IT Center Blog

New, powerful hardware for AI integrated

September 9th, 2020 | by
Artificial intelligence is gaining more and more importance - also at RWTH!

Artificial intelligence is gaining more and more importance – also at RWTH!
Quelle: Unsplash

RWTH Aachen University expands its activities in the field of artificial intelligence

In the context of the Excellence Initiative, RWTH Aachen University has procured two powerful systems for complex experiments and analyses in the field of artificial intelligence (AI). On July 10th, two NVIDIA DGX-2 systems were delivered to the IT Center, which were then installed and integrated into the RWTH cluster environment on July 15th.

Endurance test and use cases for the new systems

The NVIDIA DGX-2 are worldwide unique systems, each equipped with 2 Intel Xeon Platinum 8168 CPUs (48 cores in total) and 16 NVIDIA Tesla V100 GPUs. Each system has a peak performance of 2 PetaFlops and offers 81920 CUDA cores and 10240 Tensor computing units that can especially be used for deep learning. The RWTH cluster now includes extremely powerful systems to meet current and future challenges in AI and data science.

NVIDIA DGX-2 -a global innovation that delivers exceptional performance.

NVIDIA DGX-2 -a global innovation that delivers exceptional performance (NVIDIA).

Currently, the two DGX-2 systems are not yet integrated into the batch system, but configured as non-public dialog systems, as our colleagues from CSE-HPC, in cooperation with experts from the AI center, are subjecting the newcomers to some tests in order to develop a suitable operating concept. Among other things, experiments in the areas of Machine Learning (ML) and Deep Learning (DL) are carried out with the help of pyTorch and TensorFlow, which outsource calculations to several GPUs. Furthermore, it will be evaluated whether and how container solutions can facilitate the users’ work with the systems.

What is the goal?

The new systems form the technical basis for the activities of the RWTH AI Center to bundle and strengthen research and activities in the field of AI and machine learning. Thus, researchers of the RWTH Aachen University as well as their project partners will be granted access to the systems based on the access mechanisms for our HPC system CLAIX in order to further advance research in the field of artificial intelligence (at the RWTH Aachen University).

What’s next?

You are certainly curious now. We are also very curious about how things will continue and when the AI infrastructure at RWTH Aachen University can be put into operation. But we will keep you up to date on our social media channels.

If you want to learn more about the DGX-2 systems, please have a look at the manufacturers page. All technical information can be found in the DGX-2 data sheet and the NVIDIA Tesla V100 data sheet.

Responsible for the content of this article are Nicole Filla, Jannis Klinkenberg and Tanja Wittpoth-Richter.

2 responses to “New, powerful hardware for AI integrated”

  1. Tobi says:

    Finde ich super, dass das IT Center uns hier in Sachen GPU Computing mit dieser Hardwareintegration massiv unter die Arme greift!!
    Gibt es schon eine Roadmap, wann das System über den HPC-Zugang erreichbar sein wird?
    viele Grüße

    • Nicole Barbara Filla says:

      Hallo Tobi,

      vielen Dank für deinen Kommentar! Es freut uns sehr zu lesen, dass die Neuerungen Zuspruch bei dir finden. Aktuell liegen uns noch keine Informationen vor, da noch fleißig getestet wird. Sobald wir mehr wissen, werden wir aber auf unseren Social Media Kanälen berichten. Wir freuen uns daher sehr, wenn du uns abonnierst, um auf dem Laufenden zu bleiben.

      Viele sonnige Grüße
      Das IT Center Blog Team