蚁人

英伟达宣布Groq 3 LPX全面投产：Vera Rubin推理效率实现数倍提升_我的网站

一 |     8月25日消息，英伟达宣布，面向交互式AI推理的Groq 3 LPX处理器正式进入全面投产阶段。     该芯片是Vera Rubin平台的关键组件之一，专注于超低延迟Token生成，旨在弥补传统GPU架构在此类场景下的不足。    Farhan Akhtar’s 120 Bahadur, which released on November 21 is slowing down at the box office. The film did see a rise over the first weekend, but it could not carry that momentum forward as the week began. Directed by Razneesh Razy Ghai, this war drama tells the real story of the 120 brave soldiers from the Kumaon Regiment, led by Major Shaitan Singh, a role Farhan plays with complete honesty.           Even with strong word of mouth, the film’s limited commercial appeal seems to be stopping a wider audience from coming to the theatres. The story brings the Battle of Rezang La from the Sino Indian war to the screen with emotional moments and powerful performances, and many viewers have appreciated that effort. Still, the praise has not turned into the numbers the makers were hoping for.     The film opened with a collection of Rs 2.25 crore on day one according to Sacnilk. The weekend looked promising when the earnings jumped to Rs 3.85 crore on Saturday and went up to Rs 4 crore on Sunday. Things slowed down on Monday with the film collecting only Rs 1.4 crore, a drop of sixty five percent, and Tuesday saw only a slight rise with Rs 1.5 crore.          120 Bahadur is based on the events of the Battle of Rezang La which took place during the Sino Indian war in November 1962. This story was earlier adapted in Dharmendra’s classic film Haqeeqat where he played Captain Bahadur Singh, the same character Farhan portrays in the new film. The cast also includes Raashii Khanna, Vivan Bhatena, Ankit Siwach, Dhanveer Singh, Sahib Verma, Sparsh Walia and Eijaz Khan in important roles.Also Read: Farhan Akhtar’s 120 Bahadur Sees a Jump on Day 2, Earns Rs 4.1 Cr。    在Artificial Analysis的基准测试中，Groq 3 LPX搭载Gemma 4 31B模型、处理100,000 Token上下文时，实现了每秒3400个输出Token的吞吐量，创下该模型迄今最高纪录。     针对智能体编程等对延迟敏感的工作负载，其响应速度最高可达近期竞品平台的4倍。    英伟达指出，智能体任务与传统聊天或摘要生成存在本质差异：一次任务可能包含数百甚至数千个推理步骤，智能体需反复读取文件、调用工具、执行代码、校验结果并迭代推理，上下文长度随之累积至数十万Token。    因此，智能体AI带来两个并行的计算需求：一是高效处理持续增长的大规模上下文，二是以极低延迟持续生成每一个输出Token。    Vera Rubin平台通过多芯片协同设计应对这一挑战。其中，Rubin GPU负责大规模训练、推理及上下文处理；Groq 3 LPX则专为单用户交互场景优化Token生成速率；Vera CPU承担工具编排、代码执行、数据处理和模拟等CPU密集型工作；BlueField-4 DPU与Spectrum-6 SPX则分别负责网络与存储加速。    英伟达同步公布了Vera Rubin NVL72在真实智能体负载下的测试结果。基于SemiAnalysis的AgentX工作负载，采用DeepSeek V4 Pro模型，Vera Rubin NVL72每兆瓦（MW）吞吐量最高达到上一代GB300 NVL72的30倍，每Token成本最高降低35倍。     结合Groq 3 LPX的表现，该组合可使智能体在代码编写、测试执行、工具调用和结果验证等连续任务中获得更快的响应速度，从而显著缩短端到端任务完成时间。

二 |     英伟达强调，Groq 3 LPX全面投产的核心目的并非以专用推理芯片取代GPU，而是为Vera Rubin平台补充传统GPU架构在超低延迟Token生成方面的短板。     整个Vera Rubin平台被定位为面向智能体AI时代的AI工厂系统，涵盖Rubin GPU、Vera CPU、Groq 3 LPX、BlueField-4 DPU、Spectrum-6 SPX等七款芯片及五种专用机架，所有组件围绕计算、网络、存储和软件进行深度协同设计，确保各类硬件各司其职，而非由单一处理器包揽全部工作负载。    

Current article:http://www.laojiusidaosenleliao.bond/u02p9v/20260826/8180073.html

Published on:05:04:35