Jump to content

UHQBot

Forum Bot
 View Profile  See their activity

  • Posts

    39,330

  • Joined

  • Last visited

  • Days Won

    25

 Content Type 

Posts posted by UHQBot

    Moving Pictures: NVIDIA, Getty Images Collaborate on Generative AI

    in PC Hardware

    Posted

    As a sports commentator for a professional lacrosse team, Grant Farhall knows the value in having the right teammates.

    As the chief product officer for Getty Images, a global visual-content creator and marketplace, he believes the collaboration between his company and NVIDIA is an excellent pairing for taking generative AI to the next level.

    The companies aim to develop two generative AI models using NVIDIA Picasso, part of the new NVIDIA AI Foundations cloud services. Users could employ the models to create a custom image or video in seconds, simply by typing in a concept.

    “With our high quality and often unique imagery and videos, this collaboration will give our customers the ability to create a greater variety of visuals than ever before,  helping creatives and non-creatives alike fuel visual storytelling,” Farhall said.

    Getty Images is a unique partner, not only for its stunning images and video, but also its rich metadata, with appropriate rights. Its creative team and research bring a wealth of expertise that can deliver impactful outputs.

    For artists, generative AI adds a new tool that expands their canvas. For content creators, it’s an opportunity to create a custom visual tailored to a brand or business they’re building.

    “More often than not, it’s a visual that cuts through the noise of a busy world to capture your attention, and being able to stand out from the crowd is crucial for businesses of all shapes and sizes,” Farhall said.

    Building Responsible AI

    But, as in lacrosse, you need to play by the rules.

    The models will be trained on Getty Images’ fully licensed content, and revenue generated from the models will provide royalties to content creators.

    “Both companies want to develop these tools in a responsible way that returns benefits to creators and doesn’t pass risks on to customers, and this collaboration is testament to the fact that’s possible,” he said.

    A Time-Tested Relationship

    It’s not the first inning for this collaboration.

    “We’ve been fostering and growing a relationship for some time — NVIDIA brings the tech expertise and talent, and we bring the high quality and unique content and marketplace,” said Farhall.

    The technology, values and connections are catalysts for experiences that wow creators and users. It’s a feeling Farhall shares, sitting in front of his mic on a Saturday night.

    “There’s an adrenaline rush when the live action of a game becomes your singular focus and you’re just in the moment,” he said.

    And by training a custom model with NVIDIA Picasso, Getty Images and NVIDIA aim to help storytellers everywhere create more moments that perfectly capture their audiences’ attention.

    To learn more about what NVIDIA is doing in generative AI and beyond, watch company founder and CEO Jensen Huang’s GTC keynote below.

    Image at top courtesy Roberto Moiola/Sysaworld/Getty Images.

    View the full article

    Mind the Gap: Large Language Models Get Smarter With Enterprise Data

    in PC Hardware

    Posted

    Large language models available today are incredibly knowledgeable, but act like time capsules — the information they capture is limited to the data available when they were first trained. If trained a year ago, for example, an LLM powering an enterprise’s AI chatbot won’t know about the latest products and services at the business.

    With the NVIDIA NeMo service, part of the newly announced NVIDIA AI Foundations family of cloud services, enterprises can close the gap by augmenting their LLMs with proprietary data, enabling them to frequently update a model’s knowledge base without having to further train it — or start from scratch.

    This new functionality in the NeMo service enables large language models to retrieve accurate information from proprietary data sources and generate conversational, human-like answers to user queries. With this capability, enterprises can use NeMo to customize large language models with regularly updated, domain-specific knowledge for their applications.

    This can help enterprises keep up with a constantly changing landscape across inventory, services and more, unlocking capabilities such as highly accurate AI chatbots, enterprise search engines and market intelligence tools.

    NeMo includes the ability to cite sources for the language model’s responses, increasing user trust in the output. Developers using NeMo can also set up guardrails to define the AI’s area of expertise, providing better control over the generated responses.

    Quantiphi — an AI-first digital engineering solutions and platforms company and one of NVIDIA’s service delivery partners — is working with NeMo to build a modular generative AI solution called baioniq that will help enterprises build customized LLMs to boost worker productivity. Its developer teams are creating tools that let users search up-to-date information across unstructured text, images and tables in seconds.

    Bringing Dark Data Into the Light

    Analysts estimate that around two-thirds of enterprise data is untapped. This so-called dark data is unused partly because it’s difficult to glean meaningful insights from vast troves of information. Now, with NeMo, businesses can retrieve insights from this data using natural language queries.

    NeMo can help enterprises build models that can learn from and react to an evolving knowledge base — independent of the dataset that the model was originally trained on. Rather than needing to retrain an LLM to account for new information, NeMo can tap enterprise data sources for up-to-date details. Additional information can be added to expand the model’s knowledge base without modifying its core capabilities of language processing and text generation.

    Enterprises can also use NeMo to build guardrails so that generative AI applications don’t provide opinions on topics outside their defined area of expertise.

    Enabling a New Wave of Generative AI Applications for Enterprises

    By customizing an LLM with business data, enterprises can make their AI applications agile and responsive to new developments. 

    • Chatbots: Many enterprises already use AI chatbots to power basic customer interactions on their websites. With NeMo, companies could build virtual subject-matter experts specific to their domains.
    • Customer service: Companies could update NeMo models with details about their latest products, helping live service representatives more easily answer customer questions with precise, up-to-date information.
    • Enterprise search: Businesses have a wealth of knowledge across the organization, including technical documentation, company policies and IT support articles. Employees could query a NeMo-powered internal search engine to retrieve information faster and more easily.
    • Market intelligence: The financial industry collects insights about global markets, public companies and economic trends. By connecting an LLM to a regularly updated database, investors and other experts could quickly identify useful details from a large set of information, such as regulatory documents, recordings of earnings calls or financial statements.

    Enterprises interested in adding generative AI capabilities to their applications can apply for early access to the NeMo service.

    Watch NVIDIA founder and CEO Jensen Huang discuss NVIDIA AI Foundations in the keynote address at NVIDIA GTC, running online through Thursday, March 23:

    View the full article

    Green Light: NVIDIA Grace CPU Paves Fast Lane to Energy-Efficient Computing for Every Data Center

    in PC Hardware

    Posted

    The results are in, and they point to a new era in energy-efficient computing.

    In tests of real workloads, the NVIDIA Grace CPU Superchip scored 2x performance gains over x86 processors at the same power envelope across major data center CPU applications. That opens up a whole new set of opportunities.

    It means data centers can handle twice as much peak traffic. They can slash their power bills by as much as half. They can pack more punch into the confined spaces at the edge of their networks — or any combination of the above.

    Energy Efficiency, a Data Center Priority

    Data center managers need these options to thrive in today’s energy-efficient era.

    Moore’s law is effectively dead. Physics no longer lets engineers pack more transistors in the same space at the same power.

    That’s why new x86 CPUs typically offer gains over prior generations of less than 30%. It’s also why a growing number of data centers are power capped.

    With the added threat of global warming, data centers don’t have the luxury of expanding their power, but they still need to respond to the growing demands for computing.

    Wanted: Same Power, More Performance

    Compute demand is growing 10% a year in the U.S., and will double in the eight years from 2022-2030, according to a McKinsey study.

    “Pressure to make data centers sustainable is therefore high, and some regulators and governments are imposing sustainability standards on newly built data centers,” it said.

    With the end of Moore’s law, the data center’s progress in computing efficiency has stalled, according to a survey that McKinsey cited (see chart below).

    Power efficiency gains have stalled in data centers, McKinsey said.

    In today’s environment, the 2x gains NVIDIA Grace offers are the eye-popping equivalent of a multi-generational leap. It meets the requirements of today’s data center executives.

    Zac Smith — the head of edge infrastructure at Equinix, a global service provider that manages more than 240 data centers — articulated these needs in an article about energy-efficient computing.

    “The performance you get for the carbon impact you have is what we need to drive toward,” he said.

    “We have 10,000 customers counting on us for help with this journey. They demand more data and more intelligence, often with AI, and they want it in a sustainable way,” he added.

    A Trio of CPU Innovations

    The Grace CPU delivers that efficient performance thanks to three innovations.

    It uses an ultra-fast fabric to connect 72 Arm Neoverse V2 cores in a single die that sports 3.2 terabytes per second in fabric bisection bandwidth, a standard measure of throughput. Then it connects two of those dies in a superchip  package with the NVIDIA NVLink-C2C interconnect, delivering 900 GB/s of bandwidth.

    Finally, it’s the first data center CPU to use server-class LPDDR5X memory. That provides up to 50% more memory bandwidth at similar cost but one-eighth the power of typical server memory. And its compact size enables 2x the density of typical card-based memory designs.

    The Grace CPU is simpler and more energy efficient than current x86 CPUsCompared to current x86 CPUs, NVIDIA Grace is a simpler design that offers more bandwidth and uses less power.

    The First Results Are In

    NVIDIA engineers are running real data center workloads on Grace today.

    They found that compared to the leading x86 CPUs in data centers using the same power footprint, Grace is:

    • 2.3x faster for microservices,
    • 2x faster in memory intensive data processing
    • and 1.9 x faster in computational fluid dynamics, used in many technical computing apps.

    Data centers usually have to wait two or more CPU generations to get these benefits, summarized in the chart below.

    Grace outperforms x86 CPUsNet gains (in light green) are the product of server-to-server advances (in dark green) and additional Grace servers that fit in the same x86 power envelope (middle bar) thanks to the energy efficiency of Grace.

    Even before these results on working CPUs, users responded to the innovations in Grace.

    The Los Alamos National Laboratory announced in May it will use Grace in Venado, a 10 exaflop AI supercomputer that will advance the lab’s work in areas such as materials science and renewable energy. Meanwhile, data centers in Europe and Asia are evaluating Grace for their workloads.

    NVIDIA Grace is sampling now with production in the second half of the year. ASUS, Atos, GIGABYTE, Hewlett Packard Enterprise, QCT, Supermicro, Wistron and ZT Systems are building servers that use it.

