On this page we summarize what projects we currently work on. Not everything
here is finished but we think it is useful if we communicate with the research
community about what is brewing in the HPC group. For the more long-term work,
please check the "about us" page.
We are now easier to contact. First via our new email (
email@example.com) and second
through our weekly in-person office hours.
We have also established a new HPC newsletter you can subscribe to receive news
about training, events, support sessions, and hardware updates. See here for an
explanation on how to subscribe (and unsubscribe).
jobgraph are tools we develop to make
it easier for users to check the resource usage of their jobs graphically in
the terminal. These tools will also help us to collect resource usage data over
time and create benchmarks for future hardware procurements.
Job script generator
We have developed this nice page which can help to create job scripts:
On-boarding new users and user groups
We are helping two new user groups moving their compute work and storage to
We are helping with testing the https://jupyter.uit.no/ which is a
JupyterLab instance available to anyone with a UiT user
We are implementing the FitSM - the lightweight IT
Management framework in the NRIS
collaboration. This organisational model is developed for federated
organisations like ours to clearly define roles, introduce workflows and
processes like Service portfolio management, Incident management or Continual
service improvement management (14 processes in total).
Some of our goals with introducing the FitSM are: to clearly define
responsibilities and make them process-bound and well-documented, to keep track
of the ongoing projects and daily work, to have the complete and up-to-date
overview of services and service components, etc. At the moment the Change
Management, Continual Service Improvement Management, Incident & Service
Request Management processes are implemented and taken in use, for the rest of
the FitSM process the work is underway.
We contribute to the European Environment for Scientific Software
Installations (EESSI) project. EESSI (pronounced
as "easy") is a collaboration between different European partners in HPC
community. Goal is to provide:
- a shared stack of scientific software installations
- uniform user experience: same software available on local laptop or
workstation as in the cloud or HPC site
- no performance penalty: binaries optimized for your architectures
In the backdrop of the digital world, where each byte and command you input brings forth a dazzling array of results on your screen, there's a giant machine ticking away. With the ability to solve complex computations at an unprecedented speed, Fram isn't just a collection of circuits and wires. Fram, like every powerful machine, has its own set of challenges that we need to overcome to keep it running smoothly.
Lately, we've been tackling an issue that might not immediately come to mind when you think about supercomputing: condensation. Yes, even in the high-tech world of computational wizardry, we grapple with some decidedly low-tech problems.
Our recent weather in Tromsø has been unusually warm. While most of us enjoy the pleasant climate, Fram has been feeling the heat. Specifically, the warmer weather has been causing condensation on the water-cooling system that Fram uses to maintain optimal operating temperatures. The biggest challenge? The metal connectors that link the hoses bringing in chilly 7°C water.
So how do you solve a problem like condensation in a supercomputer? The solution came in the form of insulating foam. By adding this foam to the couplings, we've been able to limit the exposure of the cold metal to the humid summer air, thus preventing condensation from forming.
We admit, the solution sounds remarkably ordinary. By applying this layer of insulation, we've hopefully thwarted a potentially damaging condensation issue, safeguarding Fram's performance and ensuring that your data processing and computational tasks go uninterrupted.
This process may not have been as glamorous as discovering a revolutionary algorithm or creating an innovative application, but it's precisely this type of problem-solving that ensures Fram runs optimally 24/7. It's the unglamorous, mundane tasks, like insulating connectors, that can often have the biggest impact on system performance.
We share this experience to shed light on the often overlooked aspects of maintaining a supercomputer like Fram. Every component, from the sophisticated processors to the humble hose connectors, plays a vital role in the supercomputer's function. And sometimes, tackling a 'simple' problem like condensation requires just as much attention as the complex computational tasks that Fram performs daily.
We are committed to maintaining the integrity and performance of Fram, whatever the weather. We hope this insight into our work brings you a deeper understanding of the complex tapestry of tasks that go into managing a supercomputer.