-
Notifications
You must be signed in to change notification settings - Fork 0
Execution
After compiling with the desired settings, HiFiLES can be run with the following commands.
Something to keep in mind:
The input file contains the address of the mesh file. The HiFiLES
executable looks for the mesh file relative to the current directory. Then, if only the mesh file name appears in the input file, both input and mesh files need to be in the same directory, and the HiFiLES
executable needs to be called from the directory containing both files.
##Single CPU/GPU
Assume the HiFiLES
executable is in the directory /EXEC_DIR
, and the input and mesh files are in the directory /TESTCASE_DIR
. Suppose the input file is named input_file
. The following command runs the simulation:
$ /EXEC_DIR/./HiFiLES /TESTCASE_DIR/input_file
##Multiple CPUs/GPUs
The compilation should be made with MPI support
and, if using GPUs, with CUDA support
. See the Installation instructions to learn more.
###In a personal computer
$ mpirun -n 8 /EXEC_DIR/./HiFiLES /TESTCASE_DIR/input_file
Change the number 8
to the number of processors you would like to run on.
###In a computer cluster
If running in a cluster with a queue, refer to your cluster manager. If you do not have a queue system and are able to access each node directly, you will need to create a machine file. In this file, include the network name of each of the processors you would like to run on. For example, let us say we want to run HiFiLES in a two-node cluster, using 4 CPUs (or 4 GPUs) in each node. Suppose node 1 is named compute-0-0
and node 2 is named compute-0-1
. The machine file would include the following:
compute-0-0
compute-0-0
compute-0-0
compute-0-0
compute-0-1
compute-0-1
compute-0-1
compute-0-1
Suppose the machine file is called mfile
in directory /MFILE_DIR
. You can run the simulation with the following line:
$ mpirun -n 8 -machinefile /MFILE_DIR/mfile /EXEC_DIR/./HiFiLES /TESTCASE_DIR/input_file
In this case, the number of processors you want to use (8
in the example) needs to match the number of CPUs (or GPUs) you have mentioned in the mfile
.
- Home
- User Docs
- User Tutorials
- Meshing for HiFiLES
- [(2D) Flat Plate]
- [Laminar]
- [Turbulent]
- [Local Time-stepping]
- [(2D) Laminar Circular Cylinder]
- [(2D) NACA 0012, 5° AoA, Re = 60,000, M = 1.6]
- [(2D) SD7003 airfoil, 4° AoA, Re = 22,000]
- [(3D) Method of Manufactured Solutions]
- [(3D) SD7003 wing section, 4° AoA, Re = 10,000]
- (3D) Taylor-Green Vortex, Re = 1,600
- Developer Docs
- [Code Structure]
- Building and Running with Qt Creator
- Adding files to the project
- [License]
- [Contact]
- Release Notes