    Go Deep on Sustainable Computing

    To dive into the details, read this whitepaper on the Grace architecture.

    Learn more about sustainable computing from this session at NVIDIA GTC (March 20-23, free with registration): Three Strategies to Maximize Your Organization’s Sustainability and Success in an End-to-End AI World.

    Read a whitepaper about the NVIDIA BlueField DPU to find out how to build energy-efficient networks.

    And watch NVIDIA founder and CEO Jensen Huang’s GTC keynote to get the big picture.

    View the full article

    NVIDIA Announces Microsoft, Tencent, Baidu Adopting CV-CUDA for Computer Vision AI

    in PC Hardware

    Posted

    Microsoft, Tencent and Baidu are adopting NVIDIA CV-CUDA for computer vision AI.

    NVIDIA CEO Jensen Huang highlighted work in content understanding, visual search and deep learning Tuesday as he announced the beta release for NVIDIA’s CV-CUDA — an open-source, GPU-accelerated library for computer vision at cloud scale.

    “Eighty percent of internet traffic is video, user-generated video content is driving significant growth and consuming massive amounts of power,” said Huang in his keynote at NVIDIA’s GTC technology conference. “We should accelerate all video processing and reclaim the power.”

    CV-CUDA promises to help companies across the world build and scale end-to-end, AI-based computer vision and image processing pipelines on GPUs.

    Optimizing Internet-Scale Visual Computing With AI

    The majority of internet traffic is video and image data, driving incredible scale in applications such as content creation, visual search and recommendation, and mapping.

    These applications use a specialized, recurring set of computer vision and image-processing algorithms to process image and video data before and after they’re processed by neural networks.

    bing.jpgMicrosoft Bing’s Visual Search Engine uses AI Computer Vision
    to search for images (dog food, for example) within images on the Internet.

    While neural networks are normally GPU accelerated, the computer vision and image processing algorithms that support them are often CPU bottlenecks in today’s AI applications.

    CV-CUDA helps process 4x as many streams on a single GPU by transitioning the pre- and post-processing steps from CPU to GPU. In effect, it processes the same workloads at a quarter of the cloud-computing cost.

    The CV-CUDA library provides developers more than 30 high-performance computer vision algorithms with native Python APIs and zero-copy integration with the PyTorch, TensorFlow2, ONNX and TensorRT machine learning frameworks.

    The result is higher throughput, reduced computing cost and a smaller carbon footprint for cloud AI businesses.

    Global Adoption for Computer Vision AI

    Adoption by industry leaders around the globe highlights the benefits and versatility of CV-CUDA for a growing number of large-scale visual applications. Companies with massive image processing workloads can save tens to hundreds of millions of dollars.

    Microsoft is working to integrate CV-CUDA into Bing Visual Search, which lets users search the web using an image instead of text to find similar images, products and web pages.

    In 2019, Microsoft shared at GTC how they’re using NVIDIA technologies to help bring speech recognition, intelligent answers, text to speech technology and object detection together seamlessly and in real time.

    Tencent has deployed CV-CUDA to accelerate its ad creation and content understanding pipelines, which process more than 300,000 videos per day.

    The Shenzhen-based multimedia conglomerate has achieved a 20% reduction in energy and cost for image processing over their previous GPU-optimized pipelines.

    And Beijing-based search giant Baidu is integrating CV-CUDA into FastDeploy, one of the open-source deployment toolkits of the PaddlePaddle Deep Learning Framework, which enables seamless computer vision acceleration to developers in the open-source community.

    From Content Creation to Automotive Use Cases

    Applications for CV-CUDA are growing. More than 500 companies have reached out with over 100 use cases in just the first few months of the alpha release.

    In content creation and e-commerce, images use pre- and post-processing operators to help recommender engines recognize, locate and curate content.

    In mapping, video ingested from mapping survey vehicles requires preprocessing and post-processing operators to train neural networks in the cloud to identify infrastructure and road features.

    In infrastructure applications for self-driving simulation and validation software, CV-CUDA enables GPU acceleration for algorithms that are already occurring in the vehicle, such as color conversion, distortion correction, convolution and bilateral filtering.

    Looking to the future, generative AI is transforming the world of video content creation and curation, allowing creators to reach a global audience.

    New York-based startup Runway has integrated CV-CUDA, alleviating a critical bottleneck in preprocessing high-resolution videos in their video object segmentation model.

    Implementing CV-CUDA led to a 3.6x speedup, enabling Runway to optimize real-time, click-to-content responses across its suite of creation tools.

    “For creators, every second it takes to bring an idea to life counts,” said Cristóbal Valenzuela, co-founder and CEO of Runway. “The difference CV-CUDA makes is incredibly meaningful for the millions of creators using our tools.”

    To access CV-CUDA, visit the CV-CUDA GitHub.

    Or learn more by checking out the GTC sessions featuring CV-CUDA. Registration is free.

    View the full article

×
×
  • Create New...

Important Information

By using this site, you agree to our Guidelines Privacy Policy.

  I accept