[MITgcm-devel] global_sum_rl

[Martin Losch]

Hi Jean-Michel,

I am sorry, I did look closely enough, everything OK.

What's te deal with seaice_lsr/mpi/no-mpi? Can I contribute? I will be  
on vacation starting tomorrow, so I can offer to help you really only  
today.

M.

On Mar 26, 2009, at 4:16 PM, Jean-Michel Campin wrote:

> Hi Martin,
>
> I prefer the GLOBAL_SUM_TILE_RL, because it allows to get
> the same results (when GLOBAL_SUM_SEND_RECV is defined)
> with mpi and without, providing the same tile domain decomposition.
>
> But I don't see any GLOBAL_SUM in seaice_lsr.F, and regarding
> cg3d.F, I switched to GLOBAL_SUM_TILE_RL a while ago.
>
> Jean-Michel
>
> PS: I am making some progess on the seaice_lsr differences
> between mpi / no-mpi.
>
> On Thu, Mar 26, 2009 at 03:40:40PM +0100, Martin Losch wrote:
>> Hi all,
>>
>> I forgot why we do not use global_sum_rl in cg2d anymore? Does it  
>> make
>> sense to use the other variant (global_sum_tile_rl) in other  
>> components,
>> eg. in cg3d and seaice_lsr?
>>
>> M.
>>
>> _______________________________________________
>> 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