HPC Getting Started

From TAMUQ Research Computing User Documentation Wiki
Jump to navigation Jump to search


Account Request

To register for an account on RAAD / RAAD-2, please follow System Accounts Guide provided at Research Computing at TAMUQ Website.

Account Activation

​When a HPC Cluster account is created, the user will be informed via email. In this notification, the user shall also be provided with an assigned username to be used specifically to log in to the HPC Cluster. Before the account can be accessed for the first time, however, the user must reset their current windows domain password (see procedure below). Subsequent to that, the user's windows and HPC Cluster passwords shall always remain automatically synchronized (although for technical reasons the usernames must remain different).

Changing your password if you are outside TAMUQ

To reset your TAMUQ domain account password:

  1. Go to https://apps.qatar.tamu.edu/
  2. You will be asked to install CITRIX client, proceed with installation.
  3. Once done login on this portal using your TAMUQ provided user name and password
  4. Once you are logged in, Select Desktops from Upper tab.
  5. Click on Complete Desktop from new window.
  6. A new window will open and you will be logged into a virtual machine. Select Alt + Ctrl + Del from upper tab.
  7. Select change password, you will be asked for new password.
  8. Enter a new password in the "New Password" field.

This new password will be your unified password for both the windows as well as the HPC Cluster accounts. Note that you will not be able to log in to the HPC Cluster for the first time until you complete these steps.

Changing your password while in campus at TAMUQ

To change your TAMUQ domain account password:

  • Log in to your office PC or laptop using your windows username and password
  • Once logged in, press the ctrl-alt-del key combination
  • Click "Change Password" in the "Windows Security" box
  • Enter your current windows domain password in the "Old Password:" field
  • Enter a new password (8-12 characters) in the "New Password:" field
  • Enter the new password again in the "Confirm New Password:" field

This new password will be your unified password for both the windows as well as the HPC Cluster accounts. Note that you will not be able to log in to the HPC Cluster for the first time until you complete these steps.

Accessing RAAD

To Start computation on RAAD HPC Cluster, first you need to login with the credentials provided and then if required transfer source code/data to RAAD. Below guide explains logging in process for Linux, Mac and Windows OS.


Linux & MAC

Logins

In any Linux environment, you can use SSH to login to RAAD.

  1. Open Terminal and type "ssh user_name@raad.qatar.tamu.edu" without quotes and hit enter.
  2. Provide your password and hit enter.
  3. The prompt at terminal will change to [user_name@raad ~]$ indicating you are successfully logged in.

Data Transfers

To transfer data from Linux Host to your home directory in RAAD, you can use Secure Copy. For example;

  1. scp local_data user_name@raad.qatar.tamu.edu:$path

Detailed information on Secure Copy can be found at SCP


Windows

Logins

A number of software packages may be used to login to raad over the network. For users attempting logins from within our (TAMUQ) building, we recommend the X-Win32 program. Those who do not have access to an X-Win32 license (it is commercial software) may use the freeware Putty program for access or MobaXterm to launch GUI applications remotely. Users accessing HPC Cluster from outside the TAMUQ building may find X-Win32 to be sluggish due to excessive network latency, and may be better off trying putty as well. Putty provides a text-only, non-graphical interface. X-Win32/MobaXterm in contrast has the advantage of allowing the user to launch GUI-based (graphical) programs on raad while having them display on the user's monitor. Both X-Win32 and putty are Windows software packages.

MobaXterm

Our recommended way of connecting to HPC Cluster is using MobaXterm software which is free for personal use. It allows user to connect to HPC Cluster via SSH and can launch GUI applications remotely. Detailed installation and configuration information can be found here Note that, provided network latency is not an issue, MobaXterm is the preferred and recommended method for remote access to Raad as it allows for running of GUI based applications remotely on the HPC Cluster.

X-Win32

This is paid software. Users can buy their own license or may use the trial version. Instructions for installing and configuring X-Win32 are available at X-Win32 Installation

Alternatively, users can use the text-terminal based PuTTY application (freely available) for their access to Raad. Any other SSH client would do just fine as well.

Putty

It is an opensource software with which you can connect to HPC Cluster remotely via SSH. It is a text based utility and you will not be able to launch GUI sessions remotely. Users who are not in TAMUQ and experience latency over network, can use this utility to effectively launch session to HPC Cluster. Installation and configuration steps are detailed here

Data Transfers

Network R: Drive

Users using Desktop or Laptops provided my TAMUQ already have R: drive mapped on their machines. You can read/write and access all contents of your RAAD Home directory using R: drive.

SCP Clients

To transfer your data from Windows storage to HPC Cluster, you can use FTP clients for Windows. We have documented two opensource client below;

  1. WINSCP Install & Configuration Guide
  2. FileZilla Install & Configuration Guide

Submitting your First Job

Once you are connected to RAAD HPC Cluster you can start submitting jobs. But before this it is important to identify nature of your compute job with respect to parameters below;

  1. Expected Compute Time
  2. Amount of Memory Required
  3. Number of CPU's required
  4. Compiler or Library which can be required by your program

Once you have identified each parameter, you can write your job. RAAD uses Portable Batch System to schedule user jobs. Further information about this is explained in Batch processing