Wallstreetcn
2024.05.24 00:57
portai
I'm PortAI, I can summarize articles.

Hyper Race | Intel competes for Microsoft's AI PC layout

AI PCs, coveted by giants, upstream chip designers are eager

Facing Qualcomm's betrayal, Intel launched an instant counterattack.

Public information shows that some PC manufacturers are ready to cooperate with Microsoft to launch the AI PC "Copilot+PC". These computers are all equipped with Qualcomm Snapdragon X Elite or Snapdragon X Plus chips based on the Arm architecture.

As the saying goes, "No one can sleep soundly by the side of a crumbling bed." As a core member of the business value alliance "Wintel" that has been leading the industry for more than 20 years, Intel cannot sit idly by Microsoft's "betrayal". Therefore, ahead of Microsoft's "2024 Microsoft Build Conference" on May 22nd, Intel disclosed some information about its first AI laptop chip supporting "Copilot+AI" - Lunar Lake (X86 architecture) on May 21st, and officially announced that it will be launched in the third quarter of 2024.

The Qualcomm Snapdragon X Elite and Snapdragon X Plus chips are not designed for desktops, but are also used for laptops.

Intel also expects to ship over 40 million PC chips with AI capabilities this year. These chips are equipped with neural processing units (NPU) for AI functions.

According to Intel, Lunar Lake's AI performance will far exceed the current Meteor Lake model, with an NPU computing power of over 400 trillion operations per second (TOPS). In comparison, Intel claims that Lunar Lake's performance in Stable Diffusion 1.5 is 1.4 times faster than the Arm-based Qualcomm Snapdragon X Elite, and faster in core performance than AMD Ryzen 7 8840U and Snapdragon X Elite.

Based on official information from Intel and some information channels, Intel's Lunar Lake includes a CPU, integrated Xe2 GPU, and neural processing unit (NPU), making it a high-performance, low-power processor designed for mobile devices, suitable for users who require long battery life and efficient computing power.

Compared to the previous generation product Meteor Lake, Lunar Lake's AI performance will increase by more than 3 times, with NPU performance exceeding 40 TOPS, providing users with a Copilot+ experience, and the platform's computing power exceeding 100 TOPS.

Lunar Lake is designed for thin and light notebooks with a power consumption of 17W, which helps improve the battery life of laptops.

This processor adopts the Lion Cove architecture P-Core and Skymont architecture E-Core, and will use an improved low-power island design on the SoC chip, integrating an NPU with computing power exceeding 45 TOPS. The integrated graphics will use the new Xe2 architecture, which is the low-power version of Battlemage Therefore, Lunar Lake became the debut of the Battlemage GPU. Using a combination of Lion Cove P core and Skymont E core, as well as the Battlemage Xe 2-LPG graphics processor architecture, this hybrid architecture aims to provide better energy efficiency and performance.

Limited information indicates that Lunar Lake may adopt memory chips packaged on the CPU. The benefit of this design is to increase memory bandwidth and reduce power consumption.

The Lunar Lake processor is capable of achieving low power consumption, reportedly due to the use of Intel's 18A process. Intel has previously claimed that they will take a leading position in performance power ratio by 2024.

In terms of specific structure, the CPU configuration of Lunar Lake may include 8 cores, which helps to enhance multitasking capabilities while achieving low power consumption.

Lunar Lake is positioned in the flagship thin and light laptop market, competing with Apple's M series processors (excluding M Pro/Max), emphasizing energy efficiency and integration.

There is currently no more information about this AI processor, but it is highly likely that more technical details about the Lunar Lake processor will be revealed at the Computex in June and Hot Chips conference in August.

Intel's Executive Vice President and General Manager of the Client Computing Business Michelle Johnston Holthaus publicly stated, "With breakthrough energy efficiency, reliable compatibility of the X86 architecture, and a deeply integrated software catalog that combines CPU, GPU, and NPU, we will provide the most competitive solution ever with our AI PC powered by Lunar Lake and Windows 11, integrating client hardware and software."

In addition, Intel has prepared (or is still in progress?) a development schedule for the next generation AI chip Arrow Lake, following the launch of the Lunar Lake processor - as part of the Core Ultra 200 series for mobile and desktop platforms, Arrow Lake is set to be released in the fourth quarter of this year.

The reason why the "Wintel" alliance was so dominant and unshakable back then was mainly due to the strong ecological capabilities of this alliance. Therefore, this time, while Intel is launching Lunar Lake, they also plan to collaborate with more than 20 OEM manufacturers to introduce over 80 new laptops equipped with Lunar Lake processors to the market, with a specific release date expected before the Christmas shopping season this year.

Although Microsoft and Qualcomm are in a back-and-forth - clearly, laptops featuring the AI CPU version based on the Arm architecture of Qualcomm Snapdragon will be launched on June 18 (Surface Pro series), Microsoft is also planning to release AI laptops featuring Intel (including AMD) X86 architecture CPUs later this year Microsoft's Vice President of Windows and Devices, Pavan Davuluri, stated, "Thanks to our close collaboration with Intel, the launch of Lunar Lake will bring significant and substantive improvements in security, battery life, and more. We are very excited about Lunar Lake carrying a 40+ TOPS NPU to the market, providing users with a large-scale Microsoft Copilot+ experience upon its release."

Therefore, for AI PCs, whether using Qualcomm chips with Arm architecture or Intel or AMD chips with X86 architecture, it is a good thing. Once Intel's revealed Lunar Lake processor hits the market, it will undoubtedly continue to expand the global scale of AI PCs