Why the Best‑Performing Open‑Source CPU Is Chinese: The XiangShan Project
The XiangShan open‑source RISC‑V processor, praised for its ARM‑level performance and detailed micro‑architecture, has sparked worldwide discussion after a George Hotz tweet, highlighting China’s systematic development of a high‑performance, community‑driven CPU project that began in 2019 and continues to evolve.
The article reports that the XiangShan open‑source RISC‑V CPU has attracted massive attention online, with over 500 k reads and discussion on Hacker News after a tweet by George Hotz, the president of Comma AI.
Performance-wise, the latest "Kunming Lake" version achieved a SPECint2006 score of 45 at 3 GHz, comparable to ARM Neoverse N2, making it the strongest known open‑source processor.
The project was launched in 2019 under the Chinese Academy of Sciences, with the first tape‑out (Yanque Lake) in July 2021, followed by the "Nanhu" and "Kunming Lake" generations; collaborations include Tencent, Alibaba, ZTE and others.
Technically, XiangShan is written in Chisel and features an out‑of‑order six‑issue pipeline, separated load/store pipelines, L1/L2/L3 caches, and a modular backend consisting of CtrlBlock, IntBlock, FloatBlock, MemBlock and a VectorBlock for the third generation.
The processor is released under the Mulan Permissive License v2, encouraging community contributions, and is part of the broader "One Life One Chip" initiative that trains students to design and tape‑out a 64‑bit RISC‑V SoC capable of running Linux.
International observers note that while the chip is not yet in mass production and follows a roughly six‑month micro‑architecture iteration cycle, the project demonstrates China’s systematic effort to build core engineering capabilities.
IT Services Circle
Delivering cutting-edge internet insights and practical learning resources. We're a passionate and principled IT media platform.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.