[MOAB-dev] WriteNCDF and others

[Rajeev Jain]

You are right, some hexes (the ones that are initially loaded to procs-instances) are in both materials.
Below are the sets from individual procs (serially written), followed by write parallel with 2 procs having file 1 and file 2.
mbsize -ll file 1:

EntitySet 41:
Global id = 1000
  Vertex 76 - 78
  Hex 17 - 32
  EntitySet 40 -- (this has hex 1-16 belonging to file 1)
  Parent sets: (none)
  Child sets: (none)
  Sparse tags:
    MATERIAL_SET = 1000 
    CATEGORY = Material Set
  Adjacencies:
(none)

mbsize -ll file 2:
EntitySet 41:

Global id = 1
  Vertex 67 - 69
  Vertex 103 - 105
  Vertex 142 - 144
  Hex 17 - 64
  EntitySet 40 -- (this has hex 1-16 belonging to file 2)
  Parent sets: (none)
  Child sets: (none)
  Sparse tags:
    MATERIAL_SET = 1 
    CATEGORY = Material Set
  Adjacencies:
(none)

parallel writing using 2 procs each having the above two files (EntitySet 39 is repeated in both materials?):
EntitySet 42:
Global id = 1
  Vertex 67 - 69
  Vertex 103 - 105
  Vertex 142 - 144
  Hex 17 - 64
  EntitySet 39
  Parent sets: (none)
  Child sets: (none)
  Sparse tags:
    MATERIAL_SET = 1 
    CATEGORY = Material Set
  Adjacencies:

EntitySet 45:
Global id = 1000
  Vertex 196 - 198
  Hex 81 - 96
  EntitySet 39
  Parent sets: (none)
  Child sets: (none)
  Sparse tags:
    MATERIAL_SET = 1000 
    CATEGORY = Material Set
  Adjacencies:
(none)

I'm using the  "PARALLEL=WRITE_PART" option to write_file.

Rajeev



>________________________________
>From: Jason Kraftcheck <kraftche at cae.wisc.edu>
>To: Rajeev Jain <jain at mcs.anl.gov>
>Cc: "moab-dev at mcs.anl.gov" <moab-dev at mcs.anl.gov>
>Sent: Thursday, June 2, 2011 2:59 PM
>Subject: Re: [MOAB-dev] WriteNCDF and others
>
>On 06/02/2011 02:53 PM, Rajeev Jain wrote:
>> There is something wrong going on my .h5m file has 48130 hexes and .exo
>> (created using mbconvert) file has 107907 hexes. I'm sorry, this may be due
>> to some previous steps in 'parallel' coregen.
>> In this case 2 NEUMANN_SET (s) are lost.
>> The files are on MCS workstation : 
>> /homes/jain/pll/acis/MeshKit-acis/rgg/test/simple_hexflatcore
>> 
>
>The most likely cause of this is that there are hexes that exist in multiple
>material sets.  The exodusII file will contain a copy of a hex for each
>material set ('block' in exodusII terms) that contains it.
>
>- jason
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/moab-dev/attachments/20110603/31d5c7e2/attachment.htm>