Towards the end of July, four servers with four NVIDIA H100 GPUs each arrived at SDU. The GPUs have been added to the DeiC Interactive HPC (UCloud) system, meaning that the DeiC Interactive HPC service now offers 32 H100 GPUs, which are available immediately to all researchers in Denmark.
The NVIDIA H100 GPUs are among the most sought-after hardware by AI companies and researchers around the world. They offer unparalleled speed and memory for AI workloads, as well as a new architecture that enables faster data transfer between multiple GPUs.
AI and machine learning have become increasingly relevant for several research areas, including robotics, precision medicine, digital humanities, bioinformatics, material science, drug discovery and fintech – just to name a few.
“The arrival of 16 new H100 GPUs at SDU comes at an opportune moment and will help us accommodate the growing interest in this type of machine,” says Martin Lundquist Hansen, team leader for research infrastructure at the SDU eScience Center. “It seems that people are starting to get to know and experience their powerful performance, leading to a high utilization of the first 16 H100 GPUs that were added to UCloud in January this year.”
The Novo Nordisk Foundation and the Export and Investment Fund of Denmark (EIFO) recently invested in Denmark’s first AI supercomputer, Gefion, which will be equipped with more than 1500 H100 GPUs. The DeiC Interactive HPC service will offer the perfect combination of user-friendliness and state-of-the-art hardware for researchers who want to test and optimize their code for larger machines such as Gefion.
The machines bought by the DeiC Interactive HPC consortium, which besides SDU also consists of Aarhus and Aalborg University, are the HPE Cray SC XD665 and they were purchased via ATEA.
AI and quantum apps on UCloud
Two applications for AI, with a particular focus on large language models (LLMs), were recently launched on UCloud: Chat UI and Text Generation. These apps work best with powerful GPUs such as the NVIDIA H100. You can find them under the new category “Artificial Intelligence” in the App Store on UCloud. Over the next period, we plan to add more apps for AI to this category. Currently, NVIDIA NeMo and NVIDIA Triton Inference Server are scheduled to be deployed very soon.
Besides being uniquely well-suited for AI, the H100 GPUs work excellently with the two advanced quantum computing applications on UCloud: the NVIDIA CUDA Quantum Platform and the NVIDIA cuQuantum Appliance.
Try them out now!
We invite all interested researchers to try out the GPUs and experience their power and performance for themselves. You can find the H100 GPU resources on UCloud (select machine type u3-gpu) or contact your local Front Office for further guidance.
One of the four servers with four NVIDIA H100 GPUs before it was installed in SDU’s server room.
