Skip to main content


Please find some Frequently Asked Questions about LUMI below.

Questions? Please see contact details.

When did LUMI start to operate?

LUMI was installed in two phases: in the first phase, deployed during summer 2021, the CPU partition and storage systems were deployed. The CPU partition was available during autumn 2021.

The second installation in spring/summer 2022 brought LUMI to its full grandeur including the GPU partition.

Both installation phases were concluded with short pilot and acceptance phases.

LUMI was officially accepted on 31 January 2023.

Where do I find high resolution images for media?

Please have a look at the media section for images, videos and contact details. Please remember the copyright information when using the images.

What kind of research questions will be solved with LUMI?

The creativity of European research communities will eventually determine what kinds of problems are solved with LUMI. To give you some idea of the wide range of issues that could be addresses using LUMI’s resources, we foresee that LUMI will help remarkably with the following kinds of research questions:

  • More precise climate models and the interconnection of different climate models: how will living conditions change when the climate is warming?
  • The sequencing and analyzing of full genomes combined with data analysis and correlations to clinical data will bring more light to diseases and hereditary diseases: shedding light on the causes of illness and assisting with personalized treatment and medicine
  • Artificial intelligence (deep learning): analyzing large data sets (simulated and measured) and reanalyzing e.g. in atmospheric science, environmental science, climate modelling, material science and linguistics
  • Self-driving cars and vessels: the study of algorithms related to these with previously unprecedented computing power
  • Social sciences: large-scale data set analytics from social networks and the modelling of different phenomena
  • LUMI will also have a channel for urgent computing. This, so called “director’s share”, type of allocation will allow to grant some of LUMI’s resources on an ad hoc basis for time- and mission-critical simulations. This kind of simulations might be, for example, related to national or EU security, or some massive disturbance affecting the partner countries, for example a large epidemia or pandemic disease.

Can sensitive data be processed with LUMI?

Sensitive data processing with LUMI is not yet possible. LUMI aims to include a solution that enables to bring sensitive data into LUMI and to process it safely there. LUMI will build a solution to safely transport/stream sensitive data for processing in LUMI but it will not be hosting sensitive data archives or repositories. The solution is based on data encryption, virtualization and containers, in addition to sufficient isolation in the system for jobs and file systems. This is under preparation within the LUMI consortium in collaboration with the CSC sensitive data platform and the ELIXIR community.

Who are able to access LUMI?

Half of the LUMI resources belong to the EuroHPC Joint Undertaking, and the other half of the resources belong to the participating countries i.e. the LUMI Consortium countries.

Each consortium country has a share of the resources based on the country’s contribution to the LUMI funding. The shares for each of the countries are allocated according to local considerations and policies – so LUMI will be seen and handled as an extension to national resources.

The LUMI shares belonging to the EuroHPC JU will be allocated by a peer-review process (comparable to that used for PRACE Tier-0 access). In addition, up to 20% of the EuroHPC resources will be available to industry and SMEs.

A researcher affiliated with one of the LUMI Consortium countries or a company which has its headquarters in the LUMI Consortium countries are able to apply for LUMI resources via the LUMI Consortium country’s national share and, in addition, via EuroHPC’s technical and scientific peer-review process.

By partnering with European research groups and/or companies, the EuroHPC resources are available to non-European research projects as well. (Note that the Principle Investigator for any projects that apply for time on LUMI need to be based in the EU or an associated country.)

The resources of LUMI are allocated for projects in terms of three different pools: GPU-hours, CPU-hours, and storage hours. All users of the system have the access to the whole system in accordance with batch job policies, i.e. there are no dedicated hardware for any of the partners.

LUMI also has a channel for urgent computing. This, so called “director’s share”, type of allocation allows to grant some of LUMI’s resources on an ad hoc basis for time- and mission-critical simulations. This kind of simulations might be, for example, related to national or EU security, or some massive disturbance affecting the partner countries, for example a large epidemia or pandemic disease.

Can users from LUMI consortium countries apply for resources from the other pre-exascale supercomputers?

Yes, users from LUMI consortium countries can apply for resources from the other pre-exascale supercomputers, since due to different architectures, some applications will run best on another system.

How fast is LUMI?

LUMI supercomputer in Kajaani, Finland, has a sustained performance of 380 petaflops (HPL).

LUMI’s computing power is equivalent to 1,5 million modern laptops.

How big is LUMI?

LUMI takes nearly 300m2 of space, which is about the size of two tennis courts.

The weight of the system is nearly 150 000 kilograms (150 metric tons).

CSC data center has reserved 2200 m2 of floor space for the installation and the data center is expandable up to 4600 m2.

What is an exaflop?

Exascale supercomputers are the next big step in high-performance computing. One exaflop means processor computing power corresponding to 1018 floating point calculations per second or one trillion calculations per second.

How will LUMI handle big challenges of the future e.g. possible pandemics?

