Supercomputers/Systems/Santos Dumont
Brazil · operational · hpc
Santos Dumont
Operated by LNCC at LNCC (Petropolis) .
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
- 14.29 PFlop/s
- Rpeak
- -
- Rmax ÷ Rpeak
- -
- Cores
- 68,064
- Power
- 312.03 kW
- Per watt
- 45.8 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 | - BullSequana XH3000 | Eviden | - | Verified | baguete.com.brtop500.org |
| Accelerator | NVIDIA H100 SXM5 Accelerators CUDA (Hopper GH100) · 16,896C · 1.98 GHz · TSMC 4N · 700 W 62 BullSequana XH3145-H blades with four NVLink-connected H100 | NVIDIA | 248 estimated | Verified | blog.nvidia.com.brbebeez.eu |
| Accelerator | NVIDIA GH200 Grace Hopper Superchip Accelerators Arm CPU + CUDA GPU · TSMC 4N 36 BullSequana XH3515-H blades with four Grace Hopper Superchips | NVIDIA | 144 estimated | Verified | blog.nvidia.com.brtop500.org |
| Accelerator | AMD Instinct MI300A Accelerators ROCm (CDNA 3) · TSMC N5/N6 · 550 W 18 nodes with two AMD Instinct MI300A | AMD | 36 estimated | Verified | blog.nvidia.com.brbebeez.eu |
| CPU | AMD EPYC 9004 (Genoa) CPUs x86-64 · 96C · 2.4 GHz · TSMC N5 · 360 W 60 nodes with two AMD EPYC 9684X | AMD | 120 estimated | Verified | blog.nvidia.com.brbebeez.eu |
| Interconnect | NVIDIA InfiniBand NDR (Quantum-2) Interconnect Fat tree or dragonfly+ · 400 Gb/s per port · Under 0.6 microseconds NVIDIA InfiniBand NDR non-blocking fabric | NVIDIA | - | Verified | blog.nvidia.com.brtop500.org |
| Cooling | Direct liquid cooling, warm water Cooling Direct-to-chip cold plates, facility water loop · Removes 90+ percent of node heat to liquid direct liquid cooling capturing more than 98.5 percent of heat | Eviden | - | Verified | blog.nvidia.com.brbebeez.eu |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription Red Hat Enterprise Linux | Red Hat | - | Reported | top500.org |
The read
Brazil's national research supercomputer at LNCC, first installed as a Bull system in 2015 and extended in 2019. Eviden's 2024 contract (about US$19.4 million, signed 18 April 2024 per Baguete) added five BullSequana XH3000 partitions: 62 blades of four H100, 60 AMD EPYC 9684X nodes, 36 blades of four Grace Hopper Superchips, 18 MI300A nodes and four Grace nodes, on NDR InfiniBand with direct liquid cooling. DCD's account has work beginning in December 2024 and finishing in July 2025, yet TOP500 already lists a new entry at 89th on the November 2024 list, so no go-live date is recorded. TOP500 lists the machine as separate entries over the years; this row uses the 2024 BullSequana XH3000 entry (best rank 89th), and it is unclear which partitions its 14.29 PFlop/s and 68,064 cores cover, so the power figure is left out. LNCC says the expanded machine totals 18.85 PFlop/s while TOP500's Rpeak for the entry is 20.26, so no peak is recorded. TOP500 is the only source for Rmax and Rpeak, so the tier is reported.
Timeline
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Site & facility
Where sources disagree
We record conflicts instead of picking a winner quietly. All open questions.
- Peak performance: 18.85 PFlop/s for the whole machine after expansion (LNCC, via Baguete) against 20.26 PFlop/s for the 2024 entry (TOP500). The TOP500 entry alone exceeds LNCC total; no peak is recorded.
Change history
- 2026-09-22 Gained a Green500 rank pointer (#52 on the 2025-06 list)
- 2026-09-20 Added (2024 BullSequana XH3000 entry); peak disputed between LNCC and 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 |
|---|---|---|
| baguete.com.br | names system, part and count | NVIDIA H100 SXM5 · counts: 36, 120 |
| top500.org | names system, part and count | NVIDIA GH200 Grace Hopper Superchip, NVIDIA InfiniBand NDR (Quantum-2) · counts: 36 |
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
Why accelerators exist, what the memory hierarchy costs, and how the CPU and accelerator merged onto one package.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.