Germany · operational · ai training

Helma

Also known as NHR@FAU Helma

Operated by FAU Erlangen-Nuremberg at RRZE Erlangen .

Phase 1 seed dataset, compiled by hand. These rows were built from public operator, laboratory and vendor sources. A mechanical second-reader pass has since fetched every cited source: 1092 of 1153 systems have a readable citation that names them, and 40 are genuinely weakly sourced. Every claim carries its source and a confidence tier. Treat anything below verified as a lead, not a citation.

Measured performance

Rmax: 32.22 PFlop/s
Rpeak: 51.03 PFlop/s
Rmax ÷ Rpeak: 63.1%
Cores: 125,952
Power: 651.13 kW
Per watt: 49.5 GF/W

Figures are as last publicly reported for the configuration described below, not a live measurement. Where a system was upgraded in place, the post-upgrade configuration is shown and the earlier one appears in the timeline.

What this machine is made of

One row per supplier relationship. “Supplier at build” is the company that shipped the part at the time; where that company has since been acquired, the parent it rolls up to today is shown beside it. That distinction is what makes ticker-level aggregation possible across a thirty-year dataset.

Role	Part	Supplier at build	Quantity	Confidence	Source
Integrator	- MEGWARE (system integrator)	MEGWARE	-	Verified	top500.org hpc.fau.de
Accelerator	NVIDIA H200 SXM Accelerators CUDA (Hopper GH100) · 16,896C · 1.98 GHz · TSMC 4N · 700 W 96 nodes with 4 NVIDIA H200 141GB GPUs each	NVIDIA NVDANASDAQ	384 reported	Reported	doc.nhr.fau.de top500.org
Accelerator	NVIDIA H100 SXM5 94GB Accelerators CUDA (Hopper GH100) · 16,896C · 1.98 GHz · TSMC 4N · 700 W 96 nodes with 4 NVIDIA H100 94GB GPUs each	NVIDIA NVDANASDAQ	384 reported	Reported	doc.nhr.fau.de
CPU	AMD EPYC 9004 (Genoa) CPUs x86-64 · 96C · 2.4 GHz · TSMC N5 · 360 W AMD EPYC 9554 dual-socket nodes	AMD AMDNASDAQ	-	Verified	top500.org doc.nhr.fau.de
Interconnect	NVIDIA InfiniBand NDR (Quantum-2) Interconnect Fat tree or dragonfly+ · 400 Gb/s per port · Under 0.6 microseconds Infiniband NDR200 (4x NDR200 internal, 2x NDR400 external per GPU node)	NVIDIA NVDANASDAQ	-	Verified	top500.org doc.nhr.fau.de

The read

The GPU cluster of NHR@FAU at Friedrich-Alexander-Universitaet Erlangen-Nuernberg, integrated by MEGWARE: dual AMD EPYC 9554 nodes with four NVIDIA H100 or H200 GPUs each (96 nodes of each type, 768 GPUs in total per the operator's documentation), InfiniBand NDR200, alongside a CPU partition of dual EPYC 9965 nodes. The H200 half is reserved for a Bavarian foundation-model effort; funders were the State of Bavaria, the NHR programme, FAU and partner universities. TOP500 lists 125,952 cores, Rmax 32.22 PFlop/s, Rpeak 51.03 PFlop/s and 651 kW under AlmaLinux 9.5. It first entered at 79th in November 2024 and reached its best rank, 51st, on the June 2025 list; FAU calls it the fastest supercomputer at German universities. Workload is classed ai_training because the operator names AI/ML and the foundation-model project as its purpose alongside atomistic simulation.

Timeline

2024 First operational source
2025-06 List debut at #51 (June 2025) source

Public list appearances

Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.

#51: TOP500, June 2025 →

Site & facility

RRZE Erlangen

Location: Erlangen, Germany
Commissioned: 2024

top500.org

Change history

2026-10-02 Added Helma, the NHR@FAU H100/H200 cluster, 51st on the June 2025 TOP500

Source check

We fetched this system's own citations and recorded whether each page actually mentions it. This is a corroboration signal, not a fact check, and it is published so you can see how well the sourcing holds up rather than take it on trust.

Cited source	Result	Found on the page
hpc.fau.de	names system, part and count	NVIDIA H200 SXM, NVIDIA H100 SXM5 94GB · counts: 384, 384
top500.org	names system and part	AMD EPYC 9004 (Genoa), NVIDIA H200 SXM, NVIDIA H100 SXM5 94GB, NVIDIA InfiniBand NDR (Quantum-2)

Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.