First and foremost, the geometry and problem
files are taken from the previous StarCD files written with
Do not forget
to specify the maximum number of iterations for the parallel
computation... In the Navigation Center, click folder "Analysis
Preparation/Running" and then "Set
Run Time Controls" to display the
corresponding panel. Type the maximum number of iteration you
wish and then click "Apply".
In the prostar GUI, click on :
- Files -> Write Geometry
The following panel will appear :
- Choose the STAR-HPC option in
the third box under the file name
- Type the number of processors
you want to use for calculations in the box "No.
- Click on "Apply"
Now the geometry and problem files are
correctly saved. The calculation can be made with parallel StarCD
Star-HPC is run by typing prohpc
at the system prompt. The main Pro-HPC panel then appear on the
screnn as shown on picture below :
Setup is used to
set up the parallel machine (connexion to other
Setup is used to
enter the specific parameters of the problem
(that have not been given in the previous StarCD
Grid Transients is
used to compute moving grids
- Run Setup enables to run all the components set
in the following panels
useful to split up or combine data from the
parallel run for post processing or restarting
- Utilities helps in managing the generated data
- Summary sums up data of the current problem
Click on line "Machine
Setup" : the
panel shown on the left will be activated.
- In the two top boxes, choose the
Memory" and "Shared
Data" : all processors will be
attached to the same large disk and will receive
information from the others.
- In the box "Processor
Characteristics", enter the
name of the machines that are going to make the
- Select the "Queuing
System" option "None"
- In the "Message
Passing System" options, click
and then "Socket"
- In the box "Machine
Specific Settings", choose "Other".
Now specify the architecture of the network :
click on the "Remote Machine Settings ..." button. The
following panel will be displayed :
- In the "Architecture"
window, type LINUX
- In the "Root
Directory", type /logiciels-mfn/STARCD/v3150/mpp/mpich
In the panel "Machine Setup", click on "Go Up" :
you will go back to the Pro-HPC panel. Click on line "Problem Setup" : the following panel is activated :
- Choose the "Basic"
option ine the "Decomposition
Do not forget to
select the processors to be used in the corresponding
table (according to those you have mentioned in the
- When the processors are selected
and written in the right column, click on "Go
Now the simulation is going to run. In the main
Pro-HPC panel, click on the "Run setup" line.
Accept the defaults settings and
click on the "Create"
button. The calculations are going to be made.
Convergence should be achieved within the specified
maximum number of iterations.