[MITgcm-support] Differences in parrallel output

Patrick Heimbach heimbach at MIT.EDU
Wed May 24 16:52:05 EDT 2006


Ciao Gianmaria,

would be good to know whether it's a
summing problem or an MPI problem.

To find out, could you try running a 
single-processor test where you leave nPx=nPy=1
and do the decomposition using nSx, nSy?

-Patrick



On Wed, 2006-05-24 at 07:19, Gianmaria Sannino wrote:
> Hi folks,
> 
>  
> 
> I’ve downloaded the most recent version of MITgcm last week and I’ve
> run my experiment.
> 
> Comparing the output files for different parallel decompositions (e.s.
> 1x1, 2x1, 2x2)
> 
> they are different. 
> 
>  
> 
> I’m running the model on an IBM SP4, with the  useSingleCpuIO=.TRUE..
> 
>  
> 
> The differences I observe are obtained using the bdiff command.
> 
> Using a previous version (august 2005)  of MITgcm these differences
> did not appear.
> 
>  
> 
> Anyone could help me?
> 
>  
> 
> Thanks
> 
> gianmaria
> 
> 
> 
> ______________________________________________________________________
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
-- 
--------------------------------------------------------
Patrick Heimbach   Massachusetts Institute of Technology
FON: +1/617/253-5259                  EAPS, Room 54-1518
FAX: +1/617/253-4464             77 Massachusetts Avenue
mailto:heimbach at mit.edu               Cambridge MA 02139
http://www.mit.edu/~heimbach/                        USA




More information about the MITgcm-support mailing list