COMPUTE ATLAS Supercomputer supply-chain graph
1540 systems 545 sites Sourced data

Supercomputers/Systems/GSI Virgo Cluster

Germany · operational · hpc

GSI Virgo Cluster

Also known as Virgo; GSI Green Cube cluster

Operated by GSI Helmholtz Centre for Heavy Ion Research at GSI Helmholtz Centre (Darmstadt) .

Researched with AI assistance, checked by scripts. These rows were researched by AI agents reading public operator, laboratory and vendor sources, appended by scripts and checked by automated validation; no subject-matter expert has reviewed them (how this site is made). A mechanical second-reader pass has since fetched every cited source: 1350 of 1540 systems have a readable citation that names them, and 64 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
-
Rpeak
-
Rmax ÷ Rpeak
-
Cores
-
Power
-
Per watt
- 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
Accelerator -
AMD Radeon Instinct GPUs (majority of GPUs)
AMD - Reported virgo-docs.hpc.gsi.de
Accelerator -
NVIDIA Hopper GPUs, small test set
NVIDIA - Reported virgo-docs.hpc.gsi.de
CPU AMD EPYC 9004 (Genoa)
CPUs
x86-64 · 96C · 2.4 GHz · TSMC N5 · 360 W
AMD EPYC 9654 (Genoa) dual-socket nodes, 94 nodes at 384 threads
AMD 188
reported
Reported virgo-docs.hpc.gsi.de
CPU AMD EPYC 9005 (Turin)
CPUs
x86-64 · 192C · 2.7 GHz · TSMC N3E · 500 W
AMD EPYC 9555 (Turin) dual-socket nodes, 97 nodes at 256 threads
AMD 194
reported
Reported virgo-docs.hpc.gsi.de
CPU AMD EPYC 7713 (Milan)
CPUs
x86-64 · 64C · 2 GHz · 7 nm · 225 W
AMD EPYC 7713 (Milan) dual-socket nodes, 133 nodes at 256 threads
AMD 266
reported
Reported virgo-docs.hpc.gsi.de
CPU -
AMD EPYC 7662 high-memory nodes, 43 nodes at 256 threads
AMD 86
reported
Reported virgo-docs.hpc.gsi.de
CPU -
Intel Xeon Gold 6248R nodes, 10 nodes at 96 threads
Intel 20
reported
Reported virgo-docs.hpc.gsi.de
Interconnect NVIDIA InfiniBand HDR
Interconnect
Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds
NVIDIA InfiniBand HDR 200 Gb/s, 2:1 blocking fat tree, replaced FDR in June 2023
NVIDIA - Reported virgo-docs.hpc.gsi.de

The read

Main production cluster of GSI's HPC group, housed in the Green IT Cube data centre on the GSI/FAIR campus in Darmstadt and used for FAIR and GSI physics simulation and experiment data analysis. It replaced the Kronos cluster (in production 2015 to end of 2020) after pre-production access in June 2020 and entry into production in August 2020. The sinfo listing on GSI's hardware page shows 377 dual-socket nodes: 94 AMD EPYC 9654 nodes (6 submit, 88 execution), 97 EPYC 9555, 133 EPYC 7713, 43 high-memory EPYC 7662 and 10 Intel Xeon Gold 6248R nodes. The fabric moved from InfiniBand FDR to HDR in June 2023 as a 2:1 blocking fat tree, with each 200 Gb/s switch port split into two 100 Gb/s node links. GSI's cluster page claims about 70,000 CPUs and 400 GPUs for users; the docs say most GPUs are AMD Radeon Instinct cards with a small set of NVIDIA Hopper parts for testing, and since Virgo 4.0 (March 2026) the GPUs sit in a separate cluster called Hydra, with a third test cluster, Orion. The hardware page text is dated 2023 although its sinfo example is from October 2026. No HPL result has been published, so FLOPS are blank. Earlier GSI machine in the dataset: l-csc.

Timeline

Site & facility

GSI Helmholtz Centre (Darmstadt)

Location
Darmstadt, Germany
Commissioned
2014

Change history

Further reading