en.Wedoany.com Reported - On June 23, the ISC2026 conference in Hamburg, Germany, released the global TOP500 supercomputer list. China's domestically developed "Lingsheng" supercomputer topped the list with a sustained double-precision floating-point performance of 2.19 EFLOPS, ranking first globally. This marks China's return to the top of the TOP500 list after nine years and signifies that domestically produced supercomputing systems have entered a new era of exascale computing in sustained performance testing.
Deployed at the National Supercomputing Center in Shenzhen, the "Lingsheng" system achieved a measured result of 2.198 Exaflop/s in the HPL benchmark, becoming the first supercomputing system on the TOP500 list to exceed 2 Exaflop/s in sustained double-precision performance. This result surpasses the previously top-ranked US El Capitan system and pushes the performance threshold of the global public supercomputing list to new heights.
According to official list information, "Lingsheng" utilizes a customized LingKun platform, equipped with LX2 304-core processors running at 1.55GHz, with a total core count of 13,789,440. It employs the LingQi proprietary interconnect and the Kylin operating system. The system's theoretical peak performance is 2.736 Exaflop/s, with an HPL measured efficiency of approximately 80%, reflecting the overall collaborative capability of the large-scale system in processors, interconnects, software stack, and parallel computing scheduling.
The uniqueness of "Lingsheng" lies in its CPU-only architecture. Current international supercomputing and AI computing centers generally rely on GPU accelerators. However, "Lingsheng" achieves sustained double-precision floating-point performance exceeding 2 EFLOPS without relying on GPUs, which holds significant importance for traditional scientific computing, engineering simulation, and large-scale parallel computing tasks. The official TOP500 list also shows that "Lingsheng" ranks first on the HPCG list with a result of 22.00 Petaflop/s, indicating not only high Linpack computing power but also outstanding performance in data access and communication patterns closer to practical engineering calculations.
The competition in supercomputing is not just about peak numbers but also about systemic capabilities. An exascale supercomputing system requires a complete closed loop involving processors, networks, storage, operating systems, compilers, parallel algorithms, energy management, and application software. For research institutions and industrial users, higher sustained performance means that tasks such as climate simulation, fluid dynamics, materials science, life sciences, energy development, aerospace, and complex equipment design can be run at larger scales, higher resolutions, and with more complex models.
"Lingsheng" topping the TOP500 list also brings China's supercomputing back to the global top spot. In 2017, China's "Sunway TaihuLight" held the world's number one position; subsequently, systems from the US, Europe, and Japan maintained leadership on the public list. This time, "Lingsheng" returns to the top with a fully domestically developed system, demonstrating new publicly verified results in China's high-performance computing infrastructure, independent processor platforms, and large-scale system engineering.
For the domestic industrial and research systems, the significance of this milestone lies in the further strengthening of the computing power foundation. Supercomputing does not serve only a single research project; it is a crucial infrastructure for advanced manufacturing, drug development, materials design, energy equipment, engineering simulation, and AI model training. With the growing demand for industrial software, scientific intelligence, digital twins, and complex system simulation, exascale supercomputing platforms will become vital tools supporting efficiency improvements in R&D across key industries.
"Lingsheng" reaching the top is not an isolated ranking event but a concentrated demonstration of the overall capabilities of domestically produced supercomputing, from system architecture and core processors to interconnect networks and software ecosystems. As sustained double-precision performance breaks through 2 EFLOPS, supercomputing is evolving from a "national treasure" for a few top-tier scientific tasks into a foundational platform supporting the integrated development of industrial innovation, scientific discovery, and intelligent computing.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
