China's LineShine, at the National Supercomputing Centre in Shenzhen, topped the 67th TOP500 list, announced on 23 June 2026 at ISC 2026 in Hamburg, with 2.198 exaflops on the HPL benchmark.
It pushed the US system El Capitan (1.809 exaflops), No. 1 since November 2024, into second place. It is the first Chinese No. 1 since Sunway TaihuLight in 2017.
LineShine uses no GPUs: 13.79 million cores on 304-core LX2 processors (the LingKun platform), the LingQi interconnect and the Kylin operating system, drawing 42.2 MW.
Five systems now exceed one exaflop: LineShine, El Capitan, Frontier, Aurora and Germany's JUPITER Booster.
On the mixed-precision HPL-MxP benchmark, which approximates AI workloads, LineShine is only fourth (7.92 exaflops); El Capitan leads with 16.7 exaflops.
Top 10 systems on the 67th TOP500 list (HPL Rmax)
| Rank | System | Country | Rmax |
|---|---|---|---|
| 1 | LineShine | China | 2.198 exaflops |
| 2 | El Capitan | United States | 1.809 exaflops |
| 3 | Frontier | United States | 1.353 exaflops |
| 4 | Aurora | United States | 1.012 exaflops |
| 5 | JUPITER Booster | Germany | 1.000 exaflops |
| 6 | HPC7 | Italy | 571.5 petaflops |
| 7 | Eagle | United States | 561.2 petaflops |
| 8 | HPC6 | Italy | 477.9 petaflops |
| 9 | Fugaku | Japan | 442 petaflops |
| 10 | Alps | Switzerland | 434.9 petaflops |
Past TOP500 No. 1 systems
| Period at No. 1 | System | Country |
|---|---|---|
| June 2026 onward | LineShine | China |
| Nov 2024 to Nov 2025 | El Capitan | United States |
| June 2022 to June 2024 | Frontier | United States |
| June 2020 to Nov 2021 | Fugaku | Japan |
| June 2018 to Nov 2019 | Summit | United States |
| June 2016 to Nov 2017 | Sunway TaihuLight | China |
| June 2013 to Nov 2015 | Tianhe-2A | China |
| Nov 2012 | Titan | United States |
| June 2012 | Sequoia | United States |
| June 2011 to Nov 2011 | K computer | Japan |
| Nov 2010 | Tianhe-1A | China |
Ranks the 500 most powerful supercomputers by the High Performance Linpack (HPL) benchmark; established in 1993 by Jack Dongarra, Erich Strohmaier, Horst Simon and Hans Meuer
Hosts LineShine; it names no vendor for the LX2 processor
An exaflop is 10^18 floating-point operations per second. HPL measures double-precision performance solving dense linear equations. HPL-MxP measures mixed precision, which AI uses; GPU machines gain 8-11x there, but the CPU-only LineShine gained only 3.6x.
Simple Analogy: HPL is a sprint in heavy boots; HPL-MxP lets runners switch to spikes.
GS Paper III > Science and Technology > IT, computers, supercomputing
General Awareness > Science and Technology > Rankings and records
10^18 floating-point operations per second; the exascale threshold.
Double-precision benchmark that decides TOP500 rank.
Mixed-precision benchmark that approximates AI workloads; led by El Capitan in June 2026.