[MITgcm-devel] SERIOUS performance problems with latest code

[Baylor Fox-Kemper]

The CVS head just started up.  As suspected, it also is getting about  
20% efficiency on 16 processors.

   -Baylor


On Oct 31, 2006, at 11:35 AM, Dimitris Menemenlis wrote:

>> are you really running latest(!) code on columbia?
>> If not, which is your checkout or update date
>> (maybe look in doc/tag-index ).
>
> No, not latest.  I am running something just prior to  
> checkpoint58q_post, right after addition of  
> DISABLE_MPI_READY_TO_RECEIVE option.  God the code moves fast!
>
> So I am including none of the post checkpoint58q_post I/O changes yet.
>
> D.
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel