[CPMD-list] Error compilation of CPMD 3.11.1 on SGI_ORIGIN

[Axel Kohlmeyer]

On 5/17/06, Reinaldo Pis Diez <pis_diez at yahoo.com.ar> wrote:
>         Hi Axel,
>
>                 Well, first of all thanks four your useful answer.
>
> >
> > anyway, i don't have an SGI to test here,
> > but my guess is that the fix for that would
> > be to either change the declaration of TAG(2)
> > and P(2) into TAG(*) and P(*) _or_ to remove
> > the lines with:
> >       POINTER (IP_TAG,TAG),(IP_P,P)
> >
> >      CALL MEMORY(IP_P,2,'P')
> >      CALL MEMORY(IP_TAG,2,'TAG')
> >
> >      CALL FREEM(IP_P)
> >      CALL FREEM(IP_TAG)
> >
> > please let us know if any of that works, so
> > we can fix the source code accordingly.
> >
>
>                 The first attempt (change declarations) didn't succeed, compilation
> errors still arise. But the second attempt worked fine (after
> recompiling wrener.f w/o optimization)!!

yep. being unable to compile wrener.F with optimization has been
a known problem of the SGI compilers for ages. i guess after moving
to itanium processors, they don't care to fix it anymore...

>                 Then, I ran two arbitrary tests (std/wat32/inp-1 and
> linear-scaling/inp-1) and, besides some numerical diffs around the
> fifth or sixth place after the comma, results are basically the same.

several numbers can have fluctuations. mostly the total energy
should be converged similar. but you also have to pay attention to
the version of the code that ran the reference output. if it is a 3.7.x
version, please note that there were a few changes in the code, that
give a different electrostatic energy (but energy differences are still
the same). also you have to make sure, that you have the same
grid (depends a bit on the FFT library used).

> Are you interested in some special test to be safe about the fix? Just
> tell me or send me the input and I'll run it.

well, the essential test would be to run a surface hopping input on
the thusly modified code, since this part is only used for surface hopping.

perhaps one of the 'hoppers' in bochum would care to test it.

best regards,
     axel.

>                 Regards,
>
>                                         Reinaldo
>
>
>
>
> ___________________________________________________________
> 1GB gratis, Antivirus y Antispam
> Correo Yahoo!, el mejor correo web del mundo
> http://correo.yahoo.com.ar
>
>
>


-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
  Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.