[Wien] Time taken to run a 60 atom/cell program

[Peter Blaha]

Unless this a a magnetic and metallic 4f system or similar your 
parameters a way too large.

For an insulator one can even start with 1-4 k-points only, run the scf 
cycle, even do a structure minimization. Once this is done, CHECK your 
results with more k-points. Usually you reach convergence of the new scf 
cycle quickly and just compare the quantity of interest with more/less 
k-points to have an idea about its convergence.

For many properties (spectra,..) a good k-mesh is finally necessary, but 
it does not need to be done in scf, but one iteration is enough.

Similar with RKMAX: if your "smallest atom (RMT) is for a non-transition 
metal (C,N,O,..) you can even start with RKmax=4-5 in very big cases. At 
least the structurl optimization will be pretty ok with such values. 
Again, it is ABSOLUTELY NECESSARY to continue later with better RKMAX 
and COMPARE your results for the two parameter sets.

I strongly suggest to get used to such calculations with (much) smaller 
systems !!!!


Am 14.02.2012 04:38, schrieb Ghosh SUDDHASATTWA:
> 60 atoms/cell is a very large primitive cell/supercell. I once tried
> with 58 atoms /cell with 1200 k-points and RMT=8.00. 1 SCF cycle took
> almost over 3 hours on 72 processors. So I reduced my k-mesh and ran the
> SCF (non-availability of processors !!!!).
>
> If you do the same, may be on 100 processors, then you can expect 1 SCF
> cycle to take around 1-2 hours.
>
> ------------------------------------------------------------------------
>
> *From:*wien-bounces at zeus.theochem.tuwien.ac.at
> [mailto:wien-bounces at zeus.theochem.tuwien.ac.at] *On Behalf Of
> *Chee-Keong Tan
> *Sent:* Monday, February 13, 2012 8:28 PM
> *To:* wien at zeus.theochem.tuwien.ac.at
> *Subject:* [Wien] Time taken to run a 60 atom/cell program
>
> Dear all,
>
> I am new to this Wien2k program. So would you please answer me? Will it
> take only hours or 1-2 days to run?
>
> I know the question might be too trivial, but hopefully those who have
> experiences can help me out.
>
> Thanks,
>
> chee
>
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

-- 
Peter Blaha
Inst.Materials Chemistry
TU Vienna
Getreidemarkt 9
A-1060 Vienna
Austria
+43-1-5880115671