Supercomputers/Systems/Pegasus (Tsukuba)
Japan · operational · mixed
Pegasus (Tsukuba)
Operated by University of Tsukuba at Tsukuba Center for Computational Sciences .
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
- 4.34 PFlop/s
- Rpeak
- 8.13 PFlop/s
- Rmax ÷ Rpeak
- 53.4%
- Cores
- 27,000
- Power
- 130.41 kW
- Per watt
- 33.3 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 | - NEC LX 102Bk-6 | NEC | - | Verified | top500.orgccs.tsukuba.ac.jp |
| Accelerator | NVIDIA H100 80GB PCIe Accelerators CUDA (Hopper) NVIDIA H100 PCIe 80GB, one per node across 150 nodes | NVIDIA | 150 estimated | Reported | ccs.tsukuba.ac.jptop500.org |
| CPU | Intel Xeon Platinum 8468 CPUs x86-64 · 48C · 2.1 GHz · Intel 7 · 350 W Intel Xeon Platinum 8468 48C 2.1GHz (Sapphire Rapids), one per node | Intel | - | Verified | top500.orgccs.tsukuba.ac.jp |
| Interconnect | NVIDIA InfiniBand NDR (Quantum-2) Interconnect Fat tree or dragonfly+ · 400 Gb/s per port · Under 0.6 microseconds NVIDIA Quantum-2 InfiniBand NDR200, 200 Gb/s per node, full-bisection fat tree | NVIDIA | - | Verified | ccs.tsukuba.ac.jptop500.org |
| Operating system | Ubuntu Server Operating systems Linux distribution · Open source, free with paid support tiers Ubuntu 22.04 | Canonical | - | Reported | top500.org |
The read
The University of Tsukuba Center for Computational Sciences' NEC big-memory system, announced on 22 December 2022 as the first to pair NVIDIA H100 PCIe GPUs with 4th-generation Xeon (Sapphire Rapids, 48-core) CPUs and Intel Optane persistent memory, on NVIDIA Quantum-2 InfiniBand. The announcement described 120 nodes with 128 GiB DDR5 and 2 x 3.2 TB NVMe each and a 7.1 PB parallel file system; the centre's current page lists 150 nodes and a peak above 8.1 PFlop/s, matching TOP500's 27,000 cores after the expansion. TOP500 gives Rmax 4.34, Rpeak 8.13 and 130.41 kW, 296th in November 2025 (first listed 190th in June 2023). It is unrelated to the Petrobras system this dataset lists as Pegasus (Pegaso), hence the disambiguated name. No start-of-service date was found in a page we fetched, so it is blank.
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
Tsukuba Center for Computational Sciences
- Location
- Tsukuba, Japan
- Commissioned
- 1996
Change history
- 2026-10-02 Pegasus (Tsukuba) added, verified on TOP500 and the University of Tsukuba
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 |
|---|---|---|
| top500.org | names system and part | Intel Xeon Platinum 8468, NVIDIA H100 80GB PCIe, NVIDIA InfiniBand NDR (Quantum-2), Ubuntu Server |
| ccs.tsukuba.ac.jp | names system, part and count | Intel Xeon Platinum 8468, NVIDIA H100 80GB PCIe, NVIDIA InfiniBand NDR (Quantum-2) · counts: 150 |
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.