Supercomputers/Systems/Plasma Simulator Sosei
Japan · operational · hpc
Plasma Simulator Sosei
Also known as Sosei; NIFS-QST Plasma Simulator (7th generation)
Operated by NIFS at QST Rokkasho Fusion Institute .
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
- 40.38 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 | - NEC LX 204Bin-3 (CPU) and LX 401Bax-3GA (APU) servers | NEC | - | Reported | techspot.com |
| Accelerator | AMD Instinct MI300A Accelerators ROCm (CDNA 3) · TSMC N5/N6 · 550 W AMD Instinct MI300A in 70 APU servers (subsystem B) | AMD | 280 reported | Reported | techspot.comnsrp.nifs.ac.jp |
| CPU | Intel Xeon 6 6900P (Granite Rapids) CPUs x86-64 · 128C · 2 GHz · Intel 3 · 500 W Intel Xeon 6980P in 360 CPU servers (subsystem A) | Intel | 720 reported | Reported | techspot.comnsrp.nifs.ac.jp |
| Interconnect | NVIDIA InfiniBand NDR (Quantum-2) Interconnect Fat tree or dragonfly+ · 400 Gb/s per port · Under 0.6 microseconds InfiniBand with NVIDIA QM9700 switches | NVIDIA | - | Reported | techspot.com |
| Storage | DDN EXAScaler Storage Lustre appliance DDN ES400NVX2 with Lustre EXAScaler, 42.2 PB | DDN | - | Reported | techspot.comnsrp.nifs.ac.jp |
The read
The seventh-generation Plasma Simulator, run jointly by Japan's National Institute for Fusion Science and the National Institutes for Quantum Science and Technology (QST) and installed at QST's Rokkasho Fusion Energy Research Institute in Aomori. NIFS lists three subsystems totalling 478 nodes and 40.38 PFlop/s theoretical peak: 360 servers with Intel Xeon 6980P (5.89 PFlop/s), 70 servers with AMD Instinct MI300A accelerators (34.32 PFlop/s) and 48 large-memory servers with Xeon Gold 6544Y, plus 2.2 PB of SSD and 40 PB of HDD storage. Trade press reporting NEC's order announcement of November 2024 gives 720 Xeon 6900P processors and 280 MI300A accelerators in NEC LX servers, a DDN ES400NVX2 Lustre file system of 42.2 PB, InfiniBand switching (NVIDIA QM9700) and liquid cooling, with operation from July 2025 at 2.7 times the combined performance of the two earlier NIFS and QST systems. NIFS gave the machine the nickname Sosei (double star) in December 2025. It succeeds the sixth-generation Raijin system (see that row). Both sources agree on the peak figure, server counts and start date, so the tier is verified; no TOP500 entry was confirmed, so ranked is 0 and Rmax is blank.
Site & facility
QST Rokkasho Fusion Institute
- Location
- Rokkasho, Aomori, Japan
- Commissioned
- 2025
Change history
- 2026-10-01 Added Plasma Simulator Sosei (NIFS)
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 |
|---|---|---|
| nsrp.nifs.ac.jp | names system and part | Intel Xeon 6 6900P (Granite Rapids) |
| techspot.com | could not be fetched | - |
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.