|
|
|
# Parallel Grid Generation Example
|
|
|
|
|
|
|
|
The code for this example is located in `examples/GridGen`
|
|
|
|
|
|
|
|
## Brief
|
|
|
|
|
|
|
|
This example creates simple cubic or prismatic mesh
|
|
|
|
|
|
|
|
## Description
|
|
|
|
This example may run in both serial and parallel modes with `NP` processes.
|
|
|
|
|
|
|
|
The virtual IJK-structure with user defined sizes is used to cover the unit cube `[0,1]^3`.
|
|
|
|
In parallel mode this structure is split into `NP` rectangular superblocks.
|
|
|
|
Local IJK-structures are created in each superblock separately.
|
|
|
|
Depending on user defined choice in each IJK block either one cubical cell, or two triangular prismatic cells are created.
|
|
|
|
In the latter case the direction of diagonal plane is changed in staggered order.
|
|
|
|
|
|
|
|
## Arguments
|
|
|
|
```
|
|
|
|
Usage: ./GridGen ng nx ny nz [output.[p]vtk]
|
|
|
|
```
|
|
|
|
- First parameter `ng=3` stands for Prismatic generator and `ng=4` for Cubic one.
|
|
|
|
- Parameters `nx`, `ny` and `nz` are the IJK-structure sizes.
|
|
|
|
- Last optional parameter – output file name, default: `grid.vtk` for serial runs and `grid.pvtk` for parallel runs.
|
|
|
|
|
|
|
|
## Running example
|
|
|
|
If you compiled INMOST with `USE_MPI=ON` you can generate and save mesh in parallel. The following line will create 32x32x32 cubic grid using 4 processes:
|
|
|
|
```
|
|
|
|
$ cd examples/GridGen
|
|
|
|
$ mpirun -np 4 ./GridGen 4 32 32 32 /tmp/grid-32-32-32.pvtk
|
|
|
|
Cubic Grid: 32 x 32 x 32
|
|
|
|
Processors: 4
|
|
|
|
Mesh generator time: 0.14828
|
|
|
|
Save to file "/tmp/grid-32-32-32.pvtk" time: 0.060786
|
|
|
|
```
|
|
|
|
If you have ParaView installed, you can open pvtk file:
|
|
|
|
```
|
|
|
|
$ paraview --data=/tmp/grid-32-32-32.pvtk
|
|
|
|
```
|
|
|
|
You can view the following tags:
|
|
|
|
- `P_OWNER_PROCESSOR` – partitioning to processors
|
|
|
|
- `GLOBAL_ID` – global cell ID number |
|
|
\ No newline at end of file |
