ARMCI DASSERT fail


Just Got Here
Hello:

I getting the following error. I am running a b3lyp potential scan and am using a python script to calculate the rate constants. I did not write the script, but it does run just fine on the author's machine. I am using nwchem-6.1.1 with the python options. The build does not include MPI. Does anyone know what this means and how to resolve it?
Thanks

0:Segmentation Violation error, status=: 11
(rank:0 hostname:chesun11 pid:26719):ARMCI DASSERT fail. ../../ga-5-1/armci/src/common/signaltrap.c:SigSegvHandler():310 cond:0
 0: ARMCI aborting 11 (0xb).

Forum Vet
Smv,
Would you mind posting the script you are using?
Thanks, Edo

Just Got Here
Hi Edo, I had mentioned before that I did not write the script. The author would prefer for me not to post it if possible. I could still really use some advise, could I send it offline?

Forum Vet
Stephanie,
I managed to reproduce and fix the problem on my Ubuntu 12.04 box.
The problem is that the the python scientific libraries use lapack libraries and this causes a conflict with the what NWChem uses
(because of the size the integer arguments used).
Here are what you need to do

1) env. variables' settings
export BLASOPT="-llapack -lblas"
export BLAS_LIB="$BLASOPT"
export LAPACK_LIB="$BLASOPT"
export BLAS_SIZE=4
export LAPACK_SIZE=4
export USE_64TO32=Y

2)
cd $NWCHEM_TOP/src
rm -f 64_to_32 32_to_64
make 64_to_32
rm -rf tools/build tools/install
make FC=gfortran

You might get a linking error at this point.
Please execute the next step

3)
cd lapack; make FC=gfortran
cd ..
cd blas; make FC=gfortran
cd ..
make FC=gfortran link

This is what I did.
Let me know if it works for you, too.

Cheers, Edo

Just Got Here
Yes, this solved the problem. Thanks!


Forum >> NWChem's corner >> Running NWChem