[MITgcm-devel] sx8 and testing

[Martin Losch]

Jean-Michel,

There are a few experiments that are failing on sx8 for different  
reasons:
- adjustment.cs-32x32x1.nlfs fails (on purpose, or at least as  
expected, as you told me previously)
- fizhi-cs_32x32x40, fizhi-cs-aqualev20: floating point exception ->  
some coding problems which I am not responsible for
- dome, global_ocean.cs32x15.icedyn/thsice, offline_exf_seaice/ 
seaicetd (when mnc is not used) all finish correctly, but somehow  
testreport does not recognize the output.txt that is properly  
produced. I have not yet bothered to find out. I has to do with the  
fact that in my testing scheme the "NORMAL END" statement, that  
testreport is looking for, is difficult to produce.
- lab_sea (and also offline_exf_seaice when mnc is used) breaks  
because there are many netcdf files being opened. For some reason,  
opening a netcdf file requires a lot of core memory on the NEC-SX8,  
and when I ask for more, I exceed the limit of the smallest queue, so  
I decided not to bother.

So, basically there is no problem that I can not explain (o:
Martin


> Not related: Martin,
> on sx8: dome, offline_exf_seaice and offline_exf_seaice.seaicetd
> don't do very well. The MPI tests has been added recently,
> and I suspect globalFile to be the problem.
> Could you try to re-run those test exp
> on sx8 but commenting out "globalFiles=.TRUE.," ?
>
> Thanks,
> Jean-Michel
>
> On Mon, May 25, 2009 at 09:00:52AM +0200, Martin Losch wrote:
>> Ok, it appears to me that the crashes are really only related to the
>> (default) free-slip boundary conditions. I will wait for Dimitris'
>> packaged config and then check in the fix that seems to work for him
>> (after testing it a little).
>>
>> Martin
>>
>> On May 22, 2009, at 6:21 PM, Dimitris Menemenlis wrote:
>>
>>> Patrick, thanks.  this confirms you were running with free-slip sea
>>> ice boundary conditions when you crashed.
>>>
>>> Martin, I will try to take a closer look and also to package and  
>>> send
>>> you the config that fails quickly ASAP but probably not today.  I  
>>> just
>>> came back from a meeting and need to catch up on  a few things.
>>>
>>> D.
>>>
>>> On May 22, 2009, at 9:04 AM, Patrick Heimbach wrote:
>>>
>>>> <data.seaice>
>>>
>>> Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
>>> Jet Propulsion Lab, California Institute of Technology
>>> MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099, USA
>>> tel: 818-354-1656;  cell: 818-625-6498;  fax: 818-393-6720
>>>
>>> _______________________________________________
>>> MITgcm-devel mailing list
>>> MITgcm-devel at mitgcm.org
>>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>>
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel