Skip to main content

5 posts tagged with "Vera CPU"

NVIDIA Vera CPU for AI factory

View all tags

Vera CPU 拿下 Modal 首单:英伟达的 Arm CPU 开始抢 x86 的活

· 3 min read
Industry Research Team

据 10 月 19 日路透社报道,AI 云计算初创 Modal 宣布将首次采用英伟达新一代 Vera CPU,为 AI 智能体提供托管服务。这是 Vera CPU 公开的首个独立于 Rubin 平台的客户订单——英伟达的 CPU 第一次被当作「产品」而非「配件」采购。

一、订单要点​

  • 客户:Modal(AI 沙盒与智能体托管服务,为智能体安全运行生成的代码提供托管容器)
  • 性能口径:早期测试显示,高并发负载下 Vera 的任务吞吐量最高达可比 x86 硬件的 2 倍,同时保持较低的单任务延迟
  • 部署方式:不弃 x86——Vera 与现有英特尔/AMD 芯片并行部署,按任务特点调度
  • 代价:需要改造部分软件与基础设施以支持 Arm 兼容架构

二、Vera CPU 是什么​

88 个自研 Olympus 核心(Armv9.2)、176 线程、最高 1.5TB LPDDR5X、1.2 TB/s 内存带宽,通过 NVLink-C2C(1.8 TB/s)与 GPU 一致性互联。单体大芯片设计避免跨 die 延迟,详见卡片。

此前 Vera CPU 的定位是 Vera Rubin 平台的「主机 CPU」——负责数据搬运调度与系统编排。Modal 订单标志着它的独立价值开始显现。

三、为什么 AI 代理需要更好的 CPU​

这是 2026 下半年最重要的结构性变化之一:智能体工作负载(任务编排、API 调用、数据查询、每用户常驻虚拟机)主要由 CPU 承担。用户越多、后台任务越多,CPU 需求越高:

  • Intel CEO 承认其 CPU 产能只能满足约一半的 AI 相关需求
  • Meta 计划部署数千万 AWS Graviton 核心处理 AI 代理负载

Modal 的场景(AI 沙盒 = 大量短生命周期容器)恰好是高并发、低单任务延迟的典型 CPU 负载——Vera 的吞吐优势在这里最有说服力。

四、数据中心 CPU 格局的连锁反应​

  1. Graviton(AWS)、Axion(谷歌)、Vera(英伟达)、磐久(阿里):自研 Arm CPU 在云侧全面铺开,x86 双雄在 AI 负载上的份额被从边缘蚕食
  2. 英伟达的「全栈桶」更满了:GPU + CPU + 网络 + 软件,客户整套机房都可以只找英伟达——议价时其他供应商要拼的不再是一颗芯片而是整个平台
  3. Arm 生态在数据中心形成正循环:每多一个 Vera 级客户,软件适配的边际成本就降一分

延伸阅读:GPU/ASIC/CPU 三分天下 · GTC 柏林综述 · 完整对比表

Vera CPU 智能体基准实测:10.8秒拉起千个沙盒,速度达 AMD EPYC 8倍

· 4 min read
Industry Research Team

快科技10月10日消息,AI 代码执行平台 Daytona 公布了最新一轮智能体 AI 基准测试结果:NVIDIA Vera 处理器在沙盒启动速度和任务吞吐量两项核心指标上,均明显优于采用 AMD EPYC "Zen 5" 架构的产品。对正处爆发期的 Agentic AI 行业而言,这是第一次有一组公开数据,把"多智能体场景下的 CPU 选型"这个新命题摆上了台面。

Vera CPU 深度:SCF + 1.5TB LPDDR5X,NVIDIA 自研 CPU 的野心

· 7 min read
Industry Research Team

当所有人盯着 Rubin R200 的 50 PFLOPS FP4 时,真正决定这台机器推理经济性的,可能是一颗 CPU。Vera——NVIDIA 首次全面自研核心的数据中心 CPU,用 88 个 Olympus 核心、第二代 SCF 一致性结构和 1.5TB LPDDR5X,重新定义了"AI 工厂的主机处理器"。本文依据站内芯片数据库与智能体基准实测,拆解 NVIDIA 从外购 Arm 核到自研 SCF 的演进逻辑。

从 Grace 到 Vera:自研之路的三级跳​

NVIDIA 的 CPU 战略经历了三个阶段:GH200/GB200 时代的 Grace CPU 基于 72 核 Neoverse V2 公版 Arm 核心,本质是"外购核心 + 自研互联";GB300 延续了这一代 Grace;而 2026 年 GTC 发布的 Vera 则完成质变——核心、片上互连、内存子系统全部自研。据站内芯片数据库,Vera 的核心规格如下:

