Hi all,
currently I'm running the following calculation for a 34-atomic gold cluster (gemoetry data removed) on 46 cores (Intel(R) Xeon(R) CPU E5-2665 0 @ 2.40GHz):
title Au34
start Au43
charge -1
geometry units angstrom
...
...
...
end
basis spherical
* library aug-cc-pV5Z-PP
end
ecp
* library stuttgart-koeln_mcdhf_rsc_ecp
end
RELATIVISTIC
ZORA ON
end
DFT
MULT 2
XC m06
convergence energy 1e-8
iterations 100
end
DRIVER
clear
DEFAULT
maxiter 2000
end
set driver:linopt 0
task DFT optimize
For the last 24 hours, no output was written. The last lines in the output are:
....
....
....
Superposition of Atomic Density Guess
-------------------------------------
Sum of atomic energies: -3978.06149212
Renormalizing density from 646.00 to 647
That is, within 24 hours, nwchem did not even arrive at the first iteration of the self consistency loop for the first energy evaluation.
I have no experience with nwchem. Is this long computing time to be expected for this system, or is the run most probably frozen?
Best,
Elvis
Edit:
I just did a test run for a Au-dimer and here (same input file as for Au34) a complete energy evaluation (that is a complete scf cycle) just needs a few minutes. I'm really puzzled whether the Au34 runs is really supposed to be so slow, or whetether there is possibly a dead-lock or something.
|