I have to make a simple calculation with NWChem, so I downloaded it and tried to install on my notebook, here my impressions.
The first impression of NWChen is not so good, they use their own installation precedure, that often fails, I really do not understand why they don’t use the standard Autoconf .
At the beginning I tried to install NWChem with ATLAS and ScaLapack. It compiles, but the I got “Segmentation fault” and ” invalid memory reference” so I decided to remove Scalapack and use their internal BLAS and LAPACK.
Moreover if you try to compile without MPI you get an error message “You must set USE_MPI and have a working MPI installation to compile NWChem” so no way to test it in serial. First of all I installed openmpi instead of using the one of the system in the forlder /usr/local/gfortran. Then my compile script for NWChem reads:
export NWCHEM_TOP=/home/attacc/SOFTWARE/Nwchem-6.5 export NWCHEM_TARGET=LINUX64 export ARMCI_NETWORK=MPI-TS export USE_MPI=y export USE_MPIF=y export USE_MPIF4=y export MPI_LOC=/usr/local/gfortran export MPI_LIB=$MPI_LOC/lib export MPI_INCLUDE=$MPI_LOC/include export LIBMPI='-lmpi -lmpi_usempi -lmpi_mpifh -lpthread' export NWCHEM_MODULES="all" export CC=gcc export FC=gfortran export USE_NOFSCHECK=TRUE export LARGE_FILES=TRUE export MPICC=/usr/local/gfortran/bin/mpicc cd $NWCHEM_TOP/src make nwchem_config make FC=gfortran | tee make.log
In the future I will try to compile with more optimized libraries and python.
Finally some tutorial on NWChem: