1 Reply Latest reply: Apr 19, 2017 2:11 AM by Marc Ribet RSS

Seed Value?

F A

Hi,


I'm working on a lab manual and I'm supposed to assign a seed value in the simulation run, but I could not find it anywhere. Someone said that it is not in the edition I'm using (Riverbed Modeler Academic Edition 17.5) anymore.

 

Can someone tell me if there is any other possible way to assign a seed value?? or do I need to install an older version of the program and where can I find it??

 

Thank you in advance.

 

P.S. Attached is a pic of the manual.

  • Re: Seed Value?
    Marc Ribet

    I do not have the Academic Edition but the seed is set when configuring the simuation (DES -> Configure/Run DES).

     

    If that option has been removed (this is possible), you can maybe try to run the simulation from command line.

     

    1) Run the simulation from the UI and yes it is using 128 by default for the seed value

    2) Look at Help -> Session Log and you should find a line using op_runsim, for example:

     

    op_runsim -mem_optimize true -net_name TCP-slow_start_and_congestion_avoidance1 -noprompt -handle_exception true -ef TCP-slow_start_and_congestion_avoidance1-DES-1 -DESinfo TCP-slow_start_and_congestion_avoidance1-DES-1 -exec_id 1 -mgr_id 1 -opnet_port 61013 -parent_pid 17672 -tfile_dir_cache ignore -pipe_stdin

     

    There is a .ef file (environment preferences) : ef TCP-slow_start_and_congestion_avoidance1.

     

    Edit that TCP-slow_start_and_congestion_avoidance1.ef file (it should be found where the project is installed on your disk) and change the value for "seed".

     

    Save that file and copy/paste the command line from the session_log in a DOS shell from the directory where op_runsim is install, f.e:

     

    C:\OPNET\17.5.A\sys\pc_amd_win64\bin

     

    If the simulation is able to complete, go back to the UI and use the menu File -> Manage Model Files -> Refresh Model Directories (or close/reopen Modeler if not available).

     

    Note that the seed is usefull when you want to run multiple simulations to compute confidence intervals for the results. The results should be very close when using a different seed so I am not sure it really matters for your lab (unless this one wants to focus on computing confidence intervals using different seeds).