[mpich-discuss] Mpi-collective abort:

[Gus Correa]

Hi Thejna

The MPI distributed with the PGI compilers is MPICH-1, which is
too old, no longer maintained, and doesn't play well with current Linux
kernels.

I suggest that you build MPICH2 from source code and use it.
You should test if it works correctly with the cpi.c program,
before you move to CCSM, which is a very complex program.

For specific help on how to build CCSM3 or CCSM4, you may want
to ask questions to the NCAR CGD Bulletin Board:

http://bb.cgd.ucar.edu/

Most problems with CCSM are *not* related to MPI,
but with the building scripts.

I hope this helps.

Gus Correa


Thejna Tharammal wrote:
>  Hi,
> I'm using MPICH version 1.0.5, on linux cluster- with scientific linux
> (2.6.18-128.1.6.el5, x86_64)
> 
> Thank you,
> Thejna.
> ----------------original message-----------------
> From: "Rajeev Thakur" thakur at mcs.anl.gov
> To: mpich-discuss at mcs.anl.gov
> Date: Thu, 26 Aug 2010 15:54:32 -0500
> -------------------------------------------------
>  
>  
>> Which version of MPICH2 are you using and on what platform and OS?
>>
>> Rajeev
>>
>> On Aug 26, 2010, at 8:22 AM, Thejna Tharammal wrote:
>>
>>> I'm trying to run a model [CCSm, MPMD with 5 executables] on Intel xeon
> 64 bit 
>>> cluster with 6 nodes(3Gz each,ethernet) , I use Mpi2 distributed with 
>>> pgi-7.2-5 , the model compiles without errors, but during run it shows,
>>>
>>>
>>> 1: (cpl_domain_compare) domain #1 name: atm contract domain
>>> 1: (cpl_domain_compare) domain #2 name: lnd contract domain
>>> 1: (cpl_domain_compare) domain #1 name: ocn contract domain
>>> 1: (cpl_domain_compare) domain #2 name: ice contract domain
>>> rank 0 in job 1 xxxx_35825 caused collective abort of all ranks
>>> exit status of rank 0: killed by signal 9
>>>
>>> What could be the reason for this,
>>>
>>> Thank you,
>>>
>>> Thejna.
>>>
>>> _______________________________________________
>>> mpich-discuss mailing list
>>> mpich-discuss at mcs.anl.gov
>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>> _______________________________________________
>> mpich-discuss mailing list
>> mpich-discuss at mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>
> 
> 
> _______________________________________________
> mpich-discuss mailing list
> mpich-discuss at mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss