Supercomputers/Systems/LineShine
China · operational · hpc
LineShine
Deep dive: LineShine: a CPU-only 2.2 exaflop machine
Operated by National Supercomputing Centre in Shenzhen at National Supercomputing Centre in Shenzhen .
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
- 2.20 EFlop/s
- Rpeak
- 2.74 EFlop/s
- Rmax ÷ Rpeak
- 80.4%
- Cores
- 13,789,440
- Power
- 42.22 MW
- Per watt
- 52.1 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 | - Built by Shenzhen Cloud Computing Center for the National Supercomputing Centre in Shenzhen | Shenzhen Cloud Computing Center | - | Reported | top500.org |
| CPU | LingKun LX2 CPUs Armv9.2 SVE2/SME · 304C · 1.55 GHz 13,789,440 cores across 22,680 LingKun LX2-based nodes | Shenzhen Cloud Computing Center | - | Reported | top500.orgnextplatform.com |
| Interconnect | LingQi LQLink Interconnect Dual-plane multi-rail fat-tree · 1.6 Tb/s per node · ~1 microsecond LingQi LQLink dual-plane multi-rail fat-tree fabric at roughly 1.6 Tb/s per node | Shenzhen Cloud Computing Center | - | Reported | nextplatform.com |
| Operating system | Kylin Linux Operating systems Linux distribution · State-endorsed Chinese distribution Kylin OS, per the TOP500 system entry | KylinSoft | - | Reported | top500.org |
Published benchmark results
- HPL-MxP 2026-06: HPL-MxP mixed-precision (rank 4), 7.92 EFlop/s (independent, 2026-06) top500.orgtop500.org
The read
The first Chinese system to lead TOP500 since 2017 (per TOP500), and the first exascale system built without a single accelerator anywhere in it. Its entire compute story is one in-house part, NSCC-SZ's own LingKun LX2 processor (Armv9.2 with SVE2 and SME matrix units), linked by a proprietary LingQi fabric: the same short, one-designer bill of materials that Fugaku and the BlueGene line used to win without merchant accelerators, arrived at independently and two silicon generations later. It displaced El Capitan from the top of the list. We could not independently confirm a first-operational date from more than one source, so that field is left blank rather than guessed. First-operational date is the April 2026 event at which the centre presented the machine; TOP500 gives installation year 2025.
Timeline
- 2026-04 First operational source
- 2026-04-24 Announced source
- 2026-06 List debut at #1 (June 2026) 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
Change history
- 2026-10-04 AI benchmark record: hpl-mxp 2026-06
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 | LingKun LX2, LingQi LQLink, Kylin Linux |
| nextplatform.com | names system and part | LingKun LX2, LingQi LQLink |
Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.
Further reading
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.
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.