GenomeDK scales up: major investment in compute

We’re expanding our compute capacity in Q2 2026 with a significant new hardware investment, strengthening support for both traditional HPC workloads, GPU-heavy simulations, and advanced AI-driven research in all areas of science.

In total, this new investment adds 9,600 new CPU cores to the platform, making GenomeDK a 21,120 core cluster!.

We will install 50 new water-cooled compute nodes, each equipped with 2x AMD EPYC 9655 processors (96 cores per CPU), providing 192 energy-efficient CPU cores per machine. Each machine has 768 GB of memory and is integrated into GenomeDK via our high-speed InfiniBand network, ensuring low-latency communication and high throughput for tightly coupled parallel workloads and data-heavy analysis.

In addition, the expansion includes two GPU nodes, each featuring 4 NVIDIA H200 GPUs with 141 GB of memory, enabling high-performance workloads in areas such as machine learning, large-scale data analysis, and advanced simulation. Many of you have asked for more GPUs and here they are! These GPUs will complement the existing 16 NVIDIA L40S GPUs to provide a variety of GPU options.

This expansion happens only two years after our last compute expansion and demonstrates the need for continuously investing in high-performance computational infrastructure for science, as well as the continued success of GenomeDK’s open and flexible access and usage model.

This upgrade is part of GenomeDK’s on-going commitment to providing robust, scalable, and future-ready compute resources for the Danish research community and SME’s. We’re looking forward to all of the weird and wonderful science that comes out of this!

As always, we are very grateful to the Faculty of Natural Sciences, Aarhus University, the Faculty of Health, Aarhus University, the Central Denmark Region, and the Danish e-Infrastructure Consortium (DeIC) for their continued support and financing.