Intel has announced that its highly anticipated Meteor Lake chips will be launching on December 14th. The new chips will be Intel’s first CPU with different chiplets for each component, its first on its Intel 4 process node, and its first with a dedicated AI coprocessor inside. Meteor Lake is being touted as the most power-efficient client processor ever made by Intel, with up to twice the graphics performance compared to previous chips.
One of the key features of Meteor Lake is its use of tiles as building blocks for the CPU and GPU components. This approach, known as heterogeneous computing, allows for separate pieces of silicon to be printed at different sizes and grafted together. While this is a new approach for Intel, other companies like AMD and Qualcomm have been using this method for some time. The advantage of this approach is the ability to selectively cut off power to specific components, reducing overall power consumption.
Meteor Lake also introduces a “low power island” that can run tasks independently. This low power island has its own efficiency CPU core, AI coprocessor, media engine, and memory, making it ideal for low-power workloads. The other processor cores are distributed among a “Compute Tile” for performance and efficiency cores, as well as a separate Graphics Tile. Intel’s enhanced “Thread Director” intelligently manages workloads, pushing tasks to the lowest-power cores first before utilizing higher-power cores.
The GPU in Meteor Lake, called Xe LPG, features dedicated ray-tracing units and up to 8 Xe cores on the chip itself. Intel claims that the GPU offers up to twice the performance per watt compared to previous integrated graphics solutions. It also supports Intel’s XeSS, an intelligent upscaler similar to Nvidia’s DLSS and AMD’s FSR.
Additionally, Meteor Lake includes an AI coprocessor, known as an NPU, in every chip. This NPU allows for more efficient processing of AI tasks, offering options for running AI workloads on the NPU alone or in conjunction with the GPU. While Intel did not provide specific examples of how the NPU will enhance AI capabilities, it mentioned the OpenVINO Inference Engine as a tool to facilitate communication between applications and the NPU.
Intel’s announcement of Meteor Lake is seen as a step forward in the company’s turnaround plan. The company is investing heavily in the development of its new chips, with plans to ramp up capacity and continue advancing its chip-stacking technology. Intel has allocated $3.5 billion to its New Mexico facility and $7 billion to Penang, Malaysia, over the next 10 years to support chip production.
While Intel did not disclose specific performance or battery life details for Meteor Lake, the company claims that the new chips outperform previous generations and are more power-efficient. More information is expected to be revealed closer to the launch date.