#Computex2024: Lunar Lake puts Intel back in AI PC race

At Computex Taiwan, Intel CEO Pat Gelsinger unveils a wall of AI PCs running on Intel technology. During the event in June 2024, Intel unveiled cutting-edge technologies and architectures poised to accelerate the artificial intelligence ecosystem. Source: Intel Corporation

Lunar Lake's microarchitecture combines two distinct tiles interconnected by Intel's advanced Foveros packaging technology.

This shift to system-on-chip (SoC) packaging will finally do away with separate memory sticks or chips, integrating 16GB or 32GB of LPDDR5X memory directly on the dye.

#Computex2024: AMD and Qualcomm ignite the AI revolution at Intel’s expense

Lindsey Schutters  3 Jun 2024

Intel is doing this to reduce power consumption by roughly 40% because the electrons don’t need to be pushed off the processor to external components.

The company is not leaving power users who may need more hardware flexibility entirely behind, though. Intel is promising laptop CPUs based on Arrow Lake architecture later this year.

Streamlined cores

The heart of Lunar Lake lies in its Compute Tile, housing the latest generation of efficient-cores (E-cores) and performance-cores (P-cores).

These cores come with remarkable improvement in thermal efficiency compared to their predecessors, validating AMD's decision to stick with x86 architecture.

Lunar Lake's E-cores are optimised for low power consumption, making them ideal for everyday tasks and applications that demand extended battery life.

Conversely, the P-cores are engineered to handle demanding workloads, ensuring seamless multitasking and responsiveness for resource-intensive applications.

This dual approach aims to strike a balance between power efficiency and raw performance, catering to a broader spectrum of user needs, but improves on thermal performance where Meteor Lake failed to deliver on its promises.

The Compute Tile also incorporates the new Xe2 graphics processing unit (GPU), neural processing unit (NPU 4), and image processing unit (IPU).

These components collectively enhance graphic capabilities, accelerate AI computations, and refine image processing, ultimately enriching the overall user experience.

RIP hyper-threading

Intel has also walked away from hyper-threading. This decision finally accepts that SMT technology consumes more power and real estate than its benefits warrant.

With the E-cores now compact and capable, hyper-threading no longer makes sense, according to Intel.

Of course, the goal with Lunar Lake architecture is to increase AI PC capabilities, prioritising power efficiency and AI integration.

Your next computer will be AI capable, and that's fine

Lindsey Schutters  10 Apr 2024

This emphasis on power optimisation aligns with the growing demand for energy-efficient devices, particularly in the mobile computing sphere.

Microsoft promises that deeper integration of AI capabilities directly into the processor architecture could potentially enable a wider array of AI-powered applications and workloads.

It’s encouraging that Intel developed Lunar Lake in such a short time after kicking off the AI PC arms race with its Core Ultra Meteor Lake chips.

Lunar Lake-based Windows 11 AI PCs will now finally receive a free update to Copilot+ PC experiences and close the hardware gap that opened with its competitors.