Page 1 of 1

NCORE warning during NEB calculation

Posted: Tue Nov 21, 2023 12:04 pm
by julien_steffen
Dear VASP community,

I'm trying to run a NEB calculation in VASP 6.4.1 with the VTST extension and am facing issues with the NCORE setting.
Specifically, I'm trying to calculate the adsorption barrier of N2 adsorption on metal clusters.

Despite it being one of the first keywords set in my INCAR file in the root directory, the warning to set NCORE = 2
is shown at the beginning of the root's vasp.out file and of the subdirectories' OUTCARs.
Setting NCORE from my usual 26 to 2 didn't help and the calculations are running equally slow in both cases, so I'm thinking
that somehow the NCORE settings are being ignored, even though they're clearly visible in the INCAR section at the top of the OUTCAR.

I'm running the calculation on 2 nodes which both have a total of 104 cores (2 CPUs x 56), so I did consider that IMAGES needs
to be an integer of the number of cores.

I'm attaching the vasp.out from the root directory and the OUTCAR file from the first intermediate structure I'm optimizing, as well as the
POSCARs of the initial and final structure.
The INCAR is shown below.
Also out of interest, what does the vasp.out file in the root directory actually tell me?
The actual calculations are performed in the subdirectories as far as I'm aware, so what's the point except monitoring whether the calculations are done?


Best regards


INCAR:

Code: Select all

System=Ru12N2
PREC=Normal
GGA=PE
NCORE=8
ISPIN=2
MAGMOM=2*0 12*3
NELM=150
NELMIN=8
ENCUT=450
EDIFF=1e-7
LREAL=.FALSE.
ALGO=normal
ISMEAR=-1
SIGMA=0.04
IVDW=12
LWAVE=.False.
LCHARG=.false.
ISYM=-1
IBRION=1
NFREE=2
ISIF=2
EDIFFG=-0.01
NSW=2000
POTIM=0.5
ICHARG=2
AMIX=0.1 
BMIX=0.01
IMAGES=4
SPRING=-5
LCLIMB=.TRUE.

Re: NCORE warning during NEB calculation

Posted: Thu Nov 23, 2023 11:07 am
by fabien_tran1
Hi,

The warning "... NCORE = 2 ..." is displayed for both NCORE=2 and 26, because these two values may be too small and too large, respectively, to be in the reasonable range. With an intermediate value, e.g. 4 or 8, the warning should disappear and the calculation be maybe (?) faster.

A few more comments:

-At the beginning of your vasp.out file there is "MPI startup(): Warning: I_MPI_PMI_LIBRARY will be ignored since the hydra process manager was found". It is only a warning, but maybe search for info that is really harmless for performance.

-The ordering of the keywords in INCAR does not matter.

-In addition to the info on our wiki about parallelization:
https://www.vasp.at/wiki/index.php/Cate ... lelization
https://www.vasp.at/wiki/index.php/Opti ... lelization
useful comments can also be found here for instance:
https://www.vasp.at/forum/viewtopic.php?f=4&t=19211