a problem in memory settings


Just Got Here
Hello,

I’m trying to run a DFT simulation of the molecule Cr8F8(O2CCH3)16. It is a very large system containing 632 electrons and I try to calculate it with cc-pvtz and different choices of exchange.
My question is what should be the appropriate choice of memory settings for this calculation?
The cluster I’m running on contains nodes of 12 cpus with a total memory of 48 Gb for each node.
My job file includes the following lines:

  1. PBS -l nodes=8:ppn=12,walltime=12:00:00
cd $PBS_O_WORKDIR
module load NWChem/6.1.1-goolf-1.4.10-2012-06-27-Python-2.7.3
mpiexec -np 96 nwchem input 2>&1 | tee output

I tried to use different number of nodes (4-16) and number of mpi processes accordingly (nodes number * 12).
In the input file I wrote:

Memory total 3000 mb.

The input file includes also the keywords: direct and noio

Sometimes the calculations are completed successfully, but many times they collapse. I didn’t succeed to find memory settings which for sure will be sufficient for these calculations.

Can you tell me what would be the best memory choices for this calculation?

Thank you in advance,

Shira

:)


Forum >> NWChem's corner >> Running NWChem