[Wien] weird parallel problem

Peter Blaha pblaha at theochem.tuwien.ac.at
Thu Nov 16 17:27:06 CET 2006


Sorry, but I have to disagree.

Maybe it was never mentioned, but the "NFS-bug" problems NEVER appeared 
with the ifc7 executables (on our sites), so it is specific to ifort AND 
some Linux-Kernel version.

If you can, try to upgrade the Linux.

Otherwise our conclusion was to remove files before they ere rewritten.
I've put several such rm ...   statements in the scripts by now, but I'm 
not sure if I've done this for lapwsopara !
One could remove all energyso, vectorso,... files before lapwso.


Martin Kroeker schrieb:
> Given that replacing the executables is sufficient to solve the problem,
> I think NFS problems can be ruled out in your case. Perhaps the newer ifc
> versions optimize the code to a point where some variable is not properly
> initialized, but assumes whatever happens to be hanging around at is memory
> location. This might explain why it behaves differently on two systems (btw
> do two consecutive runs yield the same result ?).

-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: 
http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------


More information about the Wien mailing list