← Retour au blog
tech 29 June 2026

TOP500 at ISC'26: A New Number 1 Supercomputer

At ISC 2026, the computing world was stunned by the debut of the LineShine supercomputer at the top of the TOP500 list. Discover how this Chinese machine is redefining computational power limits.

Article inspired by the original source
TOP500 at ISC’26: We have a New Number 1 Supercomputer ↗ chipsandcheese.com

Introduction

At the International Supercomputing Conference 2026 in Hamburg, the world witnessed the unveiling of the new leader of the TOP500 list: the LineShine supercomputer, a technological marvel from Shenzhen, China. After nine years without Chinese submissions to the TOP500, LineShine makes a comeback with a CPU-only system.

Technical Specifications

The driving force behind LineShine is the LX2 processor, compliant with the Armv9 architecture and supporting SVE2 and SME. Each core boasts 32 KB of L1 cache for instructions and data. The LX2 consists of two compute dies, each with four clusters of 40 cores, two of which are disabled per cluster, resulting in 38 active cores per cluster. This totals 152 active cores per die, backed by 114 MB of L2 cache, for a total of 304 active cores and 228 MB of L2 cache for the entire processor.

Operating at 1.55 GHz, the LX2 delivers 60.3 TFLOP/s of FP64 performance at 690 watts. With eight stacks of high-bandwidth memory of 4 GB each, the processor achieves 32 GB of on-package memory with 4 TB/s bandwidth. Additionally, each LX2 is paired with 256 GB of DDR5 memory.

An Impressive Infrastructure

Each compute node in LineShine contains two LX2 processors with 800 Gbps of networking, totaling 1.6 Tbps per node. Eight nodes form a compute blade, 16 blades form a compute frame, and two frames make up a compute cabinet. With 90 cabinets, the LineShine supercomputer encompasses over 22,000 nodes and 13 million CPU cores.

Performance and Efficiency

LineShine achieves 2.198 Exaflops of sustained FP64 computing (Rmax) out of a potential 2.735 Exaflops (Rpeak), consuming 42.22 megawatts for an efficiency of 52.07 Gigaflops per watt. Although less efficient than the leader in energy efficiency, it is a remarkable feat for a CPU-only system. Beyond its LINPACK benchmark success, LineShine also leads the HPCG benchmark with 22.004 Petaflops per second.

Impact on the Global Landscape

The emergence of LineShine reshapes the supercomputing landscape, marking China's return to the peak of computational performance. Its CPU-only architecture and spectacular performance raise new questions about the future of supercomputers and strategies for optimizing energy efficiency.

Conclusion

LineShine not only pushes the boundaries of computing power but also redefines expectations for the future of supercomputers. This development highlights the importance of innovation and investment in cutting-edge technologies to remain competitive on the global stage.

Let's discuss your project in 15 minutes.

supercomputer LineShine TOP500 ISC 2026 LX2 processor
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call