Skip to content

Introduction to clusters

Learning outcomes

  • Have an understanding of the terminology of the clusters and the providers.
  • Understand that after login, one is on a login node
  • Have included HPC cluster name in Zoom username, e.g. [Dardel] Sven Svensson

This section is a beginner's guide to clusters, and provides general information about computer clusters like Tetralith, Dardel, Alvis, Cosmos, Pelle, Kebnekaise and other HPC systems, but is not directly focused on any of them.

What is a cluster?
  • A computer cluster consists of a number of computers (few or many), linked together and working closely together.
  • In many ways, the computer cluster works as a single computer.
  • All academic clusters in Sweden is so far using Linux.

  • Generally, the component-computers are connected to each other through fast local area networks (LANs).

  • The advantage of computer clusters over single computers, are that they usually improves the performance (and availability) greatly, while still being cheaper than single computers of comparable speed and size.

cluster

Where do I store my files
  • HOME folder, i.e. path ~/ or do pwd as soon as you log in.
    • This is where you get when logging in.
    • Put files with no connection with a specific project here.
    • Limited space for storage and number of files.
  • Project folder, usually something like /proj/<project-number> but varies a lot among clusters!
    • Place for data and results.
Want more information?

Do the exercises below!

The HPC clusters for academic research in Sweden

NAISS-wide

NAISS resources can be applied for by all Swedish academic researchers.

NAISS resources Centre provider Use case Availability
Alvis C3SE (Chalmers) AI Move your data NOW!
Arrhenius HPC NAISS General From Summer 2026
Arrhenius SENS NAISS Sensitive Late autumn 2026
Bianca UPPMAX (UU) Sensitive Throughout 2026
Dardel PDC (KTH) General Throughout 2026
Tetralith NSC (LiU) General Move your data NOW!

Arrhenius will replace ALL present NAISS-level systems during 2026

Local clusters

Local resources can be applied for by PIs affiliated to or with an agreement with the local university provider.

Local resources Centre provider Use case Availability
COSMOS LUNARC (LU) General End date unknown
COSMOS SENS LUNARC (LU) Sensitive End date unknown
Vera C3SE (Chalmers) General End date unknown
Pelle UPPMAX (UU) General End date unknown
Kebnekaise HPC2N (UmU) General End date unknown
Maja UPPMAX (UU) Sensitive From 2027

Warning

Dardel is unfortunately not available this week due to a service stop.

Transferring your project data to Arrhenius

If you are a present user of the Alvis or Tetralith systems, you should have received information of how you transfer your data to Arrhenius. If you missed it, here are some pages:

Exercises

Exercise 1: Rename yourself

Please change your Zoom name to start with your HPC cluster you will use today, as [name_of_hpc_cluster] your_name, e.g. [Arrhenius] Sven Svensson.

Why the name of the HPC cluster first?

It will make Zoom sort your names in a way most convenient for the course, which helps the logistics of the course.

Exercise 2: Make yourself acquainted with some terminology by reading the boxes below

What are Nodes, Cores, CPUs, and GPUs?
  • A node is the name usually used for one unit (usually one computer) in a computer cluster.
  • Each node may have one or two central processing units, or CPUs, each with many cores.
  • Each core is a single processor able to handle a single programmed task.
  • Computer nodes can also have GPUs (graphical processing units) in addition to the CPUs.

gpu node cpu node

  • Nodes are usually arranged in racks and are connected via a communication network.
  • Nearly all high-performance computer clusters use the Linux operating system.
  • Normally, clusters have some sort of batch or queuing system to handle scheduling of jobs, like Slurm (the Simple Linux Utility for Resource Management).
What are the Login and compute nodes?

  • The user logs in to login nodes via Internet through ssh or Thinlinc.

    • Here the file management and lighter data analysis can be performed.

  • The compute nodes have to be used for intense computing.

    • Any longer, resource-intensive, or parallel jobs must be run through a batch script or an interactive session.

Exercise 3: On what type of node do you connect to when you log in?

Answer
  • Login node
  • Remember to not do any compute intensive tasks here!

(Optional) Exercise 4: Learn more about clusters

(Optional) How is a job run on a computer cluster? What is a batch system?
(Optional) Which programs can be run effectively on a computer cluster?
  • The main way an HPC cluster lets you speed up computations is by letting you execute several tasks in parallel.
  • In other words, a problem must somehow be split into many tasks to gain any speed-up.
  • More info Introduction page of "Running jobs on HPC systems"
  • Another use case is when you handle much data or memory.
(Optional) What is a Supercomputer? Is it the same as a Cluster?
  • A supercomputer is simply a computer with a processing capacity (generally calculation speed) several orders of magnitude better than a typical personal computer.
  • For many years, supercomputers were single computers with many CPUs and usually large volumes of shared memory---sometimes built specifically for a certain task.
  • They have often been custom-built machines, like Cray, and still sometimes are.
  • However, since desktop computers have become cheaper, most supercomputers today are made up of many "off the shelf" ordinary computers connected in parallel.

A supercomputer is not the same as a computer cluster, though a computer cluster is often a supercomputer.