform.dylib
> > 0x00007fff9371352a _sigtramp + 26
> > [helios:47447] [ 1] 0 ???
> > 0x00000000000003e8 0x0 + 1000
> > [helios:47447] [ 2] 0 gpaw-python
> > 0x000000010b8ed035 calc_mgga + 581
> > [helios:47447] [ 3] 0 gpaw-python
> > 0x000000010b8ec613 XCFunctional_calculate + 435
> > [helios:47447] [ 4] 0 Python
> > 0x000000011051dbcd PyEval_EvalFrameEx + 26858
> > [helios:47447] [ 5] 0 Python
> > 0x00000001105170f1 PyEval_EvalCodeEx + 1583
> > [helios:47447] [ 6] 0 Python
> > 0x000000011052169a fast_function + 117
> > [helios:47447] [ 7] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [ 8] 0 Python
> > 0x000000011052172d fast_function + 264
> > [helios:47447] [ 9] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [10] 0 Python
> > 0x000000011052172d fast_function + 264
> > [helios:47447] [11] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [12] 0 Python
> > 0x00000001105170f1 PyEval_EvalCodeEx + 1583
> > [helios:47447] [13] 0 Python
> > 0x000000011052169a fast_function + 117
> > [helios:47447] [14] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [15] 0 Python
> > 0x000000011052172d fast_function + 264
> > [helios:47447] [16] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [17] 0 Python
> > 0x000000011052172d fast_function + 264
> > [helios:47447] [18] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [19] 0 Python
> > 0x00000001105170f1 PyEval_EvalCodeEx + 1583
> > [helios:47447] [20] 0 Python
> > 0x00000001105174fc PyEval_EvalFrameEx + 537
> > [helios:47447] [21] 0 Python
> > 0x00000001105170f1 PyEval_EvalCodeEx + 1583
> > [helios:47447] [22] 0 Python
> > 0x000000011052169a fast_function + 117
> > [helios:47447] [23] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [24] 0 Python
> > 0x000000011052172d fast_function + 264
> > [helios:47447] [25] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [26] 0 Python
> > 0x000000011052172d fast_function + 264
> > [helios:47447] [27] 0 Python
> > 0x000000011051daf3 PyEval_EvalFrameEx + 26640
> > [helios:47447] [28] 0 Python
> > 0x00000001105170f1 PyEval_EvalCodeEx + 1583
> > [helios:47447] [29] 0 Python
> > 0x00000001104bbfb1 function_call + 352
> > [helios:47447] *** End of error message ***
> >
> >
> > And that's where the tests stopped. I will try with lesser processors
> but do
> > you think this is a compilation/installation issue ?
> >
> > Thanks,
> > Vardha.
> >
> >
> > On Sat, Apr 9, 2016 at 2:07 AM, Varadharajan Srinivasan
> > <varadharajan.srinivasan at gmail.com> wrote:
> >>
> >> Dear Jakob,
> >>
> >> Thank you very much for your suggestion. I will try it and report back.
> >>
> >> Best,
> >> Vardha.
> >>
> >> On Sat, Apr 9, 2016 at 1:38 AM, Jakob Schiøtz <schiotz at fysik.dtu.dk>
> >> wrote:
> >>>
> >>> This is VirtualEnv doing its “magic”. I was never able to get gpaw or
> >>> asap to work with virtualenv, because they use their own python
> interpreter
> >>> (which breaks virtualenv). But strangely, it always worked for Marcin
> who
> >>> wrote the instructions. I think it was a difference between using the
> >>> build-in python or the homebrew python.
> >>>
> >>> Try without using virtualenv. But in that case it is essential to use
> >>> the homebrew Python, so you can clean up if something goes horribly
> wrong.
> >>> Try for example to follow these instructions:
> >>>
> https://wiki.fysik.dtu.dk/asap/Installing%20ASE%2C%20Asap%20and%20GPAW%20on%20a%20Mac
> >>> Just skip the installation of Asap, you do not need it.
> >>>
> >>> Best regards
> >>>
> >>> Jakob
> >>>
> >>>
> >>> > On 08 Apr 2016, at 13:43, Varadharajan Srinivasan via gpaw-users
> >>> > <gpaw-users at listserv.fysik.dtu.dk> wrote:
> >>> >
> >>> > Dear Ask,
> >>> >
> >>> > Thank you for the prompt reply.
> >>> >
> >>> > You run "gpaw info" in serial, so it has no scalapack.
> >>> >
> >>> > Try "gpaw-python $(which gpaw) info".
