[MITgcm-support] [EXTERNAL] How to make the executable with Intel compiler?

[Martin Losch]

Hi,

I think Ou already pointed at the most likely problem: the build option file.

I would start from a more generic build options file (depending your system this will probably be something like tools/build_options/linux_amd64_ifort or tools/build_options/linux_ia64_ifort (check your "uname -a” output).

For mpi-builds you may need to point the model to the MPI installation (that you loaded with “impi”), i.e. where is the mpif.h file. and set MPI_INC_DIR prior to running genmake2 (export MPI_INC_DIR=/yourMPIpath/include, your module load impi command may have set a variable like MPI_HOME, or similar, in which case it would be export MPI_INC_DIR=$MPI_HOME/include or similar). 

In case of problems, consult the genmake.log file if you can spot any obvious errors (include files or libraries not found). You can also try to figure out when exactly this error occurs. For that do not use "make -j 16”, but just “make” (will be slower, but the output will be more “linear”). 

Martin

> On 4. May 2023, at 05:38, Wang, Ou (US 329B) <ou.wang at jpl.nasa.gov> wrote:
> 
> Hello,  This may be related to your environment variables LANG and/or LC_ALL.
> Try setting it to en_US.UTF-8
> bash:
> export LANG=en_US.UTF-8
> or csh:
> setenv LANG en_US.UTF-8Also, you are using the optile linux_amd64_ifort+mpi_ice_nas, which works for NASA’s Pleiades but may not work for your machine and compilers. There are various optiles in MITgcm/tools/build_options/. Try to use one that matches your machine and compilers. You can read more about how to build MITgcm in the readthedoc MITgcm manual, section 3.5 at
> https://mitgcm.readthedocs.io/en/latest/getting_started/getting_started.html#building-the-model
>  Best,
> Ou Wang
>   From: MITgcm-support <mitgcm-support-bounces at mitgcm.org> on behalf of Zih En Tseng <zizien at umich.edu>
> Date: Wednesday, May 3, 2023 at 6:58 PM
> To: mitgcm-support at mitgcm.org <mitgcm-support at mitgcm.org>
> Subject: [EXTERNAL] [MITgcm-support] How to make the executable with Intel compiler?
> To those who may concern,
>  Hi, I’m working on the CYGNSS project and I’m trying to reproduce ECCOv4r4 data.
>  I followed the reproduce instruction (https://ecco-group.org/docs/v4r4_reproduction_howto.pdf) step by step.
>  So after cloning all essential data I did:
>  $ module load intel impi
> $ ../../../MITgcm/tools/genmake2 -rd=../../../MITgcm -mods=../code -optfile=../code/linux_amd64_ifort+mpi_ice_nas -mpi
> $ make -j16 depend
> $ make -j16 all
>  but I encountered the following error:
>  Catastrophic error: could not set locale "" to allow processing of multibyte characters
>  I searched for this error, and some forum suggested it is a problem with Intel compiler, but my IT consultant think that:
> “it has little to do with the Intel compiler. This is problem with the overly complicated build systems that derivate from standard practices.”
>  Have you encountered a problem like this? How did you solve this error? Is there any specific requirement on the version of compilers?
>  Sorry that I’m not super familiar with the Fortran code structure, but please let me know if there is any more detail that I can provide to help explain my problem clearer and perhaps make it easier for us to figure out what happened.
>  Thanks!
> -
> Alan (Zih-En Tseng)
> Graduate Student Research Assistant
> Department of Naval Architecture and Marine Engineering
> University of Michigan
> Mail: zizien at umich.edu
> Mobile: +1(737)206-9891
>  _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support