[MITgcm-support] Zeros in inactive tiles in the regional model

[Martin Losch]

Hi Yue,

I am not sure, if I understand the issue properly, but if you use “blank tiles” via “blankList” in ‘eedata’, then the inactive tiles in this list will not be computed (no processor assigned), but still part of the output field (just zeros, as you have found out yourself). This is useful, when you try to mask out parts in the middle of the domain, that do not contain any ocean (in your case maybe some tiles in the eastern part of the domain over Scandinavia or Russia), but if all grid points from j = 1 to 900 are supposed to be masked, it may make more sense to redefine the grid to start only at your j=900.

Martin

> On 19. Mar 2025, at 14:52, Matthew Mazloff <mmazloff at ucsd.edu> wrote:
> 
> Hi Yue
> 
> The MITgcm doesn’t do any compression. 
> You will have to compress (e.g. zip) after the MITgcm I/O.
> 
> Matt
> 
> 
>> On Mar 18, 2025, at 9:17 PM, WU Yue <vywu at ust.hk <mailto:vywu at ust.hk>> wrote:
>> 
>> Dear MITgcm team,
>>  
>> I’m running a regional model based on LLC1080 grid (the domain of Face 1 shown in Fig.1). The original version uses sNx = 90, sNy = 90 with 320 active tiles (also 320 processors used) for the entire domain, which takes around 6 days to run 2 experimental years. To focus more on a smaller region and for resource-saving, I masked southern part of the domain (new Face 1 in Fig.2) then uses sNx = 90 and sNy = 60 with 317 active tiles. The new version runs around 4 days for 2 years, which seems good to me that reducing grid number per tile to 2/3 of the origional version, the running time should speed by 1/3 in time.
>>  
>> However, I noticed that the output data size remains the same between the old and new versions. It appears to have values in the region with active tiles, but for those in inactive tiles the model keep assigning to zeros (see two Figs). In that sense, the new version only saves in time but doesn’t save the storage. 
>>  
>> Does anyone know whether the processors are used to computing active tiles whilst allocating zeros to inactive tiles? Or is the model do the allocation? (or other?) Are these zeros will definitely exist?
>>  
>> Best wishes,
>> Yue
>>  
>> <Fig1.png><Fig2.png>_______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org <mailto:MITgcm-support at mitgcm.org>
>> https://urldefense.com/v3/__http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support__;!!Mih3wA!DkYkZRo09RtuSD2Fp2ylcFkslD1nCFQBPtLAMtxWg3j99aN9PTymIqZ58WLpAy3VYA1RWTq9QDlntA$
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org <mailto:MITgcm-support at mitgcm.org>
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20250319/73e47af9/attachment-0001.html>