> >>> >
> >>> > I ran this and got an error :
> >>> > gpaw-python $(which gpaw) info
> >>> > Traceback (most recent call last):
> >>> > File "/Users/vardha/Virtualenvs/gpaw-trunk/bin/gpaw", line 2, in
> >>> > <module>
> >>> > from gpaw.cli.main import main
> >>> > ImportError: No module named gpaw.cli.main
> >>> >
> >>> > In fact, someone should probably add that trick to the web page if it
> >>> > isn't there already.
> >>> >
> >>> > They don't find ASE installed. Please check that you can import ase
> >>> > from gpaw-python, e.g. gpaw-python -c "import ase"
> >>> >
> >>> > I then tried gpaw-python -c "import ase" and got :
> >>> > Traceback (most recent call last):
> >>> > File "<string>", line 1, in <module>
> >>> > ImportError: No module named ase
> >>> >
> >>> > Is there something wrong with my paths? I installed GPAW in
> virtualenvs
> >>> > and used the .bash_profile below.
> >>> >
> >>> > Thanks,
> >>> > Vardha.
> >>> >
> >>> > # Set architecture flags
> >>> > export ARCHFLAGS="-arch x86_64"
> >>> >
> >>> > # personal installation of pip
> >>> > export PATH=/Users/$USER/pip_latest:$PATH
> >>> > export PYTHONPATH=/Users/$USER/pip_latest:$PYTHONPATH
> >>> >
> >>> > pyver=`python -c "from distutils import sysconfig; print
> >>> > sysconfig.get_python_version()"`
> >>> >
> >>> > # pip --user installations of packages
> >>> > export PATH=/Users/$USER/Library/Python/${pyver}/bin:$PATH
> >>> > export
> >>> >
> PYTHONPATH=/Users/$USER/Library/Python/${pyver}/lib/python/site-packages:$PYTHONPATH
> >>> >
> >>> > # homebrew
> >>> > # Ensure user-installed binaries take precedence
> >>> > export PATH=/usr/local/bin:$PATH
> >>> > export
> >>> > PYTHONPATH=/usr/local/lib/python${pyver}/site-packages:$PYTHONPATH
> >>> > # hack gtk-2.0
> >>> > export
> >>> >
> PYTHONPATH=/usr/local/lib/python${pyver}/site-packages/gtk-2.0:$PYTHONPATH
> >>> > # https://github.com/mxcl/homebrew/issues/16891
> >>> > export PKG_CONFIG_PATH=`brew --prefix
> >>> > libffi`/lib/pkgconfig:$PKG_CONFIG_PATH
> >>> > export PKG_CONFIG_PATH=/opt/X11/lib/pkgconfig:$PKG_CONFIG_PATH
> >>> > export PKG_CONFIG_PATH=/usr/local/lib/pkgconfig:$PKG_CONFIG_PATH
> >>> >
> >>> > # virtualenv
> >>> > # virtualenv should use Distribute instead of legacy setuptools
> >>> > export VIRTUALENV_DISTRIBUTE=true
> >>> > # Centralized location for new virtual environments
> >>> > export PIP_VIRTUALENV_BASE=$HOME/Virtualenvs
> >>> > # pip should only run if there is a virtualenv currently activated
> >>> > export PIP_REQUIRE_VIRTUALENV=true
> >>> > # cache pip-installed packages to avoid re-downloading
> >>> > export PIP_DOWNLOAD_CACHE=$HOME/.pip/cache
> >>> >
> >>> > Then try within mpirun.
> >>> >
> >>> > Best regards
> >>> > Ask
> >>> >
> >>> > > Thanks,
> >>> > > Vardha.
> >>> > >
> >>> > > _______________________________________________
> >>> > > gpaw-users mailing list
> >>> > > gpaw-users at listserv.fysik.dtu.dk
> >>> > > https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
> >>> >
> >>> > _______________________________________________
> >>> > gpaw-users mailing list
> >>> > gpaw-users at listserv.fysik.dtu.dk
> >>> > https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
> >>>
> >>> --
> >>> Jakob Schiøtz, professor, Ph.D.
> >>> Department of Physics
> >>> Technical University of Denmark
> >>> DK-2800 Kongens Lyngby, Denmark
> >>> http://www.fysik.dtu.dk/~schiotz/
> >>>
> >>>
> >>>
> >>
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20160420/03e0203d/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test.log
Type: application/octet-stream
Size: 17255 bytes
Desc: not available
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20160420/03e0203d/attachment-0001.obj>
More information about the gpaw-users
mailing list