--- 格式版本: 2 标题: "Blue Lion Supercomputer Will Run on NVIDIA Vera Rubin | NVIDIA Blog" 原文链接: "https://blogs.nvidia.com/blog/blue-lion-vera-rubin/" 发布日期: "2026-06-30" 发布时间校准状态: "found" 发布时间来源: "llm:strict_original_body" 发布时间证据: "time class=related-news-date nvidia-article-date datetime=2026-06-30T08:00:57-07:00: Jun 30, 2026" 发布时间校准原因: "该日期来自HTML metadata中的article-date字段，位于标题附近，符合文章发布时间的特征。" 发布时间校准置信度: "100" 发布时间候选数量: 33 发布时间严格候选数量: 12 发布时间原页读取状态: "原页面来自已抓取 HTML" 发布时间未找到原因: "候选日期无效或 LLM 未确认" 发布时间校准时间: "2026-07-20T11:49:33+08:00" 发现时间: "2026-07-20T09:25:40+08:00" 入库时间: "2026-07-20T03:58:34.928Z" 来源平台: "NVIDIA Blog 搜索" 搜索渠道: "source_template" 搜索词: "https://blogs.nvidia.com/?s=Vera%20Rubin" 匹配关键词: - "Vera Rubin" - "GPU" 相关厂家: - "NVIDIA" 相关专家: [] 内容类型: "网页" 抓取工具: "AgentKey Scrape" 清洗工具: "AgentKey Markdown + LLM 正文裁剪" 原始附件: [] AI优质: "否" AI打分: 72 AI分档: "召回候选" AI质检状态: "不通过" AI打分理由: "NVIDIA官方发布，确认Vera Rubin架构及Blue Lion/Doudna超算部署，来源权威且新颖。但内容偏向HPC超算与科学计算，对AI Rack/超节点机柜级架构、供电散热互连等细节展开不足，技术深度有限，故评为召回候选。" AI质检模型: "qwen3.6-plus" AI质检时间: "2026-07-20T11:58:34+08:00" AI主题相关性: 12 AI来源权威性: 14 AI新颖性: 16 AI技术细节: 10 AI商业部署信号: 10 AI完整性: 10 图片摘要: - "✗ ./assets/img-83d2f103.jpg | other | 通用AI基础设施/能效配图，未展示Blue Lion或Vera Rubin具体信息" - "★ ./assets/img-1f6575f1.png | diagram | 展示Vera Rubin架构中GPU与Vera CPU的协同工作及任务分配流程。" - "✗ ./assets/img-3511966c.jpg | photo | NVIDIA品牌宣传照，与正文技术内容无关" - "✗ ./assets/img-e47e99dc.png | other | 通用芯片渲染图，Alt文本指向推理软件栈，与正文硬件架构主题不符" 采集批次: "2026年7月20日9点23分34秒" 采集批次ID: "20260720-092334-062" 去重键: "https://blogs.nvidia.com/blog/blue-lion-vera-rubin" --- Germany’s [Leibniz Supercomputing Centre](https://www.lrz.de/en/), LRZ, is [gaining a new supercomputer](https://www.lrz.de/en/news/detail/2024-12-13-bluelion-en) that delivers roughly 30x more computing power compared with SuperMUC-NG, the current LRZ high-performance computer. It’s called Blue Lion. And it will run on the NVIDIA Vera Rubin architecture. That’s new. Until now, LRZ — part of the Gauss Centre for Supercomputing, Germany’s leading HPC institution — had only said its next system would use “next-generation” NVIDIA accelerators and processors. We’re confirming it: That next generation is Vera Rubin, NVIDIA’s upcoming platform for AI and accelerated science. If the name sounds familiar, it should. Last month, [Lawrence Berkeley National Lab unveiled Doudna](https://blogs.nvidia.com/blog/dell-nvidia-berkeley-doudna/), its next flagship system that will also be powered by Vera Rubin. Two continents. Two systems. Same architecture. ## What Is Vera Rubin? Vera Rubin is a superchip. It combines: - Rubin GPU — the successor to NVIDIA Blackwell - Vera CPU — NVIDIA’s first custom CPU, built to work in lockstep with the GPU Together, they form a platform built to collapse simulation, data and AI into a single, high-bandwidth, low-latency engine for science. It combines shared memory, coherent compute and in-network acceleration — and is launching in the second half of 2026. Blue Lion is built to meet it. ## About the System [HPE is building](https://www.hpe.com/us/en/newsroom/press-release/2024/12/hewlett-packard-enterprise-builds-next-generation-100-direct-liquid-cooled-supercomputer-at-the-leibniz-supercomputing-center.html) [Blue Lion](https://www.hpe.com/us/en/newsroom/press-release/2024/12/hewlett-packard-enterprise-builds-next-generation-100-direct-liquid-cooled-supercomputer-at-the-leibniz-supercomputing-center.html). It will use next-generation HPE Cray technology and feature NVIDIA GPUs in a system equipped with powerful storage and interconnect that harnesses HPE’s 100% fanless direct liquid-cooling systems architecture, which uses warm water delivered through pipes to efficiently cool the supercomputer. It’s built for researchers working on climate, turbulence, physics and machine learning, with workflows that blend classic simulation and modern AI. Jobs can scale across the entire system. Heat from the racks will be reused to warm nearby buildings. And it’s not just local. Blue Lion will support collaborative research projects across Europe. ## About the Doudna Supercomputer Meanwhile, in Berkeley, California, Doudna, the U.S. Department of Energy’s next supercomputer, will also run Vera Rubin. [Built by Dell Technologies](https://www.dell.com/en-us/dt/corporate/newsroom/announcements/detailpage.press-releases~usa~2025~05~dell-technologies-to-power-the-department-of-energy-s-next-supercomputer.htm#/filter-on/Country:en-us), the supercomputer is named for Nobel laureate and CRISPR pioneer Jennifer Doudna and will serve over 11,000 researchers when it launches next year. Doudna will be wired for real-time workflows and optimized for science results per joule of energy. Data streams in from telescopes, genome sequencers and fusion experiments and lands directly in the system via [NVIDIA Quantum-X800 InfiniBand](https://www.nvidia.com/en-us/networking/products/infiniband/quantum-x800/) networking. Processing starts instantly. Feedback loops are live. It’s designed to advance fusion energy, materials discovery and biology faster. Compared with its predecessor, it’s expected to deliver 10x more application performance, using just 2-3x the power. That’s 3-5x better performance per watt. ## Why This All Matters Blue Lion and Doudna aren’t just big machines. They’re signals of what comes next: a shift in how high-performance systems are designed, used and connected. AI is no longer an add-on. Simulation isn’t a silo. Data isn’t parked — it moves. Science is becoming a real-time discipline. The systems that power it need to keep up. Vera Rubin is built for that. ![AI Innovators Adopt NVIDIA Vera — Why Max Single-Threaded CPU at Scale Matters](./assets/img-1f6575f1.png)