The conference announced the winning enterprises of the "2024 Strong Chips China - Innovative IC" award. This award aims to select a batch of technologically leading, highly competitive and reliable high-quality chips in key fields such as AI computing power, storage, and microprocessors/controllers nationwide. It also serves to recommend high-quality domestic chips to system and complete machine enterprises as well as user units, and promote the application of domestic ICs. With its outstanding performance, economic and environmental advantages, Axera's self-developed AxeraNeurons AI Processor won the "Emerging Product Award".

 

The core of the AxeraNeurons AI Processor is the operator instruction set and dataflow microarchitecture. Its underlying layer adopts a programmable dataflow microarchitecture to improve energy efficiency and computing power density. At the same time, its flexibility ensures the completeness of the operator instruction set, supporting various AI applications. Its mature software toolchain enables developers to get started quickly. In addition, the joint design of software and hardware ensures the rapid iteration of the AxeraNeurons AI Processor, thus maintaining its competitiveness. The AxeraNeurons AI Processor has greatly reduced the development and operation and maintenance costs of AI applications, making AI intelligence more economical, efficient and environmentally friendly.

The outstanding performance of the AxeraNeurons AI Processor stems from Axera's accurate prediction of Transformer, a mainstream network architecture, in the early years. At the conference, Liu Jianwei, Co-founder and Vice President of Axera, shared his technical insights on AI processors in the intelligent era. Mr. Liu stated that AI has become a currently stable and reliable infrastructure for intelligent output, leading a cost-driven productivity revolution. Compared with traditional GPUs, AI processors can help enterprises better realize digital and intelligent transformation in a more economical, efficient and environmentally friendly way. As early as 2022, Axera laid out AI