Supercomputers/Systems/Fugaku
Japan · operational · mixed
Fugaku
Quantum: linked to RIKEN IBM Quantum System Two (156 superconducting qubits)
Deep dive: Fugaku: the CPU-only No. 1 and its successor
Operated by RIKEN at RIKEN Center for Computational Science .
Researched with AI assistance, checked by scripts. These rows were researched by AI agents reading public operator, laboratory and vendor sources, appended by scripts and checked by automated validation; no subject-matter expert has reviewed them (how this site is made). A mechanical second-reader pass has since fetched every cited source: 1350 of 1581 systems have a readable citation that names them, and 64 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
- 442.0 PFlop/s
- Rpeak
- 537.2 PFlop/s
- Rmax ÷ Rpeak
- 82.3%
- Cores
- 7,630,848
- Power
- 29.90 MW
- Per watt
- 14.8 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.
How it compares
Ranked against the other machines in this dataset that report the same figure: those that first came into service within two years of Fugaku, and those in Japan. Computed from our tables; a metric is ranked only when at least eight peers report it.
| Metric | Value | Against its era | Against its country |
|---|---|---|---|
| Rmax | 442.0 PFlop/s | #2 of 213 · top 1% | #1 of 131 · highest |
| Energy efficiency | 14.8 GF/W | #42 of 116 · top 36% | #15 of 74 · top 19% |
| Rmax ÷ Rpeak | 82.3 % | #25 of 209 · top 12% | #48 of 130 · top 36% |
| Power draw | 29,899 kW | #1 of 118 · highest | #1 of 74 · highest |
Every metric is ranked largest-first, so a high power-draw rank means a big power draw, not a good one. Rmax ÷ Rpeak uses HPL-measured results only. Rankings move as machines are added; they describe this dataset, not the world.
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 | - Fujitsu | Fujitsu | - | Verified | r-ccs.riken.jpglobal.fujitsu |
| CPU | Fujitsu A64FX CPUs Armv8.2-A SVE · 48C · 2.2 GHz · 7 nm A64FX 48C 2.2GHz | Fujitsu | 158,976 reported | Verified | r-ccs.riken.jpglobal.fujitsu |
| Interconnect | Fujitsu Tofu Interconnect D Interconnect 6D mesh/torus · 6.8 GB/s per link per direction Tofu interconnect D | Fujitsu | - | Verified | r-ccs.riken.jpglobal.fujitsu |
| Storage | Fujitsu FEFS Storage Lustre-based parallel filesystem · Roughly 150 PB across multiple volumes on Fugaku FEFS Fujitsu Exabyte File System | Fujitsu | - | Verified | r-ccs.riken.jpwiki.lustre.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 water cooling | Fujitsu | - | Reported | r-ccs.riken.jpglobal.fujitsu |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription OS Red Hat Enterprise Linux 8 | Red Hat | - | Verified | r-ccs.riken.jptop500.org |
| Operating system | McKernel Operating systems Lightweight multi-kernel (IHK-partitioned), co-resident with Linux · Research software, RIKEN-developed light-weight multi kernel operating system McKernel | RIKEN | - | Verified | r-ccs.riken.jpnetlib.org |
Engineering, funding and access
What Fugaku runs on, who paid for it and how people get time on it. Each line carries who said it and when.
