[MITgcm-support] MITgcm on BCp2: domain decomposition

[Martin Losch]

Hi Fanny,

I think that all of this could have been done without modifying linux_amd64_gfortran. Besides defining the environment variable NETCDF_ROOT (or NETCDF_HOME, NETCDF_INC/LIB)
you can define "export MPI_INC_DIR=/usr/mpi/gcc/openmpi-1.4.2-qlc/include" (see also the header of linux_amd64_gfortran)

I don't understand why 2. is required, as it is already done in the opt-file.

Martin

PS. Of course, Yuan is right and you can include this MPI_INC_DIR into the opt file.

On Dec 13, 2011, at 12:24 PM, Fanny Monteiro wrote:

> Hi,
> 
>    I have been compiling and running on parallel processors the MITgcm on the HPC of the University Bristol (BlueCrystal Phase 2).  
> 
> We had to make a few changes and I am wondering if anyone has an idea of a better way of doing it (especially in relation to point 3).
> 
> Here are the changes that our IT specialist needed to make in order to get genmake2 to run without warnings and the model to compile:
> 
> 1. Set the environment variable NETCDF_ROOT.  That's fine.
> 
> 2. In the optfile, add '-l netcdf' to the line:
>        LIBS="-L${NETCDF_ROOT}/lib -l netcdf"
> I would have thought everyone would have needed that.
> (I used linux_amd64_gfortran, but we see that, e.g. linux_ia64_ifort would have required a similar edit.
> 
> 3. After getting my Makefile, I needed to add the include path to my MPI headers:
>        INCLUDES = -I/gpfs/cluster/geog/ggdagw/netcdf/4.0/gnu_fc_4.1.2/include -I/usr/mpi/gcc/openmpi-1.4.2-
> qlc/include
> so that make depend would not give a warning about not finding "mpif.h"
> 
> 4. Then I could compile cleanly--without fixing (3), I'd get a compile error.
> 
> Thanks in advance for your help/comments.
> Cheers,
> Fanny
> 
> 
> 
> -- 
> Dr Fanny M Monteiro
> Research Associate
> School of Geographical Sciences
> University of Bristol, Bristol BS8 1SS, UK
> Phone (+44) 117 33 18351
> Email f.monteiro at bristol.ac.uk
> Web https://sites.google.com/site/fmonteir/
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support