[MITgcm-devel] linux_amd64_ifort+mpi_ice_nas

[David Ferreira]

Hi Dimitri,
I'm going to run a testreport today.  We'll see if your changes make any 
difference (probably not).

As for other options to update, I don't know. I'm running testreport 
with this:
module load comp-intel/2011.2 mpi-sgi/mpt.2.06a67 netcdf/4.0
and this works for all nodes.

That said, I think that comp-intel/2011.7.256 or higher is recommanded 
for Sandy bridge nodes, but it does change the results to change the 
compilers.
david



On 4/23/13 9:15 AM, Menemenlis, Dimitris (3248) wrote:
> Definitely!  I already changed the "comment line" to indicate 2.06a67 and meant to email you about this. Do you agree with the CC=icc change? It is needed for using memory model medium. I do not know how to run the verification experiment on Pleiades so I hope it did not break because of the CC change. Are there other updates to options that are needed for 2.06a67? I copy Devel in case other Pleiades users want to pitch in.
>
> Thanks
>
> Dimitris Menemenlis
> 818-625-6498
>
> On Apr 23, 2013, at 6:02 AM, "David Ferreira" <dfer at mit.edu> wrote:
>
>> Hi Dimitri,
>> Your check-in reminded me that we should update this optfile to the newer mpi-sgi/mpt.2.06a67 (instead of mpi-sgi/mpt.2.04.10789 now).
>> I'm already using the new version (and running the testreport with it and it doesn't change anything to any verification experiments).
>> OK with the change ?
>> cheers,
>> david
>>
>>
>> On 4/22/13 5:20 PM, Dimitris Menemenlis wrote:
>>> Update of /u/gcmpack/MITgcm/tools/build_options
>>> In directory forge:/tmp/cvs-serv25995
>>>
>>> Modified Files:
>>>     linux_amd64_ifort+mpi_ice_nas
>>> Log Message:
>>> adding "CC=icc" which is needed for -mcmodel=medium
>>>
>>>
>>> _______________________________________________
>>> MITgcm-cvs mailing list
>>> MITgcm-cvs at mitgcm.org
>>> http://mitgcm.org/mailman/listinfo/mitgcm-cvs