[MITgcm-support] file size issue with mpi

Michael Schaferkotter (Contractor) Michael.Schaferkotter at nrlssc.navy.mil
Tue Aug 24 10:05:19 EDT 2004


greetings:

i/ve been interested in the discussion about MPI and the output data.
in my case the output has 'holes' on some of the tiles.

patrick wrote:
from my experience the option
globalFiles = .TRUE.
should be used with caution, or not used at all.
>From what I know, parallel write to global file is
not a well defined operation on some/many(?) parallel platforms
and not strictly supported by MPI.
I ran into similar problem as recently as last week on an SGI Altix.
Filesize there was actually OK, but content was wrong.

So bottom line is (I think Chris/Alistair would agree)
it's implemented in the model and works on some platforms,
but by far not on all platforms.

-Patrick

the compiler/mpich tandem that i/m using is pgi5.2/mpich-1.2.6.

similar problems occurs with pgi5.1/mpich-1.2.5.2.

i found in the data file that globalFiles = .TRUE. .

could this be the problem?

i changed globalFiles to .FALSE..

the output now (it seems) is written to a separate file for each tile? (is that the way it works now?)

how is the output combined back into a single file for post processing?

michael



More information about the MITgcm-support mailing list