This page introduces the easy use of OCTOPUS, for example, how to log in to our system, make the Fortran program "sample.f90", execute it on a general-purpose CPU node, and check a result.

 

1. How to log in

Basically, you must use our system from a remote place. Please connect your PC to our "Frontend node". The Frontend node is the entrance for using the large-scale computing system and you can make a program and give an order to execute a program on the large-scale computing system, checking the result. For basic usage, you do not have to connect directly to the large-scale computing system in addition to connecting to the Frontend node.
 
We adopt 2-factor authentication to OCTOPUS. Please see the following directions:
How to log in

 

2. How to set up the environment

We installed "environment modules" on OCTOPUS and this tool simplifies a setup. For example, if you use general-purpose CPU nodes, please enter the following command:

$ module load BaseCPU

 

3. How to make a program file

To use the computer, you must first create a program.
There are various programming languages in the world, and OCTOPUS supports multiple languages. Major languages available include Fortran, C, and C++.
Programs can be created using the following methods.

 

Making a File on a Frontend node

If you make or edit or modify program files on a frontend node, you can use the editor such as vim. Please see the following:
How to use editor

 

An Example for making a program file

For this example, make a Fortran program "sample.f90" with the vim editor as follows:
 
1. Start vim and make sample.f90

    Please enter the following command on the frontend node.

    $ vim sample.f90

     

2. Edit sample.f90

    Press the i key once to enter edit mode (when -- INSERT -- or -- 挿入 -- appears at the bottom), then enter the following program sample.

    You must often copy or something edit with the vim command. Therefore, you should understand the command gradually.

 

3. Save the program file

    Please press Esc key to exit edit mode, then enter :w and next press Return key.

 

4. Close the vim editor

    Please enter :q and next press Return key.

 

4. Compiling

Previously, you made a sample program file. But you cannot execute this program file on the computing system because your program is only just a character string still. You have to convert your program file to execute the file for the computing system. This process is called "compiling".

 
note:Strictly speaking, this process has various names. Please see the following page for details about compiling:
Program coding

 
You can use C/C++/Fortran compilers developed by Intel and GNU.
 

Example of compiling

Compile the sample program with the intel Fortran compiler.
1. Compile command

    Please enter the following command on the frontend server. (note: In advance, you have to set CPU environment with "module load BaseCPU" command)

    $ ifx sample.f90

     

2. Check the result of the execution

    If you did not see the error and were able to output a.out file, this compile has been successful.
     
    * note

      If you receive an error, please check your program file again. It might have misspellings.

Reference

How to use compiler

 

5. Executing a program file

Program files can be executed in the following two ways:
 

* Batch processing

    In this way, you give the order to the computer to execute all processes together at one time. If you finish giving the order to the computer, you can logout.
    Basically, please execute the program in batch processing.

 

* Interactive processing

    In this way, you execute the process on the computer at real time through a command, etc. This way is very simple.

 

In batch processing, you have to make a "job script" file in which you write the execution process on the computer. The job script is the order to the computing system, and the execution process on the computing system is done with the job script. When you request the job script, the computing system may not execute quickly because many users use the same computing system at same time. If the computing system has many process executing at the same time, you will have you wait to execute. "The scheduler system" manages the order that executes programs on our system.

 

6. Making a job script file

A job script is constructed through the following two steps:

 
1. The head line is #PBS

    This line specifies the resources of the computing system and the environment.

2. The head line is not #PBS.

    The line is a shell script, and this specifies the process executing on the computing system.

 

An example of the job script is the following:

 

These lines beginning from #PBS specify the resources of the computing system and environment. Lines other than the #PBS line specify the process executing on the computing system (Shell script).
 

#PBS –q OCT

    This line specifies which computing system you execute. "OCT" is the job class of OCTOPUS CPU cluster. You specify the job class of the computing system on the job script. Please note that job class varies if you use under shared use or dedicated use.
     
    "OCT" is the job class just for receiving. The system has other job classes for execution. Your job script is sent to the receiving job class and is transferred automatically to the execution job class. Other job classes can be, for example, "OCT-S" to run by sharing the same node as other jobs, "DBG" to run a short job for debugging, etc.
     
