PETSC ascii input files

Hoi,
since last release new mesh files are needed to run the PETSC versions.
For the existing PETSC_GEMS benchmarks for example
"benchmarks/PETSc/TransLay2d" additional ascii msh files were added. I
have trouble run my examples with the binary mesh files (memory
allocation error...probably because I use PETSC 3.4.3 and not version
3.5), but the benchmarks seem to run with the asccii files.
I found on the wiki page how to create binary versions of these files,
but I am not able to create the ascii versions. Adding the flag -asci
does not help.
Can anyone help me with this?
Thanks, Georg

···

--
PD Dr. Georg Kosakowski * Laboratory for Waste Management
Paul Scherrer Institut * 5232 Villigen/PSI * Switzerland
Phone: +41-56-3104743 * Email: georg.kosakowski@psi.ch
homepage: http://www.psi.ch/les/georg-kosakowski

Hi Georg,

You can clone https://github.com/ufz/mesh_partition.git and download the METIS source code
1) either by git submodule command if you can access our private github repository (https://github.com/envinf) as:
       git init
       git submodule update
2) or by downloading it directly via this link: https://svn.ufz.de/ogs/attachment/wiki/PETScPage/metis-5.0.2.tar.gz, and extracting the tarball in the directory of mesh_partition.

Then you can compile the source code via cmake & make (c.f. Page: https://svn.ufz.de/ogs/wiki/PETScPage#no1).

I have successfully tested your example with commands:
partmesh --ogs2metis lag2d
partmesh --metis2ogs -np 4 -n -asci lag2d
where '-n' means 'partitiioning the mesh by nodes', which is for OGS#PETSc.

Best,

Wenqing

···

On 01/09/2015 03:53 PM, Georg Kosakowski wrote:

Hoi,
since last release new mesh files are needed to run the PETSC versions.
For the existing PETSC_GEMS benchmarks for example
"benchmarks/PETSc/TransLay2d" additional ascii msh files were added. I
have trouble run my examples with the binary mesh files (memory
allocation error...probably because I use PETSC 3.4.3 and not version
3.5), but the benchmarks seem to run with the asccii files.
  I found on the wiki page how to create binary versions of these files,
but I am not able to create the ascii versions. Adding the flag -asci
does not help.
Can anyone help me with this?
Thanks, Georg

Hi Wenqing,
thanks a lot, an update of the mesh_partitioning tool solved the problem.
I used an "old" version from 19th December last year (which I believed
was the latest one)!
Greetings, georg

···

On 01/12/2015 03:13 PM, Wenqing Wang wrote:

Hi Georg,

You can clone https://github.com/ufz/mesh_partition.git and download the
METIS source code
1) either by git submodule command if you can access our private github
repository (https://github.com/envinf) as:
      git init
      git submodule update
2) or by downloading it directly via this link:
https://svn.ufz.de/ogs/attachment/wiki/PETScPage/metis-5.0.2.tar.gz, and
extracting the tarball in the directory of mesh_partition.

Then you can compile the source code via cmake & make (c.f. Page:
https://svn.ufz.de/ogs/wiki/PETScPage#no1).

I have successfully tested your example with commands:
partmesh --ogs2metis lag2d
partmesh --metis2ogs -np 4 -n -asci lag2d
where '-n' means 'partitiioning the mesh by nodes', which is for OGS#PETSc.

Best,

Wenqing

On 01/09/2015 03:53 PM, Georg Kosakowski wrote:

Hoi,
since last release new mesh files are needed to run the PETSC versions.
For the existing PETSC_GEMS benchmarks for example
"benchmarks/PETSc/TransLay2d" additional ascii msh files were added. I
have trouble run my examples with the binary mesh files (memory
allocation error...probably because I use PETSC 3.4.3 and not version
3.5), but the benchmarks seem to run with the asccii files.
  I found on the wiki page how to create binary versions of these files,
but I am not able to create the ascii versions. Adding the flag -asci
does not help.
Can anyone help me with this?
Thanks, Georg

--
PD Dr. Georg Kosakowski * Laboratory for Waste Management
Paul Scherrer Institut * 5232 Villigen/PSI * Switzerland
Phone: +41-56-3104743 * Email: georg.kosakowski@psi.ch
homepage: http://www.psi.ch/les/georg-kosakowski