规格参数
CPU 架构ARM 架构(自研 Olympus 核心)
指令集ARMv9.2(完全兼容)
核心数量88 个 Olympus 核心
线程数量176 个线程(空间多线程)
单核性能前代产品的 2 倍
最大内存容量1.5 TB(LPDDR5X)
内存带宽1.2 TB/s
互联技术NVLink-C2C(1.8 TB/s)
片上互连第二代 SCF(3.4 TB/s 对分带宽)
发布时间2026 年 3 月 17 日
量产时间2026 年下半年

SCF:88 核无延迟扩展的底座​

SCF(Scalable Coherency Fabric,可扩展一致性结构)是 Vera 最值得细看的技术。第二代 SCF 提供 3.4 TB/s 对分带宽,采用片上网格 + 统一缓存设计,可无延迟扩展 88 个核心。

对分带宽(bisection bandwidth)是衡量片上互连质量的关键指标:它表示把芯片切成两半时,两半之间的总通信带宽。88 核的大芯片如果对分带宽不足,核心越多内耗越大。第二代 SCF 用网格拓扑让任意核心间的通信路径趋于均匀,配合统一缓存避免一致性流量的长距离穿梭——这正是 Vera 敢于采用单体计算芯片设计(而非多小芯片拼接)的底气:避免跨小芯片通信延迟,在所有核心满载时保持稳定延迟和吞吐量,性能可预测性强。

1.5TB LPDDR5X:把内存子系统当主武器​

Vera 将 SCF 与高达 1.5TB 的 LPDDR5X 内存子系统结合。对比上一代 Grace 的规格变化一目了然:

指标Grace(GB200/GB300)Vera提升
内存容量480GB LPDDR5X最高 1.5TB LPDDR5X约 3 倍
内存带宽512 GB/s1.2 TB/s约 2.3 倍
CPU-GPU 互联NV-HBI 900 GB/sNVLink-C2C 1.8 TB/s2 倍
核心数72 核 Neoverse V288 核 Olympus+22%

更关键的是系统级设计:LPDDR5X 与 GPU 的 HBM4 可视为单一一致性内存池。这意味着数据不再需要在"CPU 内存"和"显存"之间显式搬运,两个物理上分离的内存子系统在软件视角下是一体的。对推理业务,这带来两个直接收益:KV 缓存卸载——不常用的注意力上下文可以下沉到 LPDDR5X 侧,把昂贵的 HBM4 带宽留给热点数据;多模型高效执行——多个模型实例共享同一内存池,切换与加载开销大幅降低。

Daytona 实测:智能体时代的 CPU 门槛​

2026 年 10 月 10 日公布的 Daytona 智能体基准测试,给 Vera 的实际能力提供了量化注脚。智能体负载的特点是海量并发沙盒的创建与销毁——恰恰考验 CPU 的内存容量、一致性与调度效率:

测试项VeraAMD EPYC 9755差距
1000 沙盒上线10.8 秒88.4 秒约 8 倍
峰值智能体任务吞吐13.3 任务/秒——
2000 沙盒上线27.4 秒——

8 倍的沙盒上线速度差距说明:智能体负载对 CPU 的要求已与通用计算分道扬镳——容量大(1.5TB 装下海量沙盒状态)、带宽高(1.2TB/s 喂饱并发)、一致性结构强(88 核高效协作)。站内芯片数据库同时记录,Vera 的软件环境运行速度相比传统架构 CPU 提升最高 50%,满负载下强化学习评估周期可缩短 50%。

与 Groq 3 LPX 的异构分工​

Vera 所在的 Rubin 平台还有一位非常规队友:NVIDIA 收购 Groq 后推出的 Groq 3 LPX。据站内芯片数据库,LPX 机架集成 256 颗 LPU,聚合 128GB 片上 SRAM,聚合带宽高达 40 PB/s,官方口径的能效优势显著(35× perf/W vs H100 推理)。在 Rubin 平台的分工中,GPU 承担训练与 prefill,Groq 3 LPX 专攻智能体解码阶段的低延迟输出——而 Vera 则是这一切的调度中枢。prefill 吃算力、decode 吃带宽、调度吃一致性与容量,三种负载由三种架构分别消化,这正是"每 token 经济学"推导出的异构架构必然性。

Vera vs EPYC Venice:服务器 CPU 的新战场​

Vera 的出现让 NVIDIA 与 AMD 在服务器 CPU 领域正面相遇。据站内芯片数据库,AMD EPYC Venice(2026 CES 发布)是全球首款 2nm 数据中心 CPU:

