Supercomputers/Systems/RIKYU
Japan · operational · ai training
RIKYU
Also known as RIKEN AI for Science supercomputer
Operated by RIKEN at RIKEN Center for Computational Science .
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
- 64.16 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 |
|---|---|---|---|---|---|
| Accelerator | - NVIDIA GB200 NVL4 Blackwell GPUs, 1,600 across 400 compute nodes | NVIDIA | 1,600 verified | Verified | docs.r-ccs.riken.jpinvestor.nvidia.com |
| Interconnect | NVIDIA InfiniBand XDR (Quantum-X800) Interconnect Fat tree or dragonfly+ · 800 Gb/s per port NVIDIA Quantum-X800 InfiniBand up to 3.2 Tb/s | NVIDIA | - | Verified | docs.r-ccs.riken.jpinvestor.nvidia.com |
AI datacenter metrics
Figures beyond an accelerator count, each with who stated it. What can and cannot be compared across AI datacenters.
| Metric | Value | As of | Basis | Note | Source |
|---|---|---|---|---|---|
| Accelerators | 1,600 accelerators | 2026-10 | Operator stated | 400 GB200 NVL4 nodes; RIKEN documentation and NVIDIA's November 2025 release agree on 1,600 Blackwell GPUs. | docs.r-ccs.riken.jpinvestor.nvidia.com |
The read
RIKEN's AI for Science supercomputer at the Kobe campus: 400 compute nodes of NVIDIA GB200 NVL4 (1,600 Blackwell GPUs) on Quantum-X800 InfiniBand at up to 3.2 Tb/s, rated by RIKEN at over 64.16 PFLOPS in FP64 and over 15.539 EFLOPS in FP8. NVIDIA's November 2025 release independently gives the 1,600-GPU count. RIKEN's user documentation shows it in Early Access Phase 2 for production operation until mid-October 2026, so full production had not started at the check date. It and ROQUO are proxy machines for the FugakuNEXT design. The FP64 figure is a stated rating, not an HPL run.
Timeline
- 2025-11-17 Announced source
Site & facility
RIKEN Center for Computational Science
- Location
- Kobe, Japan
- Cooling
- Direct liquid cooling (warm water)
- Commissioned
- 2010
Change history
- 2026-10-07 Added RIKYU (RIKEN)
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.
None of the cited sources could be read from our network: the hosts returned bot challenges or did not respond. This check says nothing about the citations either way. See how this check works.
| Cited source | Result | Found on the page |
|---|---|---|
| docs.r-ccs.riken.jp | could not be fetched | - |
| investor.nvidia.com | could not be fetched | - |
Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.