Main dialog options:
Application:
Type the full path to the mpi application here and any arguments. This can be a local or shared location. eg "c:\temp\myapp.exe arg1 arg2" or "\\myserver\myshare\myapp.exe arg1 arg2". The local path must be valid on all the nodes.
Number of processes
Select the number of processes you want to launch.
Run
Launch and start the application.
Break
Interrupt the application
Any hosts
Choose hosts from the list in a round robin fashion.
Hosts
Choose only the highlighted hosts from the list
V button
Add a host name to the list
reset
Reset the list of hosts to the list selected by MPIConfig
Output
The output of the application shows up here. Ctrl+C will copy the output. You can also enter input here that will be sent to the root process zero.
Advanced Options:
No color output
The output will not be color coded according to the rank of the process
No mpi
Launch multiple processes that are not mpi applications - they never make any MPI calls.
Don't clear output on Run
The output in the output window does not get erased when the Run button is clicked
Always prompt for password
Don't use the saved account, prompt for user credentials every time Run is selected.
Redirect output to file
Redirect the output of the mpi application to the specified file. The contents of the file will be deleted each time Run is selected.
Slave process
Specify the path to a second executable to be launched for every process except the root process zero. This is a quick way to launch a master/slave application without using a configuration file.
Environment "var1=val1|var2=val2|var3=val3|...varn=valn"
This will set the environment variables specified in the string before each process is launched. Remember to quote the string so the command prompt doesn't interpret the vertical bar as a pipe command.
Working directory drive:\some\path
This sets the working directory for the launched processes. If this option is not specified the current directory is used.
Drive mappings drive:\\host\share
This option will map a drive on the hosts where the processes are launched. The mappings are removed after the processes exit. This option can be repeated multiple times. example: y:\\myserver\myapps;z:\\myserver\myhome
Use configuration file
Configuration files.
The config file format is as follows:
exe c:\somepath\myapp.exe OR \\host\share\somepath\myapp.exe [args arg1 arg2 arg3 ...] [env VAR1=VAL1|VAR2=VAL2|...|VARn=VALn] [dir drive:\some\path] [map drive:\\host\share] hosts hostA [#procs] [path\myapp.exe] hostB [#procs] [\\host\share\somepath\myapp2.exe] hostC [#procs] ...
Bracketed lines are optional. The # character will comment out a line. You may specify a path to an executable on each host line, thus enabling MPMD programming. If you do not specify a path, then the default is used from the exe line.
Here are two sample configuration files:
exe c:\temp\myapp.exe hosts fry 1 jazz 2
This one shows a more complicated scenario:
exe c:\temp\slave.exe env MINX=0|MAXX=2|MINY=0|MAXY=2 args -i c:\temp\cool.points hosts fry 1 c:\temp\master.exe fry 1 #light 1 jazz 2