指标NVIDIA VeraAMD EPYC Venice(Zen 6C)
核心88 个自研 Olympus最多 256 个 Zen 6C
制程未公开台积电 2nm(N2)
内存1.5TB LPDDR5X16 通道 DDR5
单路带宽1.2 TB/s1.6 TB/s
L3 缓存—1GB
卖点NVLink-C2C、CPU-GPU 一致性通用性、x86 生态

两者的定位差异明显:EPYC Venice 是通用数据中心 CPU 的极致(256 核、x86 生态、Helios 机架中与 MI455X 组队),而 Vera 是 AI 专用 CPU——它不做通用服务器生意,只为 AI 工厂的主机负载优化。Vera 甚至支持硬件级 FP8 计算,是全球首款支持 FP8 精度的 CPU,并支持全机密计算的硬件级安全隔离。竞争的天平取决于一个问题:AI 工厂的 CPU 采购,会不会整体从通用服务器预算中剥离?如果会,Vera 绑定 Rubin GPU 的销售模式将是护城河;如果不会,EPYC 的生态广度仍难撼动。

小结​

Vera CPU 标志着 NVIDIA 完成了从"用别人的核心攒超级芯片"到"自研全套主机平台"的跃迁:SCF 的 3.4TB/s 对分带宽解决了 88 核扩展难题,1.5TB LPDDR5X 与 HBM4 构成的一致性内存池重构了推理数据流,Daytona 实测的 8 倍沙盒优势则证明智能体负载确实需要一颗不一样的 CPU。它的野心不在挑战 x86 的通用服务器王座,而是重新划定一条边界——AI 基础设施中,CPU 是 GPU 子系统的一部分,而不是独立采购的商品。这条边界能否成立,将决定未来五年服务器市场的版图。

NVIDIA Vera Rubin Officially Ships: First VR200 NVL72 Delivered, Samsung HBM4 Mass Production, Rubin Ultra Cabinet Sky-High Price

· 5 min read
Industry Research Team

July 2026, NVIDIA's next-gen AI compute platform Vera Rubin officially began its first shipments, succeeding the Blackwell architecture, with large-scale mass production planned for H2 2026. First customers include Microsoft, Google, Amazon, Meta, Oracle, and other large cloud providers.

1. World's First VR200 NVL72 Delivered (Milestone)​

CoreWeave jointly with Dell announced that the world's first NVIDIA Vera Rubin VR200 NVL72 cabinet has been officially delivered and passed the L11 full-cabinet hardware diagnostics on the first try. This marks Rubin's move from roadmap to physical product, with no major bottlenecks in core supply-chain links (HBM4, advanced packaging, liquid cooling, ultra-high-power power supply).

VR200 NVL72 Core Configuration​

MetricVera Rubin VR200 NVL72
Cabinet codenameOberon
GPU72 Rubin GPUs
CPU36 Vera CPUs
Per-GPU memory288 GB HBM4
Per-CPU memory1.5 TB LPDDR5X
Total cabinet HBM420.7 TB (20,736 GB)
Total cabinet LPDDR5X54 TB
InterconnectNVLink 6 full mesh
Inference performance~3.6 exaFLOPS class
CoolingLiquid cooling
Generational improvement~3.5× per-GPU compute, ~2.8× memory bandwidth (vs Blackwell)

Vera CPU integrates 88 custom Olympus ARM cores, with 1.8 TB/s interconnect to the GPU, usable as a GPU memory expansion pool. NVIDIA completed its first Vera CPU deliveries to Anthropic, OpenAI, xAI, and Oracle Cloud in May.

2. Samsung HBM4 Mass Production: Key Bottleneck Eases​

July 8, 2026, Samsung Electronics officially started HBM4 mass production for the Vera Rubin platform, with reported HBM4 mass-production yield reaching 70% (above the initial 60-65% expectation). Confirmation of this key supply-chain link clears obstacles for Rubin's large-scale deployment.

HBM Supply Landscape (2026 Q1)Share
SK hynix45%
Samsung40%
Micron15%

HBM4 uses 8-layer stacking (12-layer design planned for 2028), priced at about 2.8× HBM3e. TrendForce predicts HBM supply will grow 65% annually, with HBM4 reaching 35% of total output by 2027 Q4.

3. Rubin Ultra Sky-High Price: HBM Cost Dominates​

Per BofA Global Research estimates, the Rubin generation will push single-server cost to a new high:

Cost ItemRubin VR200 (Oberon)Comparison
Cabinet HBM4 usage20,736 GB—
HBM4 unit price~$18.40 / GBBlackwell (HBM3e) ~$11.26 / GB
HBM4 cost alone~$382KExcluding LPDDR5X
Rubin Ultra cabinet estimated price~$21MITHome / BofA estimate

4. Rubin Ultra Design Change: Original 4-die Cancelled (per SemiAnalysis)​

Semiconductor research firm SemiAnalysis (2026-06-30) disclosed that the original 4-die Rubin Ultra GPU unveiled at GTC 2026 has been cancelled; the version actually shipping in 2027 is roughly halved in scale and performance:

  • Reason for cancellation: The original integrated 4 compute dies + 16 HBM4E in a single CoWoS-L package; the substrate warped under the 4-die config, causing compute-die-to-substrate contact failure and yield collapse; the alternative CoPoS won't reach mass production until after late 2028, missing the 2027 node.
  • New approach: Changed to dual-die (same construction as standard Rubin) + HBM4E, ~384 GB HBM4E per GPU (higher than standard Rubin's 288 GB), but total compute and bandwidth only half the original; to approach the original's aggregate compute, NVIDIA plans to assemble "2+2" board-level configs within the Kyber rack to reach four-die equivalent scale.
  • Kyber rack delay: The companion Kyber NVL144 rack is delayed 12+ months to 2028 due to midplane PCB manufacturing difficulties; the 800V DC power scheme is likewise delayed to 2028.

⚠️ Note: NVIDIA has not commented officially on the above design change; some on X argue "the chip count hasn't changed, it's old news reheated." This section is compiled from SemiAnalysis public reports, subject to final NVIDIA disclosure. We have marked "specs pending official confirmation" on the Rubin Ultra preview card.

Industry Interpretation​

  1. "Never doubt" moment realized: Rubin's first delivery passed L11 on the first try, dispelling market doubts about "Rubin delay," locking in H2 2026 AI compute supply certainty ahead of time.
  2. Designed for Agentic AI: Rubin targets agentic workflows and ultra-long-context inference, further lowering the training/inference cost curve for trillion-parameter models.
  3. HBM is the full-chain winner: 20.7 TB HBM4 per cabinet is enormous usage; SK hynix, Samsung, Micron, advanced packaging (CoWoS-L), liquid cooling, and power retrofitting all benefit across the chain, while also becoming the biggest cost and capacity constraint.

References​


This article continuously tracks Vera Rubin mass-production ramp and HBM4 supply-chain dynamics.

NVIDIA Vera Rubin Enters Full Production: The Agentic AI Factory Era Begins

· 5 min read
Industry Research Team

On June 1, 2026, NVIDIA founder and CEO Jensen Huang officially announced at COMPUTEX 2026 (Taipei) that: the Vera Rubin platform has entered full production. This marks a fundamental paradigm shift for AI hardware from "discrete accelerators" to "integrated AI factories."

Key Highlights​

  • Rubin GPU: Next-gen AI compute chip, FP4 compute is 3.6× that of Blackwell
  • Vera CPU: 88 custom Arm cores (176 threads), replacing the Grace CPU
  • NVLink 6: GPU-to-GPU interconnect bandwidth reaches 260 TB/s (double Blackwell)
  • CX8 SuperNIC: 800Gb/s network, ConnectX-9 link reaching 28.8 TB/s
  • HBM4 memory: 288GB per chip, 13 TB/s bandwidth
  • Agentic throughput: 10× over Grace Blackwell

Complete Vera Rubin Platform Specs​

Vera Rubin is not a single GPU but a complete AI factory platform comprising 7 chips:

ChipTypePurpose
Rubin GPUMain AI compute chipTraining + inference
Rubin Ultra GPUFlagship versionUltra-scale inference
Vera CPUCPU paired with RubinHost CPU + data preprocessing
NVLink 6Interconnect chipHigh-speed GPU interconnect (260 TB/s)
CX8 SuperNICNIC800Gb/s network
XDR 800G switchDatacenter networkCross-rack communication
Rubin Platform PODWhole cabinetPre-configured AI factory (144 GPUs)

Rubin GPU Detailed Specs (estimated)​

ParameterRubin GPURubin UltraBlackwell (B200)
ArchitectureRubinRubin UltraBlackwell
ProcessTSMC 3nm (est.)TSMC 3nmTSMC 4NP
Memory288GB HBM4288GB HBM4E (est.)192GB HBM3e
Memory bandwidth13 TB/s13+ TB/s8 TB/s
FP4 compute~3,600 TFLOPS (est.)~5,000 TFLOPS (est.)2,250 TFLOPS
TDP1,000W (est.)1,200W (est.)700-1000W
InterconnectNVLink 6 (260 TB/s)NVLink 6NVLink 5 (1800 GB/s)
Mass production2026 Q3H2 20272024 Q4

📌 Note: Rubin's exact specs are not fully public yet; some values above are estimates.

