[MITgcm-support] parallelizing tutorial examples
Nicolas Grisouard
grisouard at cims.nyu.edu
Thu Apr 28 13:17:12 EDT 2011
Hi Francis,
I also do this:
* when creating the makefile, I add -mpi as an option for the execution
of genmake2 (although I'm sure sure if it's necessary),
* in eedata, I add line:
usingMPI=.TRUE.,
* This last bullet is very optional, its purpose is to get output fields
that are not sliced in a number of files equal to the number of
processors. So in data, I set globalFiles to TRUE (apparently it is not
recommended) and in the same section, I add the following line:
useSingleCpuIO=.TRUE.,
That's how I've been told to proceed and I'd be interested myself to
know if all of this is necessary. But unless I forgot anything, it sure
works.
Cheers,
Nicolas.
On 4/28/11 1:05 PM, Francis Poulin wrote:
> Hello,
>
> I've just started using the MITgcm and I have a rather simple
> questions. I've been able to run many of the tutorials in serial and
> they work beautifully. To run them in parallel requires some
> alterations and I'm hoping to find a list of what needs to be done.
> What I've figured out so far is:
>
> 1) Need to change the OPTFILE. I think I've done this correctly since
> it compiles ok.
>
> 2) Need to change the SIZE.h file (I actually used the SIZE.h_mpi that
> I found in exp2).
>
> 3) When I ran it there was a problem with topog.bin. I presume this
> means that I need to recreate the data file in a way that is
> consistent with MPI?
>
> I've looked through the manual and haven't found anything that lists
> all the necessary changes. I thought I'd ask just in case someone has
> done this already.
>
> Thanks in advance for your time,
> Francis
>
>
> ------------------
> Francis Poulin
> Assistant Professor
> Department of Applied Mathematics
> University of Waterloo
>
> email: fpoulin at uwaterloo.ca <mailto:fpoulin at uwaterloo.ca>
> Web: http://www.math.uwaterloo.ca/~fpoulin/
> <http://www.math.uwaterloo.ca/%7Efpoulin/>
> Telephone: +1 519 888 4567 x32637
>
>
>
>
>
>
>
>
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
--
Nicolas Grisouard
Assistant Research Scientist
Courant Institute of mathematical Sciences
251 Mercer Street
New York, NY 10012
tel: (+1) 212-992-9920 -- room 927
http://nicolas.grisouard.free.fr
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20110428/53633de0/attachment.htm>
More information about the MITgcm-support
mailing list