跳到正文
Hacker News-首页· peter_d_sherman·· 3 小时前AI 评分50

Qualcomm Cloud AI 100 Ultra – 150W AI 加速卡(配備 128GB 及 548GB/s 带宽)

Qualcomm Cloud AI 100 Ultra – 150W AI Accelerator Card with 128GB at 548GB/s

AI 导读

Qualcomm 推出了 Cloud AI 100 Ultra 加速卡,這是一款針對生成式 AI 和 LLM 設計的 150W AI 推理解決方案。該產品配備 128GB LPR4x 內存、548 GB/s 內存带宽以及 576 MB 的 on-die SRAM。其峰值性能可達 870 TOPS(INT8)與 288 TFLOPS(FP16)。

正文 · 原文

Cloud AI 100 Ultra

Image 1

Looking for our data center roadmap?

Image 2

See Qualcomm Dragonfly™ products

Cloud AI 100 Ultra

Contact sales

On this page

OverviewQualcomm® AI Inference Suite for Cloud and On-PremVideoSpecifications

Delivers the performance and power efficiency necessary to deploy and accelerate AI inference at-scale

The Qualcomm Cloud AI 100 Ultra, the newest member of our portfolio of cloud artificial intelligence (AI) inference cards, is a performance- and cost-optimized AI inference solution, designed for Generative AI and large language models (LLMs). With up to 576 MB of on-die SRAM and 64 AI cores per card - and programmability for a wide range of workloads and acceleration techniques - Qualcomm Cloud AI 100 solutions address the unique requirements for scaling classic and generative AI workloads, ranging from computer vision and natural language processing to transformer-based LLMs.

Qualcomm Cloud AI is a product of Qualcomm Technologies, Inc., and/or its subsidiaries.

Product license agreement

Featured DocumentsCloud AI 100 Ultra Product BriefAI Inference Suite

Qualcomm® AI Inference Suite for Cloud and On-Prem

The Qualcomm AI Inference Suite is a comprehensive set of software and services for Qualcomm Cloud AI accelerators spanning across on-premises solutions and cloud deployments.

It includes ready-to-use AI applications and agents, tools, and libraries for operationalizing generative AI at scale. For enterprises developing chatbots, co-pilots, and AI agents, the AI Inference Suite features a rich set of OpenAI-compatible APIs including user management and administration capabilities. It supports chat, image generation, multi-modal AI capabilities, and retrieval-augmented generation (RAG) functionalities.

The suite also supports integration with popular generative AI models, frameworks, cloud services, and is deployed using Kubernetes or bare-metal containers. Customers can build gen AI applications and agents using open-source or proprietary models, scale and optimize workflows, and run enterprise-grade solutions.

Learn More

Image 3

Cloud AI 100 Ultra

01:26

Image 4: Cloud AI 100 Ultra

Specifications

SKU Form factor Power (TDP) Peak Integer Ops (INT8) Peak FP Ops (FP16) SRAM DRAM (w/ ECC) DRAM BW Host Interface
AI 100 Ultra PCIe FH3/4L (111.2mm x 237.9mm) 150W Up to 870 TOPS Up to 288 TFLOPS 576 MB 128 GB LPR4x 548 GB/s PCIe Gen4, 16 lanes
AI 80 Ultra PCIe FH3/4L (111.2mm x 237.9mm) 150W Up to 618 TOPS (155 TOPs/SoC) Up to 222 TFLOPS (56 TFLOPS/SoC) 576 MB 128 GB LPR4x 548 GB/s PCIe Gen4, 16 lanes
AI 100 Pro PCIe HHHL (68.9mm x 169.5mm) 75W Up to 400 TOPS Up to 200 TFLOPS 144 MB 32 GB LPR4x 137 GB/s PCIe Gen4, 8 lanes
AI 100 Standard PCIe HHHL (68.9mm x 169.5mm) 75W Up to 350 TOPS Up to 175 TFLOPS 126 MB 16 GB LPR4x 137 GB/s PCIe Gen4, 8 lanes
AI 80 Standard PCIe HHHL (68.9mm x 169.5mm) 75W Up to 190 TOPS Up to 86 TFLOPS 126 MB 16 GB LPR4x 137 GB/s PCIe Gen4, 8 lanes

Image 5

Access more.

To access more Cloud AI 100 Ultra resources, you need to be a member of a verified company. Company verification is initiated by filling out our access request form.

RegisterLearn More

Already have a Qualcomm account?Sign in

Quick links

Company info

Stay connected

Get the latest Qualcomm and industry information delivered to your inbox.

SubscribeManage your subscription

Also of Interest

: English (US)

Languages

This website processes personal data through our and third parties’ online tracking technologies, including analytics and advertising cookies. To learn more about how we and our affiliates within the Qualcomm Group may use your personal data and cookies, please review the Privacy Policy linked at the bottom of this website and Qualcomm's Cookie Policy. If you don’t want to share your website activities, including browsing behavior, with our third-party partners via these tracking technologies, click on “Cookie Settings" button below to update your preferences. You can also update your cookie preferences at any time by clicking the "Cookie Settings" link at the bottom of this website. By browsing or navigating the site or by clicking “I Understand,” you agree to our TERMS OF USE, PRIVACY POLICY (linked at the bottom of this website), and COOKIE POLICY.

IMPORTANT: Our Terms of Use contain dispute resolution provisions, including binding arbitration and class action waiver clauses, that could affect your legal rights.

Cookies Settings

I Understand

Image 6: Company Logo

Opt-Out Request Honored

Privacy Preference Center

We respect your privacy right and preference. Please select and accept your preference as described below. You can click on the different category headings to learn more about each option. Blocking some types of cookies may impact your experience of this website and the services we may be able to offer. For more details, please visit the Privacy Policy at the bottom of this website.

Allow All Manage Consent Preferences

Strictly Necessary Cookies

Always Active

These cookies are necessary for the website to function and cannot be switched off in our systems. They are usually only set in response to actions made by you which amount to a request for services, such as setting your privacy preferences, logging in or filling in forms. You can set your browser to block or alert you about these cookies, but some parts of the site will not then work.

Analytics Cookies

  • Analytics Cookies

These cookies allow us to count visits and traffic sources so we can measure and improve the performance of our site. They help us to know which pages are the most and least popular and see how visitors move around the site. If you do not allow these cookies we will not know when you have visited our site, and will not be able to monitor its performance.

Personalization Cookies

  • Personalization Cookies

These cookies enable the website to provide enhanced functionality and personalisation. They may be set by us or by third party providers whose services we have added to our pages. If you do not allow these cookies then some or all of these services may not function properly.

Targeting Cookies

  • Targeting Cookies

These cookies may be set through our site by our advertising partners. They may be used by those companies to build a profile of your interests and show you relevant adverts on other sites. If you do not allow these cookies, you will experience less targeted advertising.

Cookie List

Clear

    • checkbox label label

Apply Cancel

Consent Leg.Interest

  • checkbox label label

  • checkbox label label

  • checkbox label label

Confirm My Choices

Image 7: Powered by Onetrust

来源:Hacker News-首页 · qualcomm.com