Vera CPU: The New Host CPU Replacing Grace​

Vera CPU is NVIDIA's self-designed Arm-architecture CPU, replacing the previous Grace CPU:

ParameterVera CPUGrace CPU
Cores88 cores (176 threads)72 cores (144 threads)
ArchitectureCustom Armv9 (est.)Arm Neoverse V2
InterfaceNVLink 5.0 (1.8 TB/s)NVLink 4.0 (900 GB/s)
TDP~500W (est.)350-500W
PurposeAI factory Host CPUHPC / AI Host

Key upgrade: Vera's co-design with the Rubin GPU achieves end-to-end optimization in compute, data loading, and preprocessing, comparable to Google TPU 8t's Arm Axion integration.

Performance vs Blackwell​

NVIDIA officially claims that under the same POD configuration (144 GPU chips):

MetricGrace Blackwell (GB200 NVL72)Vera Rubin NVL144Improvement
FP4 compute1.1 PFLOPS3.6 PFLOPS3.3×
Memory capacity288GB×72 = 20.7TB288GB×144 = 41.4TB2×
Memory bandwidth8 TB/s×7213 TB/s×144~3.3×
NVLink bandwidth1800 GB/s×72260 TB/s (full POD)~2×
Agentic throughputBaseline10×10×
Performance per wattBaseline25× (vs CPU alone)25×

💡 Why "10× agentic throughput"? Agentic AI workloads differ from training/inference: one prompt may trigger multiple stages including reasoning, retrieval, tool calls, and response generation, involving thousands of steps. The Rubin platform is optimized for this long-chain, high-concurrency workload.

MGX Third-Gen Rack-Scale System​

Vera Rubin adopts the MGX third-gen open rack-scale system design:

  • Five-rack synergy: Vera Rubin NVL72 system + Vera CPU + Groq 3 LPX + Vera BlueField-4 STX storage + Spectrum-6 SPX Ethernet
  • Global supply chain: 30 countries, 350+ factories, hundreds of partners (Dell, HPE, Lenovo, Supermicro, Asus, Foxconn, etc.)
  • Spectrum-X Ethernet silicon photonics: World's first switch based on CPO (co-packaged optics) supporting 200Gb/s SerDes, now in mass production

Mass Production Timeline​

TimeEvent
Jan 2026CES 2026 first unveils Rubin platform
June 1, 2026COMPUTEX 2026 announces full production
Fall 2026Vera Rubin officially starts mass production and shipment
H2 2027Rubin Ultra launch (HBM4E upgrade)
2028Feynman architecture (next gen)

AI Factory: From Selling Chips to Selling "Smart Production Lines"​

Huang said something at the launch that shook the industry:

"Rubin's Agentic AI throughput is 10× that of Blackwell. Rubin is a complete AI factory platform."

This marks a fundamental shift in NVIDIA's business model:

  • Past: Sold GPUs (H100/B200), customers built systems themselves
  • Now: Sells "complete AI factory solutions" (Vera Rubin POD), including GPU, CPU, network, storage, software stack
  • Future: Becomes the "TSMC" of global AI infrastructure (providing smart production capacity)

vs Competitors​

VendorProductPositioningAdvantageDisadvantage
NVIDIAVera RubinComplete AI factory solutionMost complete ecosystem, most mature softwareExpensive, extremely high power
AMDMI455X (MI400 series)Training competitorPrice/performance, open ecosystemSoftware ecosystem gap
GoogleTPU 8i/8tCloud training/inferenceDeep Gemini integrationGoogle Cloud only
HuaweiAscend 910C/950Domestic substitutionChina localization, AscendMind frameworkAffected by export controls

Industry Impact​

  1. AI labs: Frontier model training time shrinks from "months" to "weeks"
  2. Cloud providers: Must decide whether to procure Vera Rubin POD (conflicts with self-developed chip strategy)
  3. Hyperscale datacenters: AI factory becomes a new competitive dimension (whoever has the strongest compute can train the strongest model)
  4. Domestic chips: Ascend 910C/950, Cambricon MLU590, etc. must catch up to Blackwell in 2026-2027, or the gap will widen to the Rubin era

References​


This article is compiled from NVIDIA official announcements and public materials. Some specs are estimates, subject to final official release.