[molpro-user] too many records error

Andy May MayAJ1 at cardiff.ac.uk
Thu Jan 13 09:55:02 GMT 2011


Kirk,

The mixed up output is almost certainly the non-master processes writing
to the master output stream. I've fixed such problems in the past and
they normally relate to what happens if one writes to unit 0 or *, or 6
when it has been closed. The behaviour is compiler dependent, so if you
can send the CONFIG file, or at least the Fortran compiler used from the
top of the output that would be great. Most likely the variable iout has
been redefined somewhere previously during the calculation, so the full
input would be useful to see exactly what path has been taken.

Thanks,

Andy

On 12/01/11 21:43, Kirk Peterson wrote:
> Dear Molpro users,
>  
> a colleague was running a relatively straightforward integral-direct DFT calculation in Molpro 2010.1 (patch level 3) and had the calculation die after iteration 69 with:
> 
>    67      0.356D-05      0.355D-06      -612.52616646   1981.496220    0.94220   -1.96679   -1.07479    9    497.7  34008.5
>    66      0.109D-04      0.647D-06      -612.52616635   1981.496296    0.94295   -1.96630   -1.07416    9    504.0  33416.6
>    68      0.216D-05      0.276D-06      -612.52616652   1981.495727    0.94255   -1.96679   -1.07377    9    496.5  34504.9
>    67      0.356D-05      0.355D-06      -612.52616646   1981.496220    0.94220   -1.96679   -1.07479    9    502.7  33919.3
>    69      0.689D-05      0.845D-07      -612.52616660   1981.495949    0.94139   -1.96853   -1.07243    9    506.6  35011.5
>  More than                       200  records on file                         1
>  ? Error
>  ? Too many records
>  ? The problem occurs in reservem
> 
> 
> I saw a similar report on the user list all the back in 2003, but no accompanying solution was posted.  There was also perhaps a similar problem posted but under the guise of lattice point charges.  I could attach the accompanying input, but the meat of it is just:
> 
> 
>  ***,es
>  memory,220,M
> 
>  gthresh,oneint=1.e-15,twoint=1.e-15,energy=1.e-10,coeff=1.e-6,grid=1.e-10
>  gdirect,thrdscf=1.e-13
>   
>  basis,default=avtz         ! 828 basis functions with 92 electrons
>   
>  geomtyp=xyz
>  symmetry,nosym
>  orient,noorient
>  angstrom
>  geometry={
>      24
> ....
> ....
> }
> 
>  wf,charge=-1,spin=1
>  {uks,pbe0;accu,13;maxit,200}
>  E1=energy
> 
> 
> It's being run 8-way parallel.
> 
> thanks in advance,
> 
> -Kirk
> 
> PS - I also just noticed that it's a bit strange that the iterations are not consecutive and with some output repeated, e.g.,  1, 3, 2, 4, 3, 5,.....   Perhaps the parallel implementation is messed up? (It's being run on the DOE NERSC computers).
> 
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user



More information about the Molpro-user mailing list