Supercomputers/Systems/MN-1
Japan · operational · ai training
MN-1
Also known as PFN MN-1
Operated by Preferred Networks at NTT Communications Tokyo No.4 Data Center .
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 1540 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
- 1.39 PFlop/s
- Rpeak
- 4.92 PFlop/s
- Rmax ÷ Rpeak
- 28.3%
- Cores
- 59,392
- 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 | - NTT Communications / NTT PC Communications GPU platform (Supermicro SuperServer 4028GR-TR2 per TOP500) | NTT Communications | - | Reported | top500.orgpreferred.jp |
| Accelerator | NVIDIA Tesla P100 Accelerators CUDA (Pascal GP100) · 3,584C · 1.48 GHz · 16 nm · 300 W 1,024 NVIDIA Tesla P100 GPUs, 8 per node | NVIDIA | 1,024 reported | Reported | preferred.jppreferred.jp |
| CPU | - Intel Xeon E5-2667v4 8C 3.2GHz | Intel | - | Reported | top500.org |
| Interconnect | Mellanox InfiniBand FDR Interconnect Fat tree · 56 Gb/s per 4x port · About 1 microsecond Mellanox FDR InfiniBand, two 56 Gb/s links per node | Mellanox NVDANASDAQ
rolls up to NVIDIA | - | Reported | preferred.jppreferred.jp |
| Operating system | Ubuntu Server Operating systems Linux distribution · Open source, free with paid support tiers Ubuntu | Canonical | - | Reported | top500.org |
The read
Preferred Networks' first private deep-learning supercomputer: 128 GPU servers with eight NVIDIA Tesla P100 each (1,024 GPUs), two 56 Gb/s FDR InfiniBand links per node in a thin tree, built on a GPU platform from NTT Communications and NTT PC Communications and run by NTT Communications exclusively for PFN in an NTT data center. PFN says it began operating in September 2017 and ranked 91st on the November 2017 TOP500 at about 1.39 PFlop/s, roughly 28 percent of the 4.92 PFlop/s peak, which PFN's press release calls 12th in the world and first in Japan among industrial supercomputers. It was used for the 15-minute ImageNet ResNet-50 training result; the V100-based MN-1b extension (64 servers, July 2018 to July 2021) is not a separate record here. Later PFN clusters are mn-2 and mn-3. TOP500 lists it until November 2020 (rank 481). The TOP500 entry names the hardware as Supermicro SuperServer 4028GR-TR2 with Xeon E5-2667v4. 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
Change history
- 2026-10-17 MN-1 added from its source pages
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 |
|---|---|---|
| preferred.jp | names system, part and count | Mellanox InfiniBand FDR · counts: 1024 |
| top500.org | names system and part | NVIDIA Tesla P100, Mellanox InfiniBand FDR, Ubuntu Server |
Checked 2026-10-18 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.