Tag: AI applications

  • Nvidia RTX Pro 6000 GPU: 96GB VRAM for Desktops, 24GB for Laptops

    Nvidia RTX Pro 6000 GPU: 96GB VRAM for Desktops, 24GB for Laptops

    Key Takeaways

    1. The RTX Pro 6000 is designed for professionals, featuring 96GB of GDDR7 VRAM and a bandwidth of 1.6 TB/s, surpassing the GeForce RTX 5090’s 32GB VRAM.
    2. It excels in AI workloads, rivaling AMD’s Ryzen Strix Halo, and is built for managing large AI models efficiently.
    3. The GPU has a thermal design power (TDP) of 400 to 600 watts and supports advanced technologies like PCIe 5.0 and DisplayPort 2.1.
    4. A laptop version of the RTX Pro 6000 is available with 24GB of VRAM, while Nvidia offers budget-friendly options with the RTX Pro 3000, 2000, 1000, and 500 series.
    5. The RTX Pro 6000 is expected to start shipping in April, with pre-built systems available from Dell, HP, and Lenovo in May, but pricing details have not yet been revealed.

    The RTX Pro 6000 marks a new high point for Nvidia’s graphics cards aimed at professionals. This GPU is mainly made for AI tasks, game creators, and other expert users who require a substantial amount of video memory. In comparison, the Nvidia GeForce RTX 5090 has “only” 32GB of GDDR7 VRAM, whereas the desktop and server editions of the RTX Pro 6000 boast an impressive 96GB of GDDR7 along with a bandwidth of 1.6 TB/s.

    Competing in AI Workloads

    With its 96GB of VRAM, the RTX Pro 6000 rivals AMD’s Ryzen Strix Halo when it comes to handling AI jobs, and this graphics card is expected to manage large AI models at a significantly quicker pace. The GPU operates with a thermal design power (TDP) ranging from 400 to 600 watts, and it supports modern technologies like PCIe 5.0 and DisplayPort 2.1. The sleeker Max-Q version may catch the eye of those looking to install multiple graphics cards within the same PC case.

    Laptop and Other Options

    Nvidia also provides a laptop version of the RTX Pro 6000, although this variant is capped at 24GB of VRAM, similar to the GeForce RTX 5090 for laptops. Additionally, Nvidia offers a range of more budget-friendly professional GPUs, like the RTX Pro 3000, 2000, 1000, and 500, which are built on the Blackwell architecture. However, Nvidia has not yet disclosed specifics about the CUDA core count or clock speeds for these new RTX Pro graphics cards.

    Release Timeline

    As of now, Nvidia has not announced the official pricing for its latest professional graphics cards. The RTX Pro 6000 is anticipated to begin shipping in April, while pre-built systems from Dell, HP, and Lenovo are expected to be available starting in May.

    Source:
    Link

    Nvidia RTX Pro 6000 GPU: 96GB VRAM for Desktops, 24GB for LaptopsNvidia RTX Pro 6000 GPU: 96GB VRAM for Desktops, 24GB for Laptops

     

    Tags: , ,
  • CEO of Baidu Cautions Against Overemphasizing Launching New LLMs Exclusively in China

    CEO of Baidu Cautions Against Overemphasizing Launching New LLMs Exclusively in China

    Baidu CEO Robin Li Yanhong has raised concerns about the current focus on creating large language models (LLMs) in China's tech industry. Li argues that this trend not only drains resources but also misses the opportunity to advance practical AI applications.

    The Value of AI Lies in Its Application

    Speaking at the X-Lake Forum in Shenzhen, Li highlighted the surge in AI model development in China, with 238 models launched as of October 2023. However, he noted a lack of AI-native applications that fully utilize the unique capabilities of AI, similar to how mobile-native apps revolutionized smartphone usage.

    Li believes that the true value of AI lies in its application, rather than just the development of foundational models. He suggests a shift in focus towards creating a wide range of AI-native applications, comparable to the era that saw the rise of popular apps like WeChat, Douyin, and Uber, which were specifically designed for mobile usage.

    Balancing Foundational Research and Application Development

    Li's remarks also address a broader issue within the industry – the balance between foundational research and application development. While foundational models are crucial for advancing AI technology, their full potential is only realized when applied in practical, everyday scenarios.

    Li further criticizes the trend of hoarding advanced semiconductors and building intelligent computing centers, which he deems an inefficient approach to AI development. He emphasizes the importance of having the right scale and training datasets to yield models with emergent abilities – the ability to perform complex tasks with minimal input.

    Baidu's Role in AI Application Development

    Baidu, the company led by Li himself, is actively engaged in developing AI applications. One example is Comate, a code-writing assistant. However, Li believes that the best AI-native applications, both in China and globally, are yet to come.

    In conclusion, Li Yanhong, the CEO of Baidu, highlights the need to shift the focus from solely creating large language models to developing a wide range of AI-native applications. He emphasizes that the value of AI lies in its application and urges the industry to strike a balance between foundational research and practical development. Baidu itself is involved in AI application development, but according to Li, there is still much more potential to be unlocked in this field.

    Tags: , ,