[gmx-users] compilation of gmx-4.0.2
Xavier Periole
X.Periole at rug.nl
Wed Nov 12 10:49:39 CET 2008
Thanks Mark and Justin,
I got the script of installation of gromacs3.3.1 from the system
manager. It works so far ... I attached it at the end (note: it is for
double precision), it might help others.
The mpicc was not found as mpcc was used instead.
Best,
XAvier
------------script------------
# 32 or 64 bit version
export OBJECT_MODE=64
# C compiler
export CC='xlc_r -q64 -qsuppress=1506-229 '
export CFLAGS='-O3 -qhot -qarch=auto -qtune=auto -qmaxmem=-1'
# C++ compiler
export CXX='xlC_r -q64'
export CXXFLAGS='-O3 -qhot -qarch=auto -qtune=auto -qmaxmem=-1'
# Fortran compiler
export FC='xlf95_r -q64'
export FFLAGS='-O3 -qhot -qarch=auto -qtune=auto -qmaxmem=-1'
# MPI compilers
export MPICC='mpcc -q64 -compiler xlc_r'
export MPIF90='mpfort -q64 -compiler xlf95_r'
# LDFLAGS linker flags, e.g. -L<lib dir> if you have libraries in a
# nonstandard directory <lib dir>
# headers in a nonstandard directory <include dir>
# CPP C preprocessor
rootdir=$PWD
echo "Rootdir: $rootdir"
export MP_HOSTFILE=$rootdir/hostfile
echo $MP_HOSTFILE
cat > ${MP_HOSTFILE} << EOF
localhost
localhost
localhost
localhost
localhost
localhost
localhost
localhost
localhost
localhost
localhost
localhost
localhost
EOF
# make fftw3 available
echo "Loading ffwt3 version"
module load fftw3
builddir=$PWD/build-dp
mkdir -p ${builddir} || exit
cd ${dir} || exit
echo "Configuring double precision mpi parallel version"
echo "./configure --prefix=${builddir} --program-suffix="_mpi" --without-x
--disable-float --enable-mpi"
./configure --prefix=${builddir} --program-suffix="_mpi" --without-x
--disable-float --enable-mpi
make mdrun
make install-mdrun
---------end script-----------------
On Wed, 12 Nov 2008 11:20:22 +1100
Mark Abraham <Mark.Abraham at anu.edu.au> wrote:
> Xavier Periole wrote:
>>
>> Dears,
>>
>> my turn to ask a stupid question:
>>
>> I am trying to install gmx402 on a IBM pSeries 575
>> CPUs: dual core IBM Power6
>>
>> when running the command:
>> ./configure --enable-mpi --prefix==/home/XXXXX/gmx402/
>>
>> I run very into the error:
>>
>> .....
>> .....
>> checking size of int... configure: error: cannot compute sizeof (int)
>> See config.log for more details.
>>
>> config.log did not help much.
>>
>> Is this a known issue? I had never seen it.
>
> Yep there's at least one known cause. By default, configure assumes that you
>are building on the same architecture on which your target will run. So when
>testing for machine properties it can use either successful compilation or
>successful execution. However in a cross-compilation environment, the latter
>will fail.
>
> There are several work-arounds. There's a tool called yodconfigure that I
>understand takes care of much of such issues, but I'm not sure how. Or you
>can trick configure into running code on the target machine somehow - this is
>tricky with GROMACS, because in src/gmxlib/nonbonded/nb_kernel you need to
>make and run on the build machine to create the generic kernels. Or you can
>tell configure you are cross-compiling with the --build and --host flag, so
>that it uses only successful compilation as a diagnostic tool.
>
> I'm surprised that lack of mpicc caused such a problem and didn't cause a
>problem when testing mpicc.
>
> Mark
> _______________________________________________
> gmx-users mailing list gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the www interface
>or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
-----------------------------------------------------
XAvier Periole - PhD
- Molecular Dynamics Group -
Computation and NMR
University of Groningen
The Netherlands
-----------------------------------------------------
More information about the gromacs.org_gmx-users
mailing list