Today, the Chinese manufacturer Xiaomi has officially introduced its Xiaomi Xring tri-chip solution in its home country, China, which includes Xiaomi Xring O3, Xiaomi Xring D100 and Xiaomi Xring O100.
Xiaomi Xring O3
The Xiaomi Xring O3 is a flagship AI SoC. It has achieved an AnTuTu score of 5.22 million. The CPU features a deca-core design with all large cores, achieving a multi-core score exceeding 15,000 for the first time, representing a 60% performance improvement.
The GPU debuts the G2-Ultra NX graphics processor, an unprecedented generational upgrade with an 85% performance increase and a 64% reduction in power consumption. The NPU boasts an impressive 200 TOPS tensor computing power and 3.13 TFLOPS vector computing power, resulting in a 45% improvement in on-device AI performance.
With the addition of dual SME2 acceleration units to the CPU, eight new NX neural network accelerators to the GPU, a built-in AINR unit to the ISP, a built-in hardware AI super-resolution unit to the DPU, and a built-in AI engine to the ADSP, the Xring O3 represents a comprehensive solution in edge AI performance, hardware-level super-resolution frame interpolation, and night scene video noise reduction.
It is revealed that the chip has undergone extensive optimisation for power consumption and smoothness, employing the Xring unified fusion bus architecture and top-level metal trace technology to achieve an industry-leading 82 ns static latency.
The Xiaomi Xring O3 is the world’s first mobile processor to support LPDDR6, boasting a bandwidth of 113.8GB/s, a 48% improvement over the Xring O1.
Xiaomi has officially announced that the Xiaomi 18 Fold will be the first to feature the Xring O3 AI flagship processor, while the Xiaomi Pad 9 Pro Max will also be equipped with this chipset. Both devices will be launching next month in China.
Xiaomi Xring D100 & Xring O100
The Xiaomi Xring D100 is China’s first 3nm high-performance AI chip for intelligent driving. It utilises an advanced 3nm process and features a powerful combination of a 20-core high-performance CPU and a 16-core high-performance NPU. It supports up to 160GB of unified memory and can locally deploy ultra-large models with up to 200B parameters.
The Xiaomi Xring O100 is a 1.22TB/s high-bandwidth AI acceleration chip that comes with an advanced near-memory AI architecture. It boasts 6nm ED wafer-level stacked advanced packaging that allows computing units and memory to be stacked vertically, creating millions of high-speed vertical channels within the chip. There are 2.58 million physical bonding nodes with a minimum bonding pitch of only 1.4µm. It can be applied to all ecosystem categories, such as mobile phones, automobiles, and robots in the future.
Xiaomi Xring D100 and O100 have been developed and will be officially commercialised next year.
