NVIDIA Launches GH200 Grace Hopper Superchip: A Game Changer In Generative AI Race
NVIDIA Launches GH200 Grace Hopper Superchip: A Game Changer In Generative AI Race
$NVIDIA (NVDA.US)$ has taken the veil off its revolutionary next-generation platform — the GH200 Grace Hopper Superchip Platform, marking a significant development in the field of accelerated computing and generative AI.
$英伟达 (NVDA.US)$ 揭开了其革命性的下一代平台 —— GH200 Grace Hopper Superchip Platform 的面纱，标志着加速计算和生成式人工智能领域的重大发展。
This platform is poised to be a cornerstone for future AI-driven applications, from virtual assistants to advanced data analytics, according to a company press release.
根据公司的一份新闻稿，该平台有望成为未来人工智能驱动的应用程序的基石，从虚拟助手到高级数据分析。
The GH200 stands out as the pioneer to harness the power of the HBM3e processor. Tailored to navigate the intricate maze of generative AI tasks, it comes in a variety of configurations to suit different needs. Its dual configuration, boasting 3.5x more memory and 3x the bandwidth of existing models, is anchored by a single server.
GH200 是利用 HBm3e 处理器强大功能的先驱。它专为在错综复杂的生成式 AI 任务迷宫中航行而量身定制，有多种配置可满足不同的需求。它的双重配置由一台服务器支撑，其内存增加了3.5倍，带宽是现有型号的3倍。
This server is packed with 144 Arm Neoverse cores and a whopping 282GB of the latest HBM3e memory. For investors, this means faster data processing and more efficient algorithmic trading models.
这台服务器装有 144 个 Arm Neoverse 内核和高达 282GB 的最新 HBm3e 内存。对于投资者来说，这意味着更快的数据处理速度和更高效的算法交易模型。
NVIDIA's CEO, Jensen Huang, emphasized the need for specialized computing platforms, stating, "The new GH200 Grace Hopper Superchip platform delivers this with exceptional memory technology and bandwidth."
英伟达首席执行官， 黄仁森，强调了对专用计算平台的需求，他说：“新的 GH200 Grace Hopper Superchip 平台凭借卓越的内存技术和带宽实现了这一目标。”
One of the standout features of the GH200 is its ability to connect with other Superchips via NVIDIA NVLink. This means enhanced collaboration for generative AI tasks, allowing for more complex computations and larger AI models.
GH200 的突出特点之一是它能够通过 NVIDIA NVLink 与其他超级芯片连接。这意味着增强生成式 AI 任务的协作，允许进行更复杂的计算和更大的 AI 模型。
Additionally, the HBM3e memory, which operates at speeds 50% faster than its predecessor, HBM3, ensures rapid data access with a combined bandwidth of 10TB/sec. For businesses, this translates to quicker insights from data and more responsive AI applications.
此外，HBm3e 存储器的运行速度比其前身 HBM3 快 50%，可确保快速的数据访问，总带宽为 10TB/秒。对于企业而言，这意味着可以更快地从数据中获得见解，并提高人工智能应用程序的响应速度。
Nvidia Entered The Trillion-Dollar Club
Nvidia 进入了万亿美元俱乐部
As Nvidia's market capitalization tripled this year, the chipmaker joined the trillion-dollar club in June, alongside $Apple (AAPL.US)$, $Microsoft (MSFT.US)$, $Alphabet-C (GOOG.US)$, and $Amazon (AMZN.US)$.
随着英伟达今年的市值增长了两倍，这家芯片制造商于6月加入了这个价值万亿美元的俱乐部， $苹果 (AAPL.US)$， $微软 (MSFT.US)$， $谷歌-C (GOOG.US)$，以及 $亚马逊 (AMZN.US)$。
On Tuesday, shares fell 1.8% amid a broader market concern sparked by a slew of downgrades on U.S. banks by the rating agency Moody's.
周二，由于评级机构穆迪对美国银行的一系列评级下调，引发了更广泛的市场担忧，股价下跌了1.8％。
Nvidia remains by far the top-performing semiconductor in 2023, up 200% year-to-date, and is the largest holding in the $VanEck Semiconductor ETF (SMH.US)$, with a 20% weight.
英伟达在2023年仍然是迄今为止表现最好的半导体，年初至今上涨了200％，并且是该半导体中最大的持股公司 $半导体指数ETF-VanEck (SMH.US)$，重量为 20%。
This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.
该内容部分是在人工智能工具的帮助下制作的，并由Benzinga编辑审查和发布。