Space Technology Company ICEYE praises LUMI user support: expert guidance step by step

For ICEYE, ensuring the necessary support for deployment posed a fundamental challenge. Top performance and a very competitive price would not have sufficed if Tapio Friberg would have had to fight with the deployment process for weeks.

Image: An artist’s rendition of an ICEYE satellite. Copyright: ICEYE.

Ease of use and user support. These are the two points for which Tapio Friberg, ICEYE Machine Learning Engineer, gives special thanks.

– And the very best experts supported us in the deployment process, Friberg says.

The Finnish space technology company ICEYE manufactures satellites and uses them to deliver real-time radar images from anywhere, at any time of the day, and under any conditions.

ICEYE makes the radar signal computationally visible and produces analytics of what the image shows. The company has imaged and investigated, among other things, floods and wildfires.

In his work, Tapio Friberg uses the LUMI supercomputer’s GPUs, i.e., LUMI-G, for product development of automated image interpretation. Training machine learning algorithms takes several weeks, so it is important to keep the data readily accessible.

– Data storage has been a particularly useful feature when working with CSC, Tapio Friberg says.

The data is kept safe until the prototype cycles, which may sometimes take up to a week, have been completed.

CSC’s support helped match software libraries with LUMI

When Friberg chooses new tools and operating methods, the factors affecting the choice are not only the price and technology but also how many working hours the deployment takes at ICEYE. When using a supercomputer, first, you must always make some effort to match your own materials and software with the supercomputer.

In the case of LUMI, according to Friberg, the hardware and memory were perfect for ICEYE’s needs, but lack of support in the deployment phase could have become a bottleneck.

– If I were to work on something on my own for a month without knowing whether it would bring results or not, I would probably spend my time on something else, Friberg said when he described the selection process at the Supercomputing Innovation Summit 2023 in Kajaani in October.

However, Friberg was not left on his own. ICEYE did not have any experts familiar with supercomputers, but with the help of CSC experts, Friberg found the right software and was able to match his own software library with LUMI.

– A CSC expert literally showed me step by step what to do and what not to do. This allowed me to match my software library with LUMI in a couple of weeks. After that, I was able to move on to doing things that are part of my core competence, he adds.

LUMI’s operating costs enable ambitious testing

It is part of ICEYE’s business philosophy to act differently from others. Even in Friberg’s own work, continuous experimentation and search for new ideas are extremely important. New approaches open up new markets.

If experimentation is fast and cost-effective enough, it is possible to run new tests just to see what happens. LUMI’s operating costs enable this kind of testing.

– The fact that the costs are significantly lower than expected enables more ambitious experimentation and thus better results, Friberg said in Kajaani.

20% of LUMI’s capacity in Finland is reserved for Finnish companies. Companies can use the LUMI supercomputer in their research, development, and innovation activities. ICEYE was one of the companies to seize the opportunity in 2023.