The most significant processor architecture change in 40 years and the addition of AI functions - Intel 4 Meteor Lake processor

At the Intel on Technology Innovation Conference held recently, Intel launched the first Meteor Lake processor platform based on the Intel 4 process technology. Thanks to the advanced Foveros 3D packaging technology, Meteor Lake adopts a separated module architecture, dividing the entire processor into functional partitions of computing module, IO module, SoC module, and graphics module, bringing 40 years of revolutionary Intel client SoC Architectural shift. The processor adopts a 3D high-performance hybrid architecture composed of performance cores (Redwood Cove), energy efficiency cores (Crestmont) and low-power energy-efficiency cores to create an excellent energy consumption ratio in the history of PC processors, and for the first time integrates a neural network processing unit (NPU) is integrated into PC processors to accelerate the popularization of artificial intelligence.

The Meteor Lake processor adopts a separated module architecture, consisting of four independent modules: computing module, SoC module, graphics module and IO module, and are connected through the industry's outstanding Foveros 3D packaging technology. Among them, the computing module uses the Intel 4 process technology for the first time, making the Meteor Lake processor the most energy-efficient client platform in Intel’s history. This innovative design represents Intel's most significant architectural shift in 40 years and lays the foundation for PC innovation in the next 10 years.

The 4 unique modules included in the separated module architecture are:

• Compute module (Compute Tile): adopts the latest generation of energy-efficiency core and performance core micro-architecture and enhanced functions. This module uses the new generation Intel 4 process technology and has achieved significant progress in energy consumption ratio.

• SoC module (SoC Tile): Innovative low-power island design integrates a neural network processing unit (NPU), bringing highly energy-efficient AI function performance to the PC, and is compatible with standardized program interfaces such as OpenVINO to facilitate AI development and The application is popular. The new low-power energy efficiency core further optimizes the balance between energy saving and performance. The SoC module also integrates a memory controller, media codec processing and display unit, supporting 8K HDR and AV1 codecs as well as HDMI 2.1 and Display Port 2.1 standards. Wi-Fi and Bluetooth are also supported, including Wi-Fi 6E.

• Graphics module (GPU Tile): This processor integrates the Intel Sharp™ graphics architecture, which can provide independent graphics card-level performance in the integrated graphics card and supports ray tracing and Intel XeSS. With a jump in graphics capabilities and higher energy efficiency, Meteor Lake delivers outstanding performance per watt.

• IO module (IO Tile): Contains industry-leading connectivity, integrating Thunderbolt™ 4 and PCIe Gen 5.0.

Packaging is an essential step for the processor. It allows electricity and signals to be transmitted between the motherboard and the chip, and protects the chip. In today's advanced packaging era, packaging technology also has added value: it can create Larger wafer complexes at different process nodes. This is not without theoretical basis. Intel founder Gordon Moore already expressed it in 1965: "When building a large system, it may be more economical to break it down into small functional modules that are individually packaged and interconnected." From Intel in 2013 ^® Core™ processors’ Packaged Chipsets (PCH), to Stratix 10 with EMIB (Embedded Multi-die Interconnect Bridge) technology in 2017, to 2020’s Active 3D Stacking Foveros technology for Lakefield processors, to 2022 with The multi-module Foveros+EMIB packaged Ponte Vecchio processor, Intel continues to explore and innovate in packaging technology.

The new Meteor Lake processor will use Foveros packaging technology to achieve extremely low power consumption and high-density chip connections within the chip. At the same time, the modular design can reduce the size of a single wafer, and more chips can be obtained from the same wafer, increasing the number of chips obtained from each wafer, and accelerating customization and market launch. In addition, a more suitable chip process can be selected for each block to achieve better cost and performance.

Intel is following Moore's Law and steadily advancing five process nodes in four years. The new Meteor Lake processor will use the Intel 4 process technology. Compared with the 408nm high-performance library height of the Intel 7 process technology, Intel 4's 240nm has achieved a 2x reduction in high-performance logic library area (the data is based on internal estimates, please check for details) Visit www.intel.com/PerformanceIndex. Results may vary). Under the Intel 4 process technology, the density of MIM has also been increased, allowing the processor to achieve more efficient power supply. At the same time, 8VTs optimized for the CPU are used to improve performance and power efficiency. In addition, the Intel 4 process technology also uses EUV extreme ultraviolet lithography technology to simplify the manufacturing process and improve manufacturing efficiency.

Nowadays, the use scenarios of AI have shifted to the edge and the terminal side, and the application scenarios and demands of AI PCs have grown rapidly. From intelligent voice noise reduction, video background blur, super-resolution to intelligent interception of game highlight moments, these are all aspects of the popularization of AI; today's large language model dialogue and AIGC scenes such as Vincentian pictures, Tusentu and Vincentian videos , which adds to the appeal of AI PC application forms, and the demand for computing power is also increasing. The new processor further innovates high-energy-efficiency AI PCs by implementing the XPU strategy.

In the Meteor Lake processor, the NPU for artificial intelligence acceleration is introduced for the first time, which is designed to bring high-energy and low-consumption performance for continuous AI and AI offloading (reducing the AI ​​workload of the CPU and GPU through the NPU). In addition to NPU, low-latency and high-response CPUs and high-performance and high-throughput GPUs also bear the demand for AI computing power. NPU, CPU and GPU collaborate with each other to continuously accelerate AI scenarios on PC.

Intel is working with industry partners to improve the AI ​​experience based on the new Meteor Lake processor.