Blue Waters

Blue Waters - From Access to Utilization

Read this page first

Step 0. Get a license for Serpent

Register for Serpent through RSICC and deposit your license documentation in the Google Drive software licenses directory.

Step 1. Communication

Have Prof Huff get you approved for Blue Waters!

Step 2. Apply for Blue Waters

You should recieve an email with a link to continue your validation process. Follow the link and fill out the application. You can either receive a virtual token or a physical token. A Virutal token is done through your mobile device, and a physical token looks like this:

alt text

Or, if you chose a soft token, install the RSA SecurID app and follow the instruction sent to you by email.

Step 3. Wait

Wait patiently for the token to arrive.

Step 4. Activating Token

For activating the token, instructions should be contained within the emails recieved. Extra information can be found here.

Step 5. Logging In

Open terminal, type

$ssh [username]@h2ologin.ncsa.illinois.edu

You will then be prompted for your password, followed by an option to select from a list of your registered tokens. After selecting the token you will use (if you only have one token, simply type 1), either enter your pin and token number or approve the login request with Duo, depending on what token type you are using.

tip) you can shortname your ssh command:

Open the terminal

$cd
$cd .ssh
$vim config

write:

        Host [desired_name]
        HostName h2ologin.ncsa.illinois.edu
        User [Blue_Waters_User_Name] 

Then you’d just have to type in

$ssh [desired_name]

For example, you might have named Blue Waters “bw”. In that case, you simply type

$ssh bw

Step 6. Congratz!

You’re in! :)

alt text

tip) running jobs

To start running jobs on Blue Waters, it is useful to look at what other people use as well as the documentation:

  1. Create a [file_name].pbs script (which contains an aprun command). A good way to find example job scripts is to search through the arfc repo for code containing “.pbs”. More information can be found here to understand how these job scripts function.
  2. Use qsub [file_name].pbs to submit the job.

Step 7. Module Loading

Various modules can be loaded and the available modules can be listed using the following commands:

Command Suffix Action
module avail   List all available modules
module avail ` |grep -r [keyword] ` Search module with keywords
module load   Load module onto environment
module swap ` [original] [new] ` replace [orignal] with [new]

tip) ~/.bashrc

If you don’t want to load your modules every time, edit ~/.bashrc and have all your ‘module load/swap’ commands.

$vim ~/.bashrc

after editing the bashrc, you must source it:

$source ~/.bashrc

The Group Directory

/projects/sciteam/bbcc

How to Download Source Code

Because Blue Waters is an HPC machine, it does not have a package manager (like Ubuntu’s apt-get). This is the reason why one has to build from source.

Also, one does not have sudo, which means that the packages should be built either in your home directory(~) or in the group directory (bahg)

Get source codes by:

  1. Github

    git clone [clone_link]
    
  2. wget

    wget [link to source code tar.xz or zip file]
    

    Expand the archive with the tar or the unzip command.

Using Moltres

Navigate to

 /projects/sciteam/bbcc/projects/

and read the readme file located there. It provides useful information to properly setup a pbs script and run Moltres.

Using Serpent

To use Serpent, one can use the command

 qsub serpent_pbs_script.pbs

where an example of a pbs script for using Serpent can be seen below.

 #!/bin/bash
 #PBS -l nodes=4:ppn=32:xe
 #PBS -l walltime=08:00:00
 #PBS -N name_of_Serpent_run
 #PBS -j oe
 #PBS -q normal

 module swap PrgEnv-cray PrgEnv-gnu

 export OMP_NUM_THREADS=32

 cd $PBS_O_WORKDIR
 aprun -n 4 -d 32 /projects/sciteam/bbcc/serpent/src2.1.32/sss2 -omp 32 ./serpent_file > output_file

Step 6 contains a link for more information on pbs scripts. The “module swap” is required in order to run Serpent since it uses the gnu compiler. The “export” line is used to set the thread count which has a default value of 1.

Tip) Cross Sections

When making your Serpent input deck, you can use the following for your cross section data

set acelib "/projects/sciteam/bbcc/serpent/xsdata/jeff312/sss_jeff312.xsdata"
set declib "/projects/sciteam/bbcc/serpent/xsdata/jeff312/sss_jeff33.dec"
set nfylib "/projects/sciteam/bbcc/serpent/xsdata/jeff312/sss_jeff33.nfy"

Tip) Viewing Progress

While you wait for your script to complete, you can run the command

 qstat | grep "name_of_Serpent_run"

which will display how long your script has been running. Once your script is done running, it should return nothing (unless someone else is running a script with the same name as yours).