Supercomputers/Systems/Hortense
Belgium · operational · hpc
Hortense
Also known as VSC Tier-1 Hortense; HPC-UGent Hortense
Operated by Flemish Supercomputer Centre at HPC-UGent (Ghent) .
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
- -
- Rpeak
- 3.30 PFlop/s
- 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 |
|---|---|---|---|---|---|
| Integrator | - Atos BullSequana XH2000 (Ghent University framework contract) | Atos | - | Reported | vscentrum.be |
| Accelerator | NVIDIA A100 SXM4 40GB Accelerators CUDA (Ampere GA100) · 6,912C · 1.41 GHz · 7 nm · 400 W Phase 1: 20 GPU nodes x 4 NVIDIA A100 SXM4 40 GB (quantity derived as 20 x 4) | NVIDIA | 80 reported | Reported | github.com |
| Accelerator | NVIDIA A100 SXM4 80GB Accelerators Phase 2: 20 GPU nodes x 4 NVIDIA A100 SXM4 80 GB with NVLink3 (quantity derived as 20 x 4) | NVIDIA | 80 reported | Reported | docs.vscentrum.begithub.com |
| CPU | - Phase 1 Rome nodes with dual 64-core AMD EPYC 7H12: 342 standard plus 42 large-memory (quantity is nodes) | AMD | 384 reported | Reported | github.com |
| CPU | AMD EPYC 7763 (Milan) CPUs x86-64 · 64C · 2.45 GHz · 7 nm · 280 W 384 Phase 2 nodes with dual 64-core AMD EPYC 7763 Milan (quantity is nodes) | AMD | 384 reported | Reported | docs.vscentrum.begithub.com |
| Interconnect | NVIDIA InfiniBand HDR Interconnect Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds InfiniBand HDR-100 in a 2:1 fat tree (dual HDR on A100 GPU nodes) | NVIDIA | - | Reported | docs.vscentrum.be |
The read
The third Tier-1 cluster of the Flemish Supercomputer Centre (VSC), hosted at Ghent University and built by Atos on BullSequana XH2000, named after Nicole-Reine Lepaute. VSC announced it on 30 September 2020 with about 44,000 cores, 80 GPUs, 100 TB of RAM, 3 PB of storage and 3.3 PFlop/s peak, which is the operator figure recorded here and refers to the original configuration. A Phase 2 expansion in May 2023 added 384 nodes with dual AMD EPYC 7763 (128 cores each) and 20 nodes with four A100 80 GB GPUs each. The VSC documentation lists 5.4 PB of shared Lustre scratch and HDR-100 InfiniBand in a 2:1 fat tree. The Phase 1 partitions (342 plus 42 Rome nodes on EPYC 7H12 and 20 nodes with four A100 40 GB GPUs) were decommissioned on 11 July 2026, so the row reflects the October 2026 state with Phase 2 hardware active. No HPL result is recorded here and no whole-system core count is stated. Both sources are VSC publications, so the tier is reported.
Site & facility
Change history
- 2026-10-04 Added Hortense, VSC Tier-1 (Ghent)
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 |
|---|---|---|
| docs.vscentrum.be | names system, part and count | AMD EPYC 7763 (Milan), NVIDIA A100 SXM4 40GB, NVIDIA A100 SXM4 80GB, NVIDIA InfiniBand HDR · counts: 384, 80, 80 |
| vscentrum.be | names system only | · counts: 80, 80 |
Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.
Further reading
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.
Inside a node
Latency-optimised versus throughput-optimised processors, NUMA, and the end of the host-to-device copy.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.