On this page we summarize what we do more long-term. We also have a page about
There are currently two different machines located at UiT. Fram is a
supercomputer with over 1000 individual nodes, each with 32 cores, capable of
1.1 PFLOP/s. A large portion of our job is to make sure that this machine
operates at peak capacity.
We also have one part of NIRD located in Tromsø. NIRD is a storage centric
machine, focusing on fulfilling needs for research that require storage and
computing on data where a single dataset can be several terabytes in size.
Although all NIRD operations are in the process of being moved to Lefdal Mine
Datacenter, we are still maintaining the "old" NIRD instance here. The future
of this system is still being discussed.
For both machines, typical tasks include replacing failing components,
rebooting machines that has stopped working, and upgrade or replace parts of
the machine when new capabilities or improved performance is available.
- GPU for HPC:
Our team is dedicated to enhancing GPU usage for both NVIDIA (Saga and Betzy) and AMD (LUMI-G) GPUs.
Part of our duties involves crafting comprehensive tutorials that facilitate understanding and optimal
utilization of both NVIDIA and AMD GPUs. We also actively participate in, and organize, training events
with the aim of fostering GPU operation skills among users. Furthermore, we provide consistent user support,
helping users navigate the complexities and solve any issues they may encounter in GPU usage.
- Research software engineering:
Within the research software engineering group at UiT,
we interact closely with users to facilitate their software development process. Our services range from
assisting in enhancing your scripts or code, to helping in organizing your code for optimal efficiency.
We also provide guidance on how to effectively share your code with others. Additionally, we offer support
aimed at improving the performance of your code to ensure it runs faster and more efficiently.
- Software team: We contribute to software installations for NRIS clusters.
We use EasyBuild as the main software installation management tool.
- Internal tooling:
We develop internal tools for HPC systems that improve the user and support experience:
- dusage shows storage quota and usage on the national HPC systems
- sonar collects data on jobs resource usage
- jobgraph uses data from sonar and plots info from jobs
Outreach and on-boarding
- We help new users getting started through the
small scale exploratory work projects.
- On this page we describe how you can get access to a test
account or a course account.
- We are planning to start later this year with regular (weekly) office hours
to provide a contact point to discuss ideas and issues with user communities.
With other NRIS centers and international collaborations we develop and deliver courses and training events:
- We provide support for NRIS users and improving usability.
- We help new users getting started through the small scale exploratory work projects
- Extended user support (support that requires few days of work)
- Advanced user support (support that requires few weeks of work)
- Improving the documentation
- Improving internal policies to make things easier for users