Dear Edo,
Thank you very much for your help !
I see, the oldtimer "golden rule" to use matching fortran and C compilers if possible to avoid underscoring/ABI/runtime problems is no longer valid. Of course underscoring was between fortran frontends, but the openmpi which GA links against was build using icc and not gcc ...
You are completely right, with gcc and ifort the "testtask" works after a fresh untaring of the source.
Still, seeing a mess like this when using compilers claiming to be standard conforming C compilers leaves me with a bad feeling.
Could the documentation be updated with a hint regarding this ? I would expect that many people would try to use a "matched team" of compilers as I did, especially after reading www.alcf.anl.gov/resource-guides/nwchem, note their
# Intel compilers
COMPILER=intel
CC=/soft/intel/11.1.059/bin/intel64/icc
FC=/soft/intel/11.1.059/bin/intel64/ifort
Best Regards
Christof
Build Environment:
tar -zxvf ~/src/NWCHEM/Nwchem-6.1.tar.gz
module add intel-12
setenv NWCHEM_TOP /local/neuro24/nwchem-6.1
setenv NWCHEM_TARGET LINUX64
setenv NWCHEM_MODULES "all"
setenv LARGE_FILES TRUE
setenv USE_NOFSCHECK TRUE
setenv LIB_DEFINES "-DDFLT_TOT_MEM=195853376"
setenv TCGRSH /usr/bin/ssh
cd $NWCHEM_TOP/src
module add openmpi-1.4.5_intel
setenv USE_MPI y
setenv USE_MPIF y
setenv USE_MPIF4 y
setenv MPI_LOC /usr/local/stow/openmpi-1.4.5_intel
setenv MPI_LIB $MPI_LOC/lib
setenv MPI_INCLUDE $MPI_LOC/include
setenv LIBMPI "-L/usr/local/stow/openmpi-1.4.5_intel/lib -lmpi_f90 -lmpi_f77 -lmpi -lpthread"
setenv BLASOPT "-L$MKLROOT/lib/intel64 -lmkl_intel_ilp64 -lmkl_sequential -lmkl_core -lpthread -lm -I$MKLROOT/include"
make nwchem_config
make FC="ifort"
|