Getting Started

This page lists information about how to get started with running applications on the HPC machines.

Getting Access

To gain access to the HPC machines, a formal application is needed. The process is explained at our page for how to apply for user account.

Logging In

Logging into the machine involves the use of Secure Shell (SSH) in a terminal. SSH login is available natively to Linux or Mac OSX (or any UNIX-based systems). On Windows machines an SSH client is needed, for example, PuTTY.

    ssh <username>@<machinename>

Replace <username> with your registered username and <machinename> with the specific machine name.

The machine names are:

• fram.sigma2.no - Fram
• saga.sigma2.no - Saga
• login.nird.sigma2.no - NIRD. See NIRD for more about NIRD.

First time you log in, and every time there is a change in the ssh-server, OS or hardware on the system you log in to, you will be asked to confirm the ssh-server fingerprint. Typical message is:

The authenticity of host 'fram.sigma2.no (158.39.114.72)' can't be established.

The procedure then is to check the stated ECDSA key fingerprint with the one printed here: ssh SHA256 fingerprint.

Remote desktop

The Fram, Saga and Stallo systems provides a remote desktop service. See here for tutorial and details.

Quickstart: Use a VNC client to log into desktop.fram.sigma2.no:5901 or desktop.saga.sigma2.no:5901 (for information on Stallo see here http://stallo-gui.uit.no). A web based remote desktop service is also available https://desktop.fram.sigma2.no:6080. Access to these services are blocked outside the Norwegian Research network, e.g. only accessible from UNINETT and partner institutions. (workarounds are described in the tutorial)

Development environment

Abel, Stallo, Fram and Saga run CentOS Linux distributions as operating system. The machines can run C/C++ or Fortran OpenMP and MPI applications, and depending on the machine, various open-source and third party applications. The Programming Environment page has more information about third-party applications.

Unix CLI

It is expected that the user is familiar with command-line interfaces (CLIs), but for those who are not familiar with commands, the UiB and UiT pages have several tutorials on the CLI environment:

• Introduction to Unix CLI - UiB

To copy files from your machine to the HPC machines, use SSH File Transfer Protocol (SFTP) or Secure Copy (SCP). They are available as command-line tools for Linux and MacOS X but Windows users need to download a separate SCP or FTP client, such as WinSCP.

For example, to copy projectfiles.tar.gz to the home directory of myusername to Fram, type:

scp projectfiles.tar.gz myusername@fram.sigma2.no:

The Transferring Files page has more information about transferring files to Fram.

Modules

To keep track of the large number of different pieces of software that is typically available on a shared HPC cluster, we use something called a software module system. This allows us to have many different versions of compilers, libraries and applications available for different users at the same time without conflicting each other. By default when you log in to the cluster you will get a clean environment with nothing but standard system compilers and libraries. In order to make your favourite software application available to you, you need to load its module into your environment, which is done using the module command

module [options] [module name]

Some of the more common options include:

• avail - list the available modules
• list - list the currently loaded modules
• load <modulename> - load the module called modulename
• unload <modulename> - unload the module called modulename
• show <modulename> - display configuration settings for modulename

More details on the module system, as well as a list of currently available software modules, can be found in the Software section.

Running Applications

The HPC machines provide compute nodes for executing applications. To ensure fair access to the resources, the HPC machines run applications as jobs in a queue system, which schedules the tasks and process to run on compute nodes. Abel, Stallo, and Fram use the Slurm queue system.

A job is described by a batch script, which is a shell script (a text file) with SBATCH options to specify the needed resources and commands to perform the calculations. All batch scripts must contain at least the following SBATCH options:

#Project name
#SBATCH --account=<project_name>

#Max running time. The wall clock time helps the scheduler assess priorities and tasks.
#SBATCH --time=<wall_clock_time>

To submit a job, use the sbatch command followed by the batch script's name:

sbatch <scriptfile>

See the "JOBS" section in the menu for documentation about jobs and the queue system.

Account information

Information on available CPU-hours in your accounts:

cost

More Information

The menu on the left list the pages with more information about developing and running applications.