Politics

GPU Cluster for Artificial Intelligence and Data Science

The TU Ilmenau has put a state-of-the-art GPU cluster into operation at its university computer center. It combines several GPUs, i.e. powerful graphics processing units (GPU), into a cluster in such a way that their computing power is multiplied. The parallel storage and processing of huge amounts of data is the basis for the complex learning processes of machine learning, robotics and artificial intelligence (AI).

Detailausschnitt vom GPU-Cluster mit Metallteilen und leuchtenden Lämpchen Johann Lembach
Detailausschnitt vom GPU-Cluster.

The GPU cluster is part of TU Ilmenau's activities to strengthen its work in the fields of artificial intelligence and data science. With the commissioning of the GPU Cluster in June 2021, the university will take a further step towards making Ilmenau a premium AI location. The President of TU Ilmenau, Professor Kai-Uwe Sattler:

The number of data-intensive applications has really exploded in the course of digitalization and demands ever more powerful computing technology. With such technology, the TU Ilmenau can compete internationally in the fields of artificial intelligence and data science.

Funding by the State of Thuringia

The establishment of the GPU cluster was funded with 630,000 euros by the Thüringer Aufbaubank. Previously, the university's proposal "NVIDIA GPU Cluster Machine Learning and Data Science" had successfully won the competition announced in 2019 as part of the "Guideline of the Free State of Thuringia for the Promotion of Research". The project was jointly initiated by the three Groups of Databases and Information Systems (Prof. Kai-Uwe Sattler), Neuroinformatics and Cognitive Robotics (Prof. Horst-Michael Groß) and Software Engineering for Safety-Critical Systems (Prof. Patrick Mäder) from the Faculty of Computer Science and Automation. The then junior professor Prof. Mäder has since been appointed university professor and head of the Department of Data-intensive Systems and Visualization at the Faculty of Computer Science and Automation, thus consolidating his research work in this field.

State-of-the-art technology for research and teaching

The two-rack GPU cluster is equipped with hardware from NVIDIA, the world's leading provider of graphics processors. In addition to 320 CPU cores, the seven GPU servers contain a total of 36 NVIDIA A100 Tensor-Core GPUs with a total of 1.44 TB of RAM. Connected via a low-latency network with a bandwidth of 200 GBit/s, these GPU servers allow efficient processing of even the largest data volumes. The new GPU cluster thus offers excellent technological conditions for research work on highly ambitious current projects in the future field of artificial intelligence and data science.

These include the project "Engineering for Smart Manufacturing (E4SM)", funded by the Carl Zeiss Foundation, in which innovative methods for the development and operation of machine learning-based assistance systems for smart manufacturing are being researched. In the inter-university project "ThurAI", funded by the Free State of Thuringia, concrete pilot projects for the application and transfer of AI methods in practice are to be initiated and scientifically accompanied, building on the activities of the Thuringian Center for Learning Systems and Robotics (TZLR).

However, TU Ilmenau does not only want to advance the fields of artificial intelligence and data science in research. The new disciplines are also increasingly finding their way into the university's teaching. With the GPU cluster, the university has the ideal prerequisites to meet the requirements for training future generations of AI experts.

With the completion of the Ilmenau location of the "IT Center of the Thuringian Universities" (HS-ITZ) on the campus of the TU Ilmenau, the GPU cluster will move from its current location in the server room of the University Computing Center to the new building along with other state-of-the-art storage and computer server hardware.

[Translate to English:]

[Translate to English:]

[Translate to English:]

[Translate to English:]

[Translate to English:]