Quote:Edoapra Nov 13th 3:44 pmNWChem 6.6 contains a bug that makes the (T) part crash when the run is oversubscribed (i.e. you are using too many processors). Please just four processors as in the example below
#!/bin/bash -l
SBATCH -J a1_avtz
SBATCH -C haswell
SBATCH -p regular
SBATCH -N 1
SBATCH -t 05:00:00
SBATCH -o Sc.o%j
SBATCH --mail-type=BEGIN,END,FAIL
SBATCH --mail-user=nothez@mail.com
module load nwchem
srun -n 4 nwchem Sc.nw >& Sc.out
Dear Edoapra,
I've try to re-run the jobs with your new TCE section and new run file.
It works for Sc quartet state, but doesn't for any other spin state and other transition metals.
usually it ends up with the error like that, this particular one is for neutral Sc (sextet) atom:
Multipole analysis of the density wrt the origin
------------------------------------------------
L x y z total open nuclear
- - - - ----- ---- -------
0 0 0 0 0.000000 -5.000000 21.000000
1 1 0 0 0.000000 0.000000 0.000000
1 0 1 0 0.000000 0.000000 0.000000
1 0 0 1 -0.000000 -0.000000 0.000000
2 2 0 0 -23.929351 -20.140909 0.000000
2 1 1 0 0.000375 -0.001093 0.000000
2 1 0 1 -0.000073 0.000211 0.000000
2 0 2 0 -23.579043 -21.160950 0.000000
2 0 1 1 -0.067669 0.197042 0.000000
2 0 0 2 -23.916280 -20.178970 0.000000
------------------------------------------------------------------------
tce energy failed 0
------------------------------------------------------------------------
------------------------------------------------------------------------
current input line :
32: task tce energy
------------------------------------------------------------------------
------------------------------------------------------------------------
This type of error is most commonly associated with calculations not reaching
convergence criteria
------------------------------------------------------------------------
For more information see the NWChem manual at
http://nwchemgit.github.io/index.php/NWChem_Documentation
For further details see manual section:
[0] Received an Error in Communication: (-1) 0:tce energy failed:
Rank 0 [Tue Nov 14 15:36:48 2017] [c6-5c2s7n1] application called MPI_Abort(comm=0x84000004, -1) - process 0
forrtl: error (76): Abort trap signal
Image PC Routine Line Source
nwchem 0000000005476B25 Unknown Unknown Unknown
nwchem 0000000005474747 Unknown Unknown Unknown
nwchem 000000000541F924 Unknown Unknown Unknown
nwchem 000000000541F736 Unknown Unknown Unknown
nwchem 00000000053BAE06 Unknown Unknown Unknown
nwchem 00000000053C19A8 Unknown Unknown Unknown
nwchem 00000000051B2300 Unknown Unknown Unknown
nwchem 00000000053AC94B Unknown Unknown Unknown
nwchem 000000000549BC58 Unknown Unknown Unknown
nwchem 0000000005235132 Unknown Unknown Unknown
nwchem 00000000052081E9 Unknown Unknown Unknown
nwchem 0000000002DC41F4 Unknown Unknown Unknown
nwchem 0000000002D6DFB7 Unknown Unknown Unknown
nwchem 0000000000B14E4B Unknown Unknown Unknown
nwchem 0000000000413D31 Unknown Unknown Unknown
nwchem 000000000040AA80 Unknown Unknown Unknown
nwchem 000000000040A59E Unknown Unknown Unknown
nwchem 0000000005495B00 Unknown Unknown Unknown
nwchem 000000000040A487 Unknown Unknown Unknown
srun: error: nid12829: task 0: Aborted
srun: Terminating job step 8366663.0
slurmstepd: error: *** STEP 8366663.0 ON nid12829 CANCELLED AT 2017-11-14T15:36:48 ***
srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
srun: error: nid12829: tasks 1-3: Killed
These jobs are only a single point energy calculations of bare one atom.
What about the optimization or frequency calculation for similar but bigger systems, like MCl or MCl3?
Is it possible to use more then 4 cpu's for ROHF approximation with CCSD(T) method ?
Could you help me and provide me the all-purpose TCE section and run-script for OPT and FREQ jobs also?
Thank you,
best wishes
Nothez
|