Hardware
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Compute nodes | 158,976 nodesOne A64FX CPU per node; 432 racks, 396 with 384 nodes and 36 with 192 nodes | 2021 | Operator stated | r-ccs.riken.jp |
| Total memory | 5,460 TBRIKEN states 4.85 PiB total (32 GiB HBM2 per node); converted to TB here | 2021 | Operator stated | r-ccs.riken.jp |
| Interconnect and topology | Tofu Interconnect D, 6D mesh/torus with RDMA, six network interfaces at 6.8 GB/s each per node (40.8 GB/s) | 2021 | Operator stated | r-ccs.riken.jp |
| Cooling design | Water cooling for 9 parts and air for 1 part of rack heat, using expanded chillers, heat exchangers and pumpsK computer used a 2:1 water to air split | 2021 | Operator stated | r-ccs.riken.jp |
Software
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Parallel filesystem | Three layers: LLIO job-local storage, FEFS (Lustre-based) shared file system, and commercial cloud storage | 2021 | Operator stated | r-ccs.riken.jp |
| Programming and software stack | Fujitsu Software Technical Computing Suite V4.0 (compiler, math library, MPI) | 2026-06 | Operator stated | top500.org |
| Programming and software stack | Red Hat Enterprise Linux 8 with McKernel, Spack and Singularity | 2021 | Operator stated | r-ccs.riken.jp |
| Programming and software stack | Fujitsu MPI (Open MPI based) and MPICH-Tofu (MPICH based) | 2021 | Operator stated | r-ccs.riken.jp |
Funding
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Funded by | Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan | 2021-03 | Vendor stated | info.archives.global.fujitsu |
| Funding programme | High-Performance Computing Infrastructure (HPCI), with Fugaku as its core system | 2021-03 | Vendor stated | info.archives.global.fujitsu |
| Procurement | Developed jointly by RIKEN and Fujitsu from 2014; all racks delivered May 2020, trial use from April 2020 | 2021-03 | Vendor stated | info.archives.global.fujitsu |
Access
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| How researchers get time | HPCI projects reviewed by RIST: general and junior researcher projects, industrial use, and Fugaku Trial AccessOpen to researchers in Japan and overseas | 2021 | Operator stated | r-ccs.riken.jp |
| Access page | www.hpci-office.jp/folders/english | 2021 | Operator stated | r-ccs.riken.jp |
Results
| Item | Detail | As of | Basis | Source |
|---|---|---|---|---|
| Notable result | 2025: ranked first on Graph500 for the 11th consecutive term at ISC 2025 | 2025-06 | Operator stated | r-ccs.riken.jp |
| Notable result | 2020: ranked first in TOP500, HPCG, HPL-AI and Graph500 in both June and November 2020 | 2021-03 | Vendor stated | info.archives.global.fujitsu |
The read
158,976 A64FX nodes, no accelerators, no separate NIC: the Tofu D interconnect is on the CPU package. The shortest bill of materials of any modern leadership system: one company supplies the processor, the memory interface, the fabric and the integration. It is the counterexample to the assumption that top-end compute necessarily implies a multi-vendor supply chain, and the reason our accelerator-share chart has to handle a null accelerator role rather than treating every system as heterogeneous.
Timeline
- 2014 Announced source
- 2020-04 Trial use of Fugaku began ahead of full delivery source
- 2020-05 Delivery of all Fugaku racks to RIKEN was completed source
- 2020-06 List debut at #1 (June 2020) source
- 2020-06 First operational source
- 2020-06 Fugaku took first place on TOP500, HPCG, HPL-AI and Graph500 in the same list cycle source
- 2021-03-09 Shared use of Fugaku opened to researchers through HPCI source
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Site & facility
RIKEN Center for Computational Science
- Location
- Kobe, Japan
- Cooling
- Direct liquid cooling (warm water)
- Commissioned
- 2010
Change history
- 2026-09-22 Gained a Green500 rank pointer (#54 on the 2023-11 list)
Nearby in scale
The machines closest to Fugaku by Rmax, smaller to larger.
Related systems
Same site
- FugakuNEXT · Japan, planned
- RIKYU · Japan, operational
- ROQUO · Japan, operational, 2026
- K computer · Japan, decommissioned, 2011
Same operator
- HOKUSAI BigWaterfall · Japan, operational, 2017
- HOKUSAI BigWaterfall 2 · Japan, operational, 2023
- RAIDEN · Japan, operational, 2017
- HOKUSAI GreatWave · Japan, decommissioned, 2015
- Shoubu · Japan, operational, 2015
- RICC · Japan, operational, 2009
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 |
|---|---|---|
| global.fujitsu | could not be fetched | - |
| r-ccs.riken.jp | could not be fetched | - |
Checked 2026-10-18 by pnpm verify. Re-run it and the table changes with the web.
Related news
- Fugaku takes the top position with no accelerators · RIKEN Center for Computational Science158,976 A64FX nodes, on-package HBM, and the Tofu fabric integrated onto the processor die. The shortest bill of materials of any modern leadership system.
Further reading
Power and cooling
Why air ran out at 20 kW a rack, how warm-water cooling removes the chillers, and what PUE conceals.
Inside a node
Latency-optimised versus throughput-optimised processors, NUMA, and the end of the host-to-device copy.
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.