China's 'LineShine' takes first place in the world supercomputer rankings for the first time.

In the 67th 'TOP500' world supercomputer rankings announced in June 2026, Lawrence Livermore National Laboratory's 'El Capitan,' which had held the top spot for several consecutive years, dropped to second place, and China's 'LineShine (霊晟)' took its place as number one.

June 2026 | TOP 500

LineShine Debuts at No. 1 as the TOP500 Enters a New Global Exascale Era | TOP500

NVIDIA Powers Over 400 of the World's 500 Fastest Supercomputers | NVIDIA Blog
https://blogs.nvidia.com/blog/top500-green500-supercomputers-isc-2026/?linkId=100000427693245

China's Super Calculation 'Ling Sheng' Ranked in Global Super Calculation 500 New Year Book - New Book
https://www.news.cn/tech/20260624/37f2b0c76a0c494388560f0a657b0df8/c.html

LineShine is a supercomputer located at the National Supercomputing Shenzhen Center in Shenzhen, China, and was built by the Shenzhen Cloud Computing Center.

It is based on a 1.55GHz 304-core LX2 processor and a custom 'LingLun' platform with proprietary LingQi interconnect and Kylin OS, and has a power consumption of approximately 42.2MW.

In the HPL benchmark, using 13,789,440 cores, it achieved 2,198 Exaflops/s, nearly 80% of the theoretically highest value of 2,736 Exaflops/s. It became the first system in the TOP500 to surpass 2,000 Exaflops using only the CPU and to record sustained double-precision performance. This is the first time a Chinese system has topped the TOP500 since 2017.

In addition to LineShine, other consistently high-performing systems such as El Capitan (Lawrence Livermore National Laboratory), Frontier (Oak Ridge National Laboratory), Aurora (Argonne National Laboratory), and Jupiter Booster (Zurich Supercomputing Center) have also recorded 1000 Exaflop/s in the HPL benchmark. This marks the first time that exascale systems are present in three locations: Asia, North America, and Europe.

Furthermore, LineShine also ranked first in the HPCG rankings with 22 Petaflops/s. However, in the HPL-MxP benchmark, which measures mixed precision performance, it came in fourth place with 7.92 Exaflops/s. Since it is a CPU-only design, the speed-up multiplier was a modest 3.6x.

Incidentally, 81% of the supercomputers ranked in the TOP500 list use NVIDIA technology. If we consider only the newly added systems, the usage rate of NVIDIA rises to 90%.

Japan's fastest supercomputer, 'Fugaku,' is ranked 9th in the TOP500 list.