AI 自主设计芯片：从规格到硬件仅需两周
An AI system can now take a chip from a written spec to working hardware on its…
An AI system can now take a chip from a written spec to working hardware on its own, and this paper argues that is what finally makes chip design fast, not better code generation.
一个AI系统现在可以独立将芯片从书面规格转化为可工作的硬件，本文认为这才是最终让芯片设计变快的关键，而非更好的代码生成。
Redwood is a small, low-power inference chip. 2 human architects wrote the spec, and the AI system generated everything below it: the chip design, the tests, the firmware, and the kernels.
Redwood是一款小型、低功耗的推理芯片。2位人类架构师编写了规格，AI系统生成了其下所有内容：芯片设计、测试、固件和内核。
That means a spec change was back on real hardware in under 48 hours.
这意味着规格变更在不到48小时内就能回到真实硬件上运行。
Redwood ran Qwen3-0.6B on an AMD FPGA board at 12.1 tokens/s, against 28 on an NVIDIA Jetson Orin Nano. Built as a real chip on Samsung 8 nm, the team projects 49 tokens/s at 1.335 W, or 3.4x more tokens per watt than the Jetson.
Redwood在AMD FPGA板上以12.1 tokens/s的速度运行Qwen3-0.6B，而NVIDIA Jetson Orin Nano为28 tokens/s。作为三星8纳米工艺的真实芯片制造，团队预计在1.335瓦功耗下可达49 tokens/s，即每瓦token数比Jetson高3.4倍。
More importantly, the speed did not come from skipping tests. Every block reached 95% coverage, and the first design sent to the FPGA had 0 bugs.
更重要的是，速度并非通过跳过测试实现。每个模块都达到了95%的覆盖率，且首次发送到FPGA的设计零缺陷。
Hardware could start moving at the pace of model releases, instead of models waiting on hardware.
硬件可以开始以模型发布的速度前进，而不是模型等待硬件。
– arxiv. org/abs/2608.26418
– arxiv.org/abs/2608.26418
Title: "Redwood: A Frontier AI Accelerator Designed, Verified, and Deployed from Scratch in 2 Weeks by AI"
标题：“Redwood：一款由AI在2周内从零设计、验证并部署的前沿AI加速器”
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力