Exporting large mesh

Coreform Cubit Meshing
#1

I’m exporting a particularly large mesh, Coreform Cubit 2022.9, its a mesh with 112m hexes in it, its meshes ok, but when exporting to exodus, we get a crash.

Sweeper Mesh Gen: 189.840 seconds
Sweeper Output: 100.500 seconds
Volume 1 meshing completed using scheme: sweep
Generated 112128000 hexes
Meshing time: 384.300000 

Finished Command: mesh volume 1

%>cubit.cmd('set exodus netcdf4 off')

Finished Command: set exodus netcdf4 off

%>cubit.cmd('set large exodus file on')

Finished Command: set large exodus file on

%>cubit.cmd('export mesh "/home/dc-davi4/rds/rds-ukaea-ap001/dc-davi4/mesh/mesh_112m.e"')


Using ExodusII API version 8.03 and DB version 8.03 to export "/home/dc-davi4/rds/rds-ukaea-ap001/dc-davi4/mesh/
mesh_112m.e".
Writing Exodus file
0   |    |    |    |    50   |    |    |    |  100
coreform_cubit: /home/ci/cfconda/conda-bld/netcdf_1657893597849/work/mesh/cubitclaro/cubitcomp/cubit/libs/netcdf
-c-4.7.4/libsrc/posixio.c:1211: ncio_spx_get: Assertion `extent < 2147483647' failed.
**********...Interrupt Detected. CUBIT Exiting...

Any advice? Should I go to netcdf instead?

Thanks

Andy

#2

I should note, using netcdf it worked fine.

#3

Hi Andy,

The default for Exodus should be netcdf4 on. The setting is retained for legacy purposes, but all new Exodus files are generated using netcdf.

My first thought was 112M elements isn’t very big at all. It’s not big because netcdf handles the large format files.

Thanks,
Karl