Wageningen University & Research (WUR) has given Anunna, its high-performance computing (HPC) facility on campus, a major investment boost. Anunna is part of a European network of supercomputers. The investment will enable the computer to handle more data and facilitate wider deployment of Artificial Intelligence (AI). This will make this campus facility future-proof, which is important, because Anunna is useful for all areas of research conducted at WUR. Thanks to a grant from RegioDeal Foodvalley, the HPC is available for shared use via WUR's Shared Research Facilities.
More graphical computing power for data science and AI applications
The Wageningen HPC was in need of an overhaul, and a sum of €2.1 million has been earmarked for it, explains Alexander van Ittersum, Product Owner Compute & Platform Solutions and Research IT Solutions. “This will ensure the long-term security and reliability of the system. But it also no longer had sufficient storage capacity. In addition, we will be updating and increasing the graphics processing power, allowing AI applications to be deployed to much better use in the cluster.” Anunna is used to calculate climate models and map genomes, for example. “That happens at the Animal Breeding and Genomics, Environmental Research, and Plant Sciences research groups. But we are also seeing more and more application of data science in all fields, which requires computing power. Useful applications can be found in all research groups. The use of the HPC will also have an impact on key research themes such as Connected Circularity and the Protein Transition.”
The additional graphics power is needed because of the high demand by all research groups for large-scale image processing capacity. “An example is the analysis of video footage produced by drones. Conducting such research using a public cloud can be an expensive affair. This investment will allow us to keep everything in-house and improve our cost efficiency.”
Anunna and Animal Breeding and Genomics
Martien Groenen, head of the Animal Breeding and Genomics chair group, has been using Anunna with his group from day one, so for some ten years now. “We needed a lot of memory back then to map genomes, and so this genome research was soon transferred to the Anunna server. Breeding companies also showed an interest, and they too now store data on the HPC as part of various partnerships. The quantity of terabytes is only increasing, so the expansion is needed. Both computing power and storage space are very essential for the work we do. Of course, you don’t want that storage space in the cloud, particularly if it involves sensitive data.”
Machine learning and AI are increasingly applied in the Animal Sciences department’s Next Level Animal Sciences programme. “We conduct a lot of measurements on animals using sensors. There is a lot field data to record and analyse. When I started, it was mostly lab work in our department. Today, the majority of our PhDs analyse datasets.”
Network of supercomputers and partnerships
Anunna is part of a network of supercomputers in Europe. The Netherlands’ national supercomputer, Snellius, is located in Amsterdam and falls under the national Tier-1 system. Switzerland is home to a European supermodel (Tier-0). Anunna is a Tier-2 regional system.
“There is still too little cohesion between the HPC facilities of the various universities. We want to foster more mutual cooperation. For major projects, the Tier-1 system in Amsterdam is required. Often, this requires reinstalling the software and describing the code. We want to improve the cooperation in this area to make that process easier.”
Groenen recognises the problem: transferring data is time-consuming, and there are often waiting times to use Snellius. “That makes researchers more reluctant to use it. That’s why we are glad that our own HPC is being expanded. It will save us from having to move all that data around. It also makes us a more interesting partner for other parties in the Netherlands and in Europe. We have a clear position in the supercomputer landscape.”
According to Van Ittersum, the expansion may well attract more universities and companies. “However, our aim is not to compete with other Tier-2s. It is all about cooperation. It’s for good reason that Shared Research Facilities is the owner of the HPC: our goal is indeed to share this facility.”
Supercomputer also available for students
Anunna also plays a role in education. “Data science and AI are being deployed in more and more fields,” says Van Ittersum. “We think it is important to introduce students to Anunna, because they need to gain sufficient experience with this technology. Whatever the industry they end up in, it is bound to play a role.” A new professor in AI will be starting soon, and the HPC will play a role there too.
Anunna will also play a role at the Wageningen Data Competence Centre (WDCC). At the WDCC, education, research, value creation, infrastructure and data management come together. The WDCC facilitates developments in these areas. The development of AI is high on the agenda of Wageningen University & Research. The appointment of three AI professors within the Wageningen Data Competence Centre confirms this importance.
Availability and use
The supercomputer will be available for use from October 2023. The new components for the HPC will be installed in one of WUR’s two data centres. These new machines are water-cooled instead of air-cooled, which is much more efficient. Systems with water cooling operate up to 80% more energy-efficiently. The HPC thus scores highly for energy efficiency, which is important both for environmental reasons and in light of the current energy prices.
Not everyone is aware of Anunna’s capabilities, or they only think they are, Van Ittersum says. “There is plenty of support available; we have a team dedicated to doing just that. They sit down with the researchers to discuss the opportunities for research using Anunna. We also provide HPC training, from basic to advanced courses.” These are scheduled for next September and October. The exact dates are shown in the calendar.