    Please see the following page for details about the job class:
    OCTOPUS Job class table

#PBS --group have to specify your group name on OCTOPUS. If you do not forget your group name, you can get group name with "id" command or "groups" command.

$ id
uid=11111(u6a123) gid=22000(ocean) groups=22000(ocean),12345(G12345)

In the above case, your group name is "G12345".
 

#PBS –l elapstim_req=1:00:00

    This line specifies the resources of the computing system. This designation of the example is the elapse time (elapstime_req), which is one hour. This parameter needs to be specified. If your program use is more than the parameter of resources you specified, the calculation will stop. You have to look for a better parameter to fit your program.

 

module load BaseCPU is same command in step 2. You also need to include it in your job script.
 
cd $PBS_O_WORKDIR

    This line means that you need to go to the directory where you were at the time you sent the job script. $PBS_O_WORKDIR is the environment variable that was automatically set at the directory where you were at the time you sent the job script.
     

 

./a.out

    This line is used to execute the file compiled on "4. Compiling".

 

Next, we make a job script file "nqs.sh" with the vim editor.

1. Start vim and make nqs.sh

    Please enter the following command on Frontend node:

    $ vim nqs.sh

     

2. Edit nqs.sh

    Press the i key once to enter edit mode (when -- INSERT -- or -- 挿入 -- appears at the bottom), then enter the following example.
    ※ Please change the part [your group name] to your group's name.

     

3. Save the job script file

    Please press Esc key to exit edit mode, then enter :w and next press Return key.
     

4. Close the emacs editor

    Please enter :q and next press Return key.

 

7. Submit a job script file

You must send a job script file and request the computing system for the executing process. This procedure is called "submit a job script".
 
You can submit a job script with the qsub command. How to submit job script that you create on "6. Making a job script file":

$ qsub nqs.sh

When you enter the above command, you will get the following message:

Request 11111.oct submitted to queue: OCT.

This message means that your job script was correctly received by the computing system. The job script submitted will be attached to a Request ID. If you could not get the above message, your job script might have an error. Please check again!

 

If you want to get the status of a submitted job script, you can use the "qstat" command below:

$ qstat

 

RequestID ReqName UserName Queue Pri STT S Memory CPU Elapse R H M Jobs
--------------- -------- -------- -------- ---- --- - -------- -------- -------- - - - ----
11111.oct nqs.sh user-id O1 0 RUN - 1.01M 0.2 1 Y Y Y 1

You should check "STT" in this information. STT will often display the following two statuses:

 

    QUE -> queued : The program is waiting for execute
    RUN -> running : The program is executing

 

Of course, STT will sometimes display other statuses besides these two . To understand the details of the qsub and qstat commands or other commands, please see below:

 

About scheduler commands

 

8. Check the results

When the computing system finishes executing your program, the result of the execution will be transferred to the Frontend server. If you did not specify the destination of the output on your job script file, the result of the execution will be the standard output. If you did not specify the name of the standard output file and the error output file, the standard output will be named [Jobscript name].o[RequestID] and the standard error output [Jobscript name].e[RequestID]. (Of course, you can change thenames of these files!)
 
In the case of the above example program, you will get files named "nqs.sh.o11111" and "nqs.sh.e11111". How to check these files.
 

1. Check the standard output file

    $ cat nqs.sh.o11111

    An example of the result:

     Hello World!

     

2. Check the standard error output file

    $ cat nqs.sh.e11111

    An example of the result:

    Loading BaseCPU/2025
      Loading requirement: tbb/latest compiler-rt/latest umf/latest compiler/latest
        mpi/latest advisor/latest dal/latest intel_ipp_intel64/latest mkl/latest
        vtune/latest debugger/latest dpl/latest inteloneAPI/2025.2.0

     

If you got the result in the above example, and if the standard error output file did not have an error message, the execution was a success.
 
 

Conclusion

We have finished introducing how to login for novice users. However, if you have any questions, please see below:
Consultation/Inquiry