LUMI will for example speed up the development of new medicines and the simulation of different viruses, and finding cure to them will be much faster.

If there will be new pandemics or other similar phenomena where time- and mission-critical simulations are related to national or EU security, or some massive disturbance affecting the partner countries, LUMI will have a “director’s share” type of allocation which will allow to grant some of LUMI’s resources for urgent computing. Also LUMI’s national allocation pools may be used for urgent operations which have a high societal impact.

A supercomputer requires supercooling, how do you handle this?

LUMI uses warm-water cooling, which enables its waste heat to be utilized in the district heating network of the city of Kajaani, and thus replaces heat produced by fossil fuels.

The waste heat from LUMI that can be used in Kajaani’s district heating network. Annually, hundreds of households in the city of Kajaani are heated with LUMI’s waste heat.

The need for cooling is also reduced by the fact that the outdoor temperature and, as a result, the building’s thermal stress is much lower in Kajaani than what it would be in Southern Europe, for instance.

The carbon footprint of the supercomputer is further decreased by the fact that it can be placed in an existing building near existing infrastructure.

Does LUMI mean something as a word?

LUMI is an abbreviation of Large Unified Modern Infrastructure. The LUMI consortium countries are Finland, Belgium, the Czech Republic, Denmark, Estonia, Iceland, Norway, Poland, Sweden, and Switzerland.

LUMI means snow in Finnish and Estonian. Snow has also inspired the look of the LUMI data center (see the media section for high. res. images).

Three important things which everyone should know about supercomputers?

A supercomputer is a computer with very high performance and among the fastest in the world. One criteria for a supercomputer is regarded to be the top500 list of the world’s fastest computers.

The lifetime of a supercomputer is typically from four to six years.

If compared to cars, supercomputers are like Formula 1 cars compared to normal passenger cars. Supercomputers perform mathematical operations millions of times faster than regular computers. Supercomputers are needed in doing research in various data and computing intensive disciplines such as drug modelling and in climate research.

Where can I look for HPC training that will enable me to work efficiently on LUMI?

There are multiple training options available for beginners, intermediate and advanced HPC users. In order to help you navigate through the broad spectrum of training opportunities, we have collected the links to the most notable training portals for you. Training offered by LUMI user support team can be found at LUMI´s website under Events and Training. In addition, see PRACE training portal, CSC´s training calendar, EuroCC training portal, High-Performance Computing in Europe to find suitable HPC training for the needs of your research.

Is LUMI supporting quantum computing?

I am using an application that currently runs on Nvidia GPUs. Can I run it in LUMI-G?

Not out of the box, many applications written specifically for Nvidia GPUs will not work in LUMI-G without modifications. Our goal is to provide and maintain a list of software packages that have been tested. For now, you can check the applications listed in AMD’s InfinityHub. The applications which are labelled “AMD Instinct MI200” are most likely to work on LUMI-G.

I am developing an application that currently runs on NVIDIA GPUs. How can I port it to AMD GPUs in LUMI?

If your application uses CUDA, you can convert it to HIP. There are tools (e.g. “hipify”) that can partially automate most of this process. OpenMP GPU offloading (C/C++ and Fortran) and OpenACC (only Fortran) are also supported by the compilers that will be available on LUMI. In general, the porting procedure contains many steps. AMD provides a porting guide. We also have published some blog posts, such as porting CUDA codes and OpenMP offloading. Further information on porting applications to AMD GPUs will be available in the LUMI documentation.

My application uses CUDA Fortran: can it also be translated to HIP?

Yes, but it works differently than with CUDA. The kernels written in CUDA Fortran have to be converted to C/C++ CUDA and AMD provides an interface (“hipfort”) to call HIP kernels from Fortran.

I am developing an application that currently runs only on CPUs. What are the choices I have for when porting to AMD GPUs in LUMI?

We recommend that you look into the following programming languages and frameworks: HIP (which is similar to CUDA), OpenMP with GPU offloading, OpenACC (but only for Fortran), and for C++, the Kokkos, RAJA, SYCL and ALPAKA frameworks.

What kind of help can I get with porting my application?

You can contact LUMI User Support for general information on where to get help in your specific country, or if you have technical questions about LUMI-G. You can also contact the EuroHPC Competence Center in your country. They can often assist with applying for grants from e.g. the European Union for porting applications.

LUMI will organize regular AMD GPU Hackathons, where you can will be able to get hands-on help from experts from AMD and HPE.

LUMI runs a porting program where software developers can apply for help with porting to AMD GPUs. There are regular calls issued, typically 1-2 times per year.

How to cite LUMI in a scientific publication?

Please see the acknowledgement page for information how to cite LUMI in a scientific journal. The page includes information about how you may refer to LUMI in your presentation.

My question was not answered in the FAQ. Where to find help?

If you have further questions, please contact our user support team by filling out a form on the user support page.