Getting Started
What is an HPC Cluster
An HPC (High-Performance Computing) cluster is a collection of powerful computers (called nodes) that are connected through a fast internal network and work together to perform complex computations. These systems are designed to handle tasks that are too large, too slow, or too data-intensive for standard desktop or laptop computers.
At sciCORE, the HPC cluster provides researchers across the University of Basel and partner institutions with access to advanced computing resources, optimized for large-scale simulations, data analysis, machine learning, and more. The cluster runs a Linux-based operating system and is managed through a job scheduler that efficiently distributes computing tasks across available nodes.
Using a cluster allows users to:
-
Run computations in parallel across many CPUs or GPUs
-
Handle large data sets that do not fit in standard workstation memory
-
Perform long-running tasks reliably and reproducibly
-
Share resources efficiently among multiple research groups
When to use an HPC Cluster
You should consider using the SciCORE HPC cluster when your computing tasks exceed the capabilities of your local machine. Typical use cases include:
-
Large-scale simulations (e.g., fluid dynamics, molecular dynamics, physics-based modeling)
-
Data-intensive workflows (e.g., genomics, transcriptomics, large image processing)
-
Machine learning and AI that require multiple GPUs or significant compute power
-
Batch processing of many jobs that can run in parallel
-
Long-running tasks that must be completed without interruption
-
High-memory workloads that demand more RAM than available on a personal computer
In general, a cluster is not intended for tasks that are interactive, lightweight, or faster to execute on a local machine (e.g., editing documents, plotting small datasets, or running simple scripts).
Additional information
However, at sciCORE we offer some interactive capabilities that be launched within the scheduler framework. See Interactive computing at sciCORE.
If you’re unsure whether your project is a good fit for the HPC environment, sciCORE offers user support to help you evaluate your needs and get started with the appropriate resources.
HPC Cluster of the University of Basel: sciCORE
sciCORE currently operates a high-performance computing infrastructure, divided in different environments tailored to specific scientific needs. The infrastructure is composed of near 200 InfiniBand and Ethernet 100G-interconnected nodes, and around 13500 CPU cores, providing 70 TB of distributed memory and a high-performance (GPFS) cluster file system with a disk-storage capacity of over 11 PB.
The sciCORE cluster is updated on a regular basis to match the growing needs in computational research. Nowadays, almost 30 million of CPUh are consumed per year by our 800 users, summing up to more than 14 million of jobs run per year.
Account at sciCORE
Warning
You need an UniBasel account and e-mail to access sciCORE.
If you already have a UniBasel account and e-mail address
Make sure you are on the university network or VPN, go to https://support.scicore.unibas.ch/ and login with your Uni Basel email credentials. Then click on Account creation and fill out the form.
Info
Undergraduate students can also obtain a sciCORE account, please contact the sciCORE team for details using the same steps described above.
If you do not have a UniBasel acccount and e-mail address
The following applies only for external users (non-UniBasel):
Request your account through this link. Select “Register by using your personal e-mail address” and use your current institutional address (e.g. xxx@fhnw.ch) for registration. Make sure to indicate an authorized approver, the list of approvers is provided here. See the Access for Associates and Guests page for more information.
Info
Although sciCORE is listed as an approving body, we only create accounts in exceptional cases that cannot be handled by the faculties.
If you need an account on sciCORE without having an affiliation to the University of Basel or associated institutes, please contact the sciCORE team directly (scicore-admin@unibas.ch).
In general, the university rules for usage of IT apply – for more information visit the guidelines.
Important
Your user account is personal and must not be shared. You are personally responsible for the usage of your account. If you need to share large-scale data with your colleagues, please contact us. Do not share your password in any circumstance!
Details on your sciCORE user account
Your sciCORE username and password is identical to your Uni Basel user account (not the e-mail address, but the shorter associated username).
Your standard home directory path ($HOME) in the sciCORE systems will be:
where “groupname” is the user name of your PI (with some few exceptions).
Note
The home path of some user accounts created before 2014 may have a different structure.
Note
By default, your login shell is set as /bin/bash, which is the recommended shell. If you want to work with a different default shell, please contact the sciCORE team.
Warning
If you leave the university, your user account will be deactivated automatically at the same time as your Uni Basel e-mail address. However, your data will be preserved for a limited time after deactivation. See Leaving sciCORE for more details.
Logging into sciCORE
The sciCORE HPC environment is accessible exclusively via SSH (secure shell).
Warning
Login nodes are meant to be used for interactive development and low-burden testing. It is strictly forbidden to execute demanding (in terms of CPU, RAM, or I/O) codes/scripts/commands directly on the login nodes. Heavy calculations (longer than a few minutes, with more than one core, and more than a few GB of memory) MUST be submitted as cluster jobs to the queue (see the SLURM user guide). Any attempt to circumvent this will be detected and interrupted, and in case of repeated offense, sanctioned. In the case of file/data management tasks, such as compressing/decompressing, searching for files, or synchronizing directories, users must use the specifically-dedicated login node for data transfers and heavy I/O operations: transfer12.scicore.unibas.ch.
Access from within the Uni Basel network
The SSH command to access login nodes is:
Where <username> is the username that sciCORE assigned to you. Usually the same as your short name for you university e-mail account.
If you are using graphical applications requiring X11 tunneling (f.e. interative plotting), you will need to explicitly activate the X11 transport:
However, if interative graphical usage is important for you (f.e. Jupyter or R notebooks), a much more efficient and comfortable approach is to use interative computing via Open On Demand. Also note that demanding graphical-based operations in the login node are forbidden and will be killed without any warning.
Access from outside the Uni Basel network
The sciCORE environment can only be accessed from within the Uni Basel network. If you are located somewhere else, you will need to access using a VPN client. Visit the ITS instructions here for more information.
Recommended terminal clients for non-Linux computers
Windows
Windows Subsytem for Linux
Currently, Windows offers a native implementation of a Linux subsystem (WSL) that allows users to install their preferred Linux distribution under the umbrella of a Windows operating system. This solution provides the most complete and robust experience for Linux-based developers and users. See the Microsoft website for details on how to install it.
Terminal emulators
MobaXterm provides many useful capabilities, including a drag-and-drop FTP server to transfer files from/to the sciCORE server. If you don’t have administration rights on your local computer, you can still download the portable version of MobaXterm, which requires no installation.
Putty is a very lightweight SSH client, which in combination with Xming provides X11 support.
Macintosh / MacOSX
Recent versions of MacOSX do not necessarily support X11 by default. The free library XQuartz can be used to provide X11 functionality if your version of MacOSX lacks this functionality.
Line terminators issue in ASCII files
Text files generated under Windows (and sometimes MacOSX) have different line terminators than in standard Unix. The command to convert text files from Windows format to Unix format is dos2unix textfile.txt If you notice strange behaviors when accessing within the sciCORE Linux system a Windows or MacOSX generated file, we recommend using the command above.
Training
sciCORE provides regular training sessions focused on the usage of the cluster, best practices, and specific courses of highly-demanded computational methods. You can check our current training offer here. See Training and references section for more resources.