Steve Makris reports on Intel’s new Meteor Lake processor technology which he, among other global media, was shown one month ago at the Intel Tech Tour Malaysia 2023, under embargo. There, Steve Long, Corporate Vice President and General Manager for the Intel Pacific Region introduced Meteor Lake technology, Intel’s first deep artificial intelligence sojourn embedded in a multi-tasking processor for PC’s. Malaysia, which is Intel’s first off-shore site outside the United States 50 years ago, was also first to open it’s doors to highly secure chip building and testing sites to us in Penang Island and Kulim. Long cited unimagined road conditions in Malaysia’s Intel upstart with 100 employees where even Intel co-founder Andy Grove pushed cars through muddy roads. Now with more than 15,000 employees (40% female) Malaysia Intel plays an important “first” role amid Intel’s operations in a dozen countries world wide. We had one full day of immersive Meteor Lake topics and demos, followed by a combined 6km walk in two highly secure chip plants. “We think that seeing is believing and I would trust this exact experience is going to help you understand a few things about our Grovian culture…it’s alive and well and what is driving this transformation”, said Long. Indeed we did.
Today Intel announced the Intel Core Ultra, code named Meteor Lake processor infused with AI technology in a multi-function laptop CPU with smart multiple processing cores and updated game-ready Arc graphics in laptops. Meteor Lake is Intel’s 14th generation of Intel Mobile Processors. Core Ultra will deliver power-efficient AI acceleration and local inference on the PC keeping it working at it’s best and working with third party AI capable apps or the cloud for efficient handling of demanding apps like graphics, video, audio editing, highe-res gaming while monitoring power use and extending battery life. Meteor Lake/ Core Ultra launches Dec 14.
“AI represents a generational shift, giving rise to a new era of global expansion where computing is even more foundational to a better future for all,” said Intel CEO Pat Gelsinger. “For developers, this creates massive societal and business opportunities to push the boundaries of what’s possible, to create solutions to the world’s biggest challenges and to improve the life of every person on the planet.”
The biggest challenge software companies are facing in implementing AI are standards. Intel is teaming with Microsoft embracing and enabling an open ecosystem. Office applications and Windows Studio Effects are using direct ML enabling Meteor Lake’s NPU (Neural Processing Unit) to work together. Conversely, Adobe whose apps use the CPU and GPU extensively is allowing for ML-based IP asses AI functions with the Meteor Lake’s NPU.
What Will Meteor Lake Laptops Do?
Up to 40% lower power consumption and memory management in every function of the multipackaged processor is independently controlled. The new NPU AI consumes a fraction of the power the CPU and GPU consume as it performs data-based inference tasks.
Tasks like scheduling a last minute appointment on a busy day, creating an image based on your textual input by subtracting information from a garbled image or removing a singer’s voice from a song does not require as much power as resizing an 8K movie or playing a game with intricate graphics effects at higher resolution and frame rate.
The upgraded game-ready graphics give the user a sense of faster all-round computing as various apps, media and graphics quickly snap on and off the screen. Basically power when you need it and battery life saving when you don’t. Here are some of the features on a Meteor Lake laptap between the CPU, GPU and new NPU AI processors delivering performance at scale.
- Optimizing battery efficiency againt high performance needs by accounting users short and long time history use
- First built-in low power NPU AI engine maximizes AI use of the onslaught of AI-enabled applications and monitoring the laptop for most efficient use of power
- Assigning the right combination of New efficient E-core and performance P-core architecture and new low power island E-cores for maximizing performance with the least battery draw
- Native HDMI 2.1 and DP 2.1 standards
- Latest PCIe Gen 5 storage memory
- Thunderbolt 4
- WiFi 7
- 8K HDR video and AV1 support
- New Intel Arc Xe LPG graphics up to 2X GFX performance/ watt
- App IP’s have independent paths to memory and can independantly power on/off
- Media seperated from graphics for more efficient use of memory
- New high scalability for 128 BGs connectivity
- Improved Intel Thread Director is more responsive and efficient off-loading computing tasks between the new low power island E-cores and the new efficient E-core and performance P-cores.
- Inteligent display has autonomous low rerfesh rate depending on rendered content and dynamic visual enhancements like contrast and brightness based on content.
Foveros Technology: Intel’s Secret Weapon
Intel developed Foveros (from the Greek word meaning fearsome and awesome) advance 3D packaging technology to package chiplets, that provide a specific function on a PC. One such chiplet for examlpe, the new NPU AI component on Meteor Lake is packaged much like Lego pieces along with chiplets such as the CPU, GPU and internal memory components to form a complete system-on-chip (SoC) also called “computer on a chip”.
This results in higher density energy efficient circuits with low latency in chiplets before they are packaged in the finished processor that computer builders install on PC motherboards. Unlike chiplets being placed side buy side Foveros places chiplets on top of each other saving space as well. This technology can be a chip package inside a PC, server, phone, car or pretty much anything with silicon inside. Here is cool video showing how Foveros 3D packaging works. https://www.intel.com/content/www/us/en/silicon-innovations/6-pillars/process.html
Meteor Lake for Laptops, What About Desktops?
Meteor Lake leapfrogged Intel’s historically expected new 14th Generation desktop processor by holding it back in favour of Meteor Lake’s mobile processor launch. Instead, a “refresh” overclocked version of last year’s 13th Gen Raptor Lake CPU albeit with a questionable 14th Gen sticker, is expected to launch soon. But no sexy AI, just more traditional raw performance gains with off course, unlimited power supply and heat.
The Meteor Lake processor is manufactured using the new Intel 4 architecture process, second of five Intel scheduled manufacturing “node” processes successively shrinking the size of billions of transistors and chiplets into smaller spaces. It succeeds last year’s Raptor Lake Intel 7 process technology, using smaller 7 nanometer size transistors using more accurate EUV lithography processes on the silicon waffer and Foveros 3D packaging technology. Additional Intel 3, Intel 20A and Intel 18 Architecture processes will soon follow “on schedule” according to Intel with even more transistor size shrinking through to 2024.
“We will ship ten’s of millions of AI labelled PC’s into the market delivering billions of AI functions throughout 2024. And that’s just the first year” said Intel Michelle Johnston Holthaus Executive VP and GM of the Intel Client Computing Group.
What Other Processors Will Use Intel AI
Intel is incorporating Meteor Lake AI smarts to the upcoming 5th Gen Intel Xeon Scalable processors, code-named Emerald Rapids. It also announced the Intel Developer Cloud, built on a foundation of advanced central processing units (CPUs) that are purpose-built for AI, graphics processing units (GPUs), and Intel Gaudi2 processors for Deep Learning, along with open software and tools easy enough to accommodate experienced scientists or beginner non-technical practitioners everywhere. The cloud development environment also provides access to the latest Intel hardware platforms.
Intel believes that we are at an inflection point betting much of all AI interpretations will take place outside the data centres and the cloud, iright down to personal computers in the near future. Thus their rush to arm Intel PC’s with AI capabilities. “By 2025 more than half of enterprise managed data will be created and processed outside the data centre or cloud, while more than 60% of AI developers are already adding AI to existing applications” said Holthaus.