[MOAB-dev] Setting tags of non-owned entities

Grindeanu, Iulian R. iulian at mcs.anl.gov
Fri Jul 15 15:19:39 CDT 2016


Sorry, I missed the question about ghosts.
One issue is that ghost entities are created one by one, as they get generated;
If there is no empty space on a sequence, it will create a default sized sequence

the entity space might appear "broken" 
(usually, entity sequences are created at reading time, and they are "contiguous"; if you use "delete" reading options, then some entities might get deleted, and leave open spaces that will be reused by ghosts, later)

When you set a dense tag on an entity, the memory for the tag is allocated the first time an entity from that sequence is set with that tag. 

Iulian


________________________________________
From: moab-dev-bounces at mcs.anl.gov [moab-dev-bounces at mcs.anl.gov] on behalf of Alexander Nikolas Breuer [anbreuer at ucsd.edu]
Sent: Friday, July 15, 2016 12:56 AM
To: MOAB dev
Subject: [MOAB-dev] Setting tags of non-owned entities

Dear MOAB-team,

I have a (hopefully) quick questions on the User Guide:

> Owning Processor: Each shared entity is owned by exactly one processor. This processor has the right to set tag values on the entity and have those values propagated to any sharing processors.

Would like to create an (globally) inconsistent state by setting values (dense tag) of non-owned entities. Don’t need this info on neighboring ranks -> I’ll never propagate that info through pcomm. Are there any technical reasons existing against doing so?

Should I initialize the ghost entities remotely first so that the memory gets allocated properly?

> Storage for a given dense tag is not allocated until a tag value is set on an entity; memory for a given dense tag is allocated for all entities in a given sequence at the same time.


Thanks,

Alex

PS: [6] “MOAB Users Email List.”, moab at mcs.anl.gov. in http://ftp.mcs.anl.gov/pub/fathom/moab-docs/contents.html#references seems to be broken:

> The following message to <moab at mcs.anl.gov> was undeliverable.
> The reason for the problem:
> 5.1.0 - Unknown address error 550-'#5.1.0 Address rejected.'


--
Alexander Breuer
University of California, San Diego
San Diego Supercomputer Center
9500 Gilman Drive MC 0505
La Jolla, California 92093-0505, USA
http://hpgeoc.sdsc.edu
+1 (858) 822-5692





More information about the moab-dev mailing list