Xiaomi Xuanjie Chip: Breakthrough and Technical Analysis of 3nm Process Technology

Recently, Xiaomi officially released its first self-developed chip equipped with 3nm process technology - Xuanjie O1, marking a crucial step for domestic chips in the field of advanced processes, with innovative CPU architecture, powerful GPU performance, and efficient AI processing capabilities. This article will analyze the key technologies and industry significance of the Xuanjie O1 chip.

Advanced process technology

The Xuanjie O1 chip adopts TSMC's second-generation 3nm process (N3E), integrating approximately 19 billion transistors with a chip area of only 109mm ². This manufacturing process is currently at the forefront of the mobile phone chip field, with a transistor density increase of about 12% and power consumption reduction of about 15% compared to previous generation processes. Smaller process technology means that more transistors can be integrated on the same chip area, achieving higher computing power and lower power consumption performance. This is of great significance for improving the overall performance of chips and the endurance of terminal devices.

Innovative CPU architecture

The Xuanjie O1 adopts a "2+4+2+2" ten core four cluster architecture, including 2 Cortex-X925 ultra large cores (with a frequency of up to 3.9GHz), 4 Cortex-A725 performance cores (with a frequency of 3.4GHz), 2 Cortex-A725 low-frequency energy efficiency cores (with a frequency of 1.9GHz), and 2 Cortex-A520 ultra-high energy efficiency cores (with a frequency of 1.8GHz). This unique architecture design breaks the traditional 8-core framework of mobile SoC and enables flexible core scheduling based on different task loads, ensuring peak performance while achieving ultimate energy efficiency.

In terms of actual performance, the Xuanjie O1's single core Geekbench 6 score reaches 2709 points, and its multi-core score reaches 8125 points, surpassing the Snapdragon 8 Gen3 and approaching the level of the Dimensity 9400. Its dual core Cortex-X925 ultra large core can provide powerful performance support in instantaneous burst tasks, while the quad core Cortex-A725 performance large core can cope with high load multi-threaded scenarios, ensuring smooth system operation. At the same time, the presence of low-frequency and ultra-high energy efficiency cores enables chips to effectively reduce power consumption and extend device endurance in daily use and standby scenarios.

Powerful GPU configuration

The Xuanjie chip is equipped with a 16 core Immortalis-G925 GPU, which supports dynamic performance scheduling technology and can dynamically adjust the GPU's operating status according to the running scenario. As Arm's latest generation of high-end GPU, Immortalis-G925 introduces hardware level ray tracing and scalable rendering units in its architecture, which can bring users more realistic game graphics and smoother graphics processing experience. Whether it's high frame rate games or complex graphics rendering tasks, Xuanjie O1's GPU can easily handle them, meeting users' needs for high-performance graphics processing. In addition, its 6-core NPU has a computing power of 44 TOPS and supports multi algorithm parallel computing. Combined with Xiaomi's third-generation end-to-end model, AI processing speed is improved while power consumption is reduced.

A complete heterogeneous computing framework

Xuanjie O1 has built a complete heterogeneous computing framework, in which high-performance CPUs complete general task scheduling, GPUs and ISPs are responsible for graphics and image processing, NPUs undertake intelligent inference tasks, and ultimately achieve a smooth system experience through on-chip high-speed interconnection and memory scheduling. This heterogeneous computing architecture can fully leverage the advantages of various functional modules, improve the overall performance and energy efficiency of chips, and provide stronger computing support for terminal devices.

The release of Xiaomi Xuan Jie O1 chip represents a key progress in the research and development of domestic high-end chips. Its TSMC second-generation 3nm process, innovative architecture, and powerful AI capabilities have set a new benchmark for the industry. This not only promotes the high-end development of domestic chips, but also inspires domestic enterprises to increase their research and development efforts, enhance global market competitiveness, and have profound significance in promoting the independent controllability of domestic chips, helping to upgrade intelligent devices, and promoting the intelligent development of the technology industry.