[MITgcm-support] combining multi-PE NetCDF files into a single file

Martin Losch mlosch at awi-bremerhaven.de
Mon Feb 7 10:13:22 EST 2005


Eli,
I think I have checked in a version with a hack that handles multiple 
processors (NOT tiles) for a lat-lon (exch1 type) domain.
You'll find it under
utils/matlab/mnc_assembly.m
and not in Ed's development area. This version does not have the fix 
for Matlab6.5. If you find that it  works after applying the fix, tell 
me and I'll check it in.

Ed, sorry for this confusion, we now have different version of 
mnc_assembly.m in the repository, on in MITgcm and one in 
delevelopment/edhill/mnc_assembly
I don't know what you want to do about it, but I suggest to have the 
one im MITgcm/utils/matlab  as the "reference" version.

Martin

On Feb 7, 2005, at 3:42 PM, Eli Galanti wrote:

>
> Ed, Martin,
>
> mnc_assembly works now with Matlab6.5, thanks. but... I get the same 
> problem as
> Martin:
>
>> Another problem:
>> For my output mnc_assembly does not assemble anything, but writes only
>> one tile to the output file. I don't know why this is, but I suspect 
>> is
>> has to do with the bi,bj's in my netcdf files:
>> they are all 1. Ed, I don't know whether I have this because don't use
>> the latest version of the code (haven't tried it, yet), or because
>> bi,bj=1 if you use only 1 tile per processor? Your examples all have
>> nPx=nPy=1 and nSx,nSy>1, whereas I have nSx=nSy=1 and nPx,nPy>1
>
> I tried to bypass it by making mnc_assebly use nPx,nPy instead of 
> bi,bj but
> compined file didn't look very nice...
>
> thanks, Eli.
>
> -- 
> Dr. Eli Galanti
> Department of Geophysics and Planetary Sciences
> Tel-Aviv University, Ramat Aviv 69978, Israel.
> Tel:  972-3-6408274   Fax:  972-3-6409282
> E-mail: galanti at tau.ac.il
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
>
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support




More information about the MITgcm-support mailing list