[MITgcm-support] Computation time

[chris hill]

Hi Goran,

  There is a description in the online docs in the parallelism section 
6.18.3. I'm surprised you haven't read that bit!

  There is also an example in MITgcm/util/exch2/matlab-topology-generator

  Both pieces use the cube grid to illustrate what is done.

Chris
Göran Broström wrote:
> Hi  Chris
> 
> How do you use tiling to remove land points. Are there any code examples 
> where this is used or any written ducumentation?
> 
> Göran Broström
> 
> 
> 
>> Hi Eli,
>>
>>  The code computes over land points in general.
>>
>>  For configurations with large numbers of points e.g. 1/6 and higher 
>> resolution globally you can use tiling to throw away regions with no 
>> land.
>>
>>  For smaller configurations the overhead of skipping land points in do 
>> loops actually costs more than doing null computations, so that the 
>> shortest time to solution comes with computing that includes land.
>>
>>
>>  What size of problem are you running and how long are your runs taking?
>>
>> Chris
>>
>> eliyahu biton wrote:
>>
>>> Hello everyone!
>>>
>>> I just want to be sure, is the computation time for 'land grid point' 
>>> (those indicated by zero values within the bathimetry file)  the same 
>>> as the 'sea grid point' ?
>>>
>>> And if so what can I do to reduce this computation time of these 
>>> 'land grid point' ?
>>>
>>> Thanks Eli
>>>
>>>
>>> _______________________________________________
>>> MITgcm-support mailing list
>>> MITgcm-support at mitgcm.org
>>> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
>>>
>>
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
>>
>