China has officially seized the title of the world’s fastest supercomputer for the first time since 2017. Shaking up the global rankings, a previously unlisted Chinese machine named LineShine clocked a staggering 2.198 Exaflops of performance. The feat comfortably dethrones the previous American champion, El Capitan, which sits at 1.809 Exaflops at the Lawrence Livermore National Laboratory.
Read: Samsung Galaxy Buds 4 Review: Brilliant brains, flawed physics
LineShine is the first system in history to exceed two exaflops of sustained double-precision performance, and it did so utilizing CPUs alone.
LineShine’s architectural victory is particularly notable given ongoing global technology embargoes. While western frontrunners rely heavily on graphics processing units (GPUs) to accelerate massive AI and mathematical workloads, China’s National Supercomputer Centre designed its new champion entirely around a custom 304-core CPU.
- The Core Network: The system welds together a massive 13.79 million processing cores running at 1.55GHz, all bound by a high-speed, proprietary internal interconnect.
- Power & Efficiency: LineShine pulls roughly 42.2 megawatts of power, yielding an efficiency rating of 52.07 Gigaflops per watt.
“It’s an impressive system,” Top500 organizer Dr. Jack Dongarra told The New York Times. “They upped us by developing a system that is not reliant on GPUs.”
While China claims the pinnacle of the list, the global race continues to accelerate. The latest Top500 rankings now feature five distinct supercomputers that have cleared the elite exascale threshold:
- LineShine (China): 2.198 Exaflops
- El Capitan (USA): 1.809 Exaflops
- Frontier (USA): 1.353 Exaflops (Oak Ridge National Laboratory)
- Aurora (USA): 1.012 Exaflops (Argonne National Laboratory)
- Jupiter Booster (Germany): 1.000 Exaflop (Jülich Supercomputing Centre)
See also
This elite bracket showcases an era of immense architectural diversity. Rather than a single dominant hardware path, top-tier vendors are successfully mixing and matching custom-accelerator approaches alongside silicon from AMD, Intel, and NVIDIA.
Beijing frequently shields its breakthrough supercomputer blueprints from the public eye due to geopolitical sensitivities and government restrictions. However, LineShine represents a rare exception because it was financed entirely through private corporate backing rather than state funding.
Because of its private nature, the system’s designers felt cleared to submit LineShine to the rigorous Top500 standardized tests. Even so, certain competitive mysteries remain intact: the creators have flatly declined to reveal which semiconductor foundry manufactured the custom 304-core CPUs, or the underlying microarchitecture used to print the chips.
