Otus - Technical Description
In the second half of 2025, PC2 has inaugurated the new Lenovo HPC system, called Otus, which is delivered, installed and maintained by pro-com Datensysteme GmbH. The computer system is available for nationwide use as part of the National High Performance Computing (NHR) initiative and supports the Atomistic Simulation research priority, among others. The current Noctua 2 system at PC2 is thus complemented by another HPC system with approximately twice the computing power.
In the Top500 list 06/2025, Otus is on position #164 (CPU partition) and on position #255 (GPU-H100 partition).
Otus ranks 5th on the Green500 list from June 2025 and also on the current list from November 2026.
The overall setup of the Otus supercomputer with insights into the operation of the cluster from a hardware, software and facility perspective is described in the Otus paper.
Here you can find the inauguration video of Otus.
Apply for free access to our supercomputers for research purposes. Details here.
| Manufacturer | Lenovo |
|---|---|
| Number of Processor Cores | 142,656 <coming soon> additional 12,864 |
| Number of GPUs | 108 <coming soon> additional 204 |
| Total Main Memory | 593.3 TiB <coming soon> additional 77 TiB |
| Floating-Point Performance | CPU: 11.44 PFLOPS DP Peak (8.42 PFlop/s Linpack) GPU: 7.45 PFLOPS DP Peak (4.65 PFlop/s Linpack) <coming soon> additional GPU: 8 PFLOPS DP* Peak (480 PFLOPS TF32-Peak w/ sparsity) |
| Cabinets | 11 racks - all components with direct liquid cooling, 2 racks - all processors with direct liquid cooling, 6 racks - air cooling <coming soon> additional 4 racks |
| Compute Nodes | 636 nodes, each with • 2x CPU, AMD EPYC 9655 "Turin" • 2x 96 cores, 2.6 GHz up to 4.5 GHz • 768 GiB main memory <coming soon> additional 20+ nodes |
| Compute Nodes (Large Memory) | 48 nodes, each with • 2x CPU, AMD EPYC 9655 "Turin" • 2x 96 cores, 2.6 GHz up to 4.5 GHz • 1,536 GiB main memory <coming soon> additional 6 nodes |
| Compute Nodes (Huge Memory) | <coming soon> 2 nodes, each with • 2x CPU, AMD EPYC 9655 "Turin" • 2x 96 cores, 2.6 GHz up to 4.5 GHz • 3,072 GiB main memory |
| GPU Nodes | 27 nodes, each with • 2x CPU, AMD EPYC 9655 "Turin" • 2x 96 cores, 2.6 GHz up to 4.5 GHz • 768 GiB main memory • 4x NVIDIA H100-94GB <coming soon> 48 nodes, each with NVIDIA GB200 NVL4 (2x ARM Neoverse V2, each 72 Cores + 4x B200-186GB) <coming soon> 3 nodes, each with 2x AMD EPYC 9655 and 4x NVIDIA H200-141GB (exclusively for hpcCluster.nrw) |
| Accelerator Nodes (FPGA, …) | 32 nodes, each with • 2x CPU, AMD EPYC 9655 "Turin" • 2x 96 cores, 2.6 GHz up to 4.5 GHz • 768 GiB main memory • FPGA configuration (pilot installation with 3x AMD Alveo V80, coming soon Altera Agilex) • GPU configuration (10 nodes with 3x NVIDIA A40) • Tenstorrent configuration |
| Communication Network | Nvidia NDR InfiniBand fabric, blocking factor 2:1 • compute/FPGA nodes connected with 200Gbps • GPU nodes connected with 2x400Gbps |
| Storage System | IBM Storage Scale (GPFS) 5,4 PB capacity (Flash + Hard disks) <coming soon> additional 2,9 PB capacity (Hard disks) |