These AI Accelerator Cards Hope To Be The Next 3dfx

PCWorld
13 Jan 202403:48

TLDRAt CES 2024, Lenovo's ThinkPad Neo Ultra was showcased, featuring the first PCI Express-based AI accelerator cards, including Memory X's MX3 and a product from Canara. These cards aim to enhance AI processing power for desktops, with the MX3 capable of 10 teraflops per chip, rivaling major tech companies' offerings. The AI accelerators are primarily used for vision processing and running large language models, potentially transforming PC applications and gaming. This marks the beginning of a new era for AI technology in PCs, with many vendors competing to become the next big player in the field.

Takeaways

  • πŸ˜ƒ CES 2024 introduces the first era of AI accelerator cards for PCs, marking a significant advancement in desktop computing technology.
  • πŸ’» Lenovo unveils the ThinkPad Neo Ultra, a standalone desktop featuring the first PCI Express-based AI accelerator cards, priced at $1,000 and set to ship in June.
  • πŸ” Two pioneering AI accelerator cards are highlighted: MemoryX and a product by Canara, aiming to boost desktop AI processing capabilities.
  • πŸ‘¨β€πŸ’» MemoryX's MX3, an M.2 form factor PCI Express card, is equipped with four chips, each capable of 10 teraflops of AI processing power, totaling around 40 to 50 teraflops.
  • πŸ“Έ The MemoryX card is demonstrated as a Vision Processor, with a use case featuring avatar representation of a cameraman, showcasing its potential applications in vision processing.
  • πŸ€– These AI accelerator cards are designed for a variety of applications, including local language models, chatbots, and potentially enhancing gaming experiences.
  • πŸ“š PCWorld.com is mentioned as a source for more detailed information about the capabilities and specifications of these new AI technologies.
  • πŸš€ Lenovo hints at a burgeoning market with up to a dozen PC card vendors competing in the AI technology space, likening the potential impact to the historical influence of 3Dfx on 3D graphics.
  • 🧠 The performance and scalability of these AI accelerator cards are still in early stages, with expectations of growth following Moore's Law and increasing application integration.
  • πŸ•Ή The emergence of AI accelerator cards like MemoryX and Canara represents the dawn of a new era in computing, reminiscent of the transformative early days of 3D gaming graphics.

Q & A

  • What is the significance of the AI accelerator cards introduced at CES 2024?

    -The AI accelerator cards introduced at CES 2024 mark the beginning of a new era for PC hardware, specifically designed to accelerate AI processing. They are the first PCI Express based AI accelerator cards, indicating a shift towards dedicated AI processing capabilities for desktop computers.

  • Who is presenting this information about AI accelerator cards?

    -Mark Hawkman from PC World is presenting the information about the new AI accelerator cards at CES 2024.

  • What is the Lenovo ThinkPad Neo Ultra?

    -The Lenovo ThinkPad Neo Ultra is a standalone desktop that is showcased at CES 2024. It is priced at about 1,000 dollars and is set to ship in June. It features the first PCI Express based AI accelerator cards.

  • What are the two AI accelerator cards mentioned in the script?

    -The two AI accelerator cards mentioned are the Memory X and a product from a company called Canara.

  • What is the Memory X MX3?

    -The Memory X MX3 is a PCI Express card in an M.2 form factor. It contains four chips, each capable of performing about 10 teraflops of AI processing power.

  • How does the Memory X MX3 compare to AI capabilities of Intel, Qualcomm, and AMD processors?

    -The Memory X MX3, with each chip capable of 10 teraflops, is in the same neighborhood as the AI capabilities of Intel, Qualcomm, and AMD processors, which are discussing capabilities in the range of 40 to 60 teraflops.

  • What is the primary use of the Memory X AI accelerator card?

    -The Memory X AI accelerator card is primarily being used as a Vision Processor. It can interpret visual data, such as camera feeds, and is also applicable for running large language models, chatbots, and potentially gaming applications.

  • How many PC card vendors are competing to be the next rendition of 3dfx in the AI space according to Lenovo?

    -Lenovo has stated that there are about 8 to 10, perhaps even a dozen, PC card vendors competing to become the next major player in the AI technology space, similar to what 3dfx was for 3D graphics.

  • What does the future hold for AI accelerator cards according to the script?

    -The future of AI accelerator cards is still uncertain, but with Moore's Law coming into play, the performance and capabilities are expected to improve significantly. This could lead to more sophisticated applications and a new era of AI technology for PCs.

  • How can one learn more about the Memory X and Canara AI accelerator cards?

    -For more detailed information about the Memory X and Canara AI accelerator cards, one can visit pcworld.com to read about the specifications and offerings of these new chips.

  • What is the expected price and shipping time for the Lenovo ThinkPad Neo Ultra?

    -The Lenovo ThinkPad Neo Ultra is expected to cost around 1,000 dollars and is scheduled to start shipping in June.

Outlines

00:00

πŸš€ Introduction to AI Accelerator Cards at CES 2024

The script opens with a welcome to the new era of AI accelerator cards for PCs, with Mark Hawkman from PC World at CES 2024. The focus is on the Lenovo ThinkPad Neo Ultra, a standalone desktop expected to ship in June at a price point of around $1,000. The excitement revolves around the first PCI Express based AI accelerator cards, with two main ones showcased: Memory X and a product from Canara. The discussion emphasizes the current lack of desktop AI acceleration options and how startups are stepping in to fill this gap. The Memory X MX3 is highlighted as a PCI Express card with an m.2 form factor, featuring four chips capable of 10 teraflops of AI processing power each.

Mindmap

Keywords

πŸ’‘AI accelerator cards

AI accelerator cards are specialized hardware designed to speed up artificial intelligence tasks, such as machine learning and deep learning. In the context of the video, these cards are being introduced for PCs to enhance their AI processing capabilities. The video discusses the first generation of these cards, highlighting their potential to revolutionize PC performance in AI-related tasks.

πŸ’‘PCI Express

PCI Express (Peripheral Component Interconnect Express) is a high-speed interface standard for connecting a computer to various hardware devices. It is used in the video to describe the type of connection the new AI accelerator cards use to interface with the PC, indicating a faster and more efficient data transfer rate compared to older technologies.

πŸ’‘Memory X MX3

Memory X MX3 is a specific model of AI accelerator card discussed in the video. It is a PCI Express card in the M.2 form factor, containing four chips capable of performing approximately 10 teraflops of AI processing power each. This represents a new class of hardware aimed at providing significant computational resources for AI tasks on personal computers.

πŸ’‘Teraflops

Teraflops (tera floating point operations per second) is a unit of measurement for computational speed, indicating the ability of a computer to perform a trillion (10^12) floating-point operations per second. In the context of the video, teraflops is used to quantify the processing power of AI accelerator cards, comparing their performance with other industry standards.

πŸ’‘Vision Processor

A vision processor is a type of AI accelerator designed specifically for processing and analyzing visual data, such as images and videos. In the video, the AI accelerator cards are described as being used primarily as vision processors, which implies their application in tasks like image recognition, video analysis, and other vision-related AI functions.

πŸ’‘Large language models

Large language models refer to advanced AI systems capable of understanding, generating, and predicting human language on a massive scale. These models can be used for various applications, including chatbots, translation services, and content creation. In the video, the AI accelerator cards are mentioned as potential hardware to run such models locally on a PC, enhancing their performance and responsiveness.

πŸ’‘Chatbots

Chatbots are computer programs designed to simulate conversation with human users, typically over the internet. They can be used for customer service, information provision, or entertainment. In the context of the video, chatbots are mentioned as one of the applications that could benefit from the AI processing power provided by the new accelerator cards, potentially leading to more sophisticated and responsive chatbot interactions.

πŸ’‘Moris Law

Moris Law, often misspelled as Moore's Law, is an observation made by Gordon Moore, co-founder of Intel, that the number of transistors on a microchip doubles approximately every two years, leading to an increase in computing power while costs remain the same. In the video, this concept is used to suggest that as AI accelerator cards continue to evolve, their performance will improve exponentially, leading to more powerful and efficient AI capabilities.

πŸ’‘3dfx

3dfx was a company known for its 3D graphics accelerator cards in the 1990s, which played a significant role in the development of 3D graphics in computer games and other applications. In the video, the comparison is made to suggest that the new AI accelerator cards could have a similar transformative impact on AI technology as 3dfx had on 3D graphics.

πŸ’‘Canara

Canara is mentioned in the video as one of the companies competing in the emerging market of AI accelerator cards for PCs. While not much detail is provided about Canara, its mention signifies the existence of a competitive landscape in the AI hardware space, with multiple vendors aiming to establish themselves as leaders in this new era of technology.

πŸ’‘Mark Hawkman

Mark Hawkman is the reporter from PC World who is presenting the information in the video. He provides insights into the new era of AI accelerator cards and their potential impact on the PC industry. His role is to inform viewers about the latest developments and to offer a perspective on the significance of these technological advancements.

Highlights

Introduction to the new era of AI accelerator cards for PCs at CES 2024.

Mark Hawkman from PC World introduces the Lenovo ThinkPad Neo Ultra.

The ThinkPad Neo Ultra is a standalone desktop priced at approximately $1,000, shipping in June.

Lenovo showcases the first PCI Express based AI accelerator cards.

Two AI accelerator cards are featured: Memory X and a product by Canara.

AI accelerator cards fill the gap for desktop AI acceleration where current processors fall short.

Memory X's MX3 is a PCI Express card with an m.2 form factor and four chips.

Each chip in the Memory X MX3 can perform about 10 teraflops of AI processing power.

Intel, Qualcomm, and AMD are also developing AI processing capabilities, with a focus on teraflops.

AI accelerator cards are primarily used as Vision Processors and for running large language models.

The Memory X is being used to interpret camera inputs and create avatars in real-time.

AI accelerators are expected to enhance applications such as chatbots and potentially gaming.

The CPU and GPU will work alongside the NPU, with the NPU taking over specific tasks for optimization.

Lenovo suggests there are 8-10, possibly a dozen, PC card vendors aiming to be leaders in AI technology.

The industry is in the early days of AI, with performance and applications still developing.

Moris Law's impact on scaling AI capabilities is expected to lead to more interesting applications in the future.

The potential for AI in gaming and graphics is compared to the early days of 3D effects and movies like Jurassic Park.

Mark Hawkman encourages viewers to stay tuned for more information on AI accelerator cards from Memory X and Canara.

Casual Browsing

Could This App be the NEXT AI Rising Star? | Imagine Art

2024-05-17 11:35:01

This Might be The Next AI Animation Trend | IPIV’s Morph img2vid AnimateDiff Tutorial

2024-05-18 03:20:01

Attempting to Play w/ AI Generated Cards feat. @Hypnocorn

2024-03-31 04:10:00

The End of Lightroom? | Polarr Next Review | Using AI to Edit Photos

2024-04-22 13:05:00

Why Prusa is the next Blackberry

2024-04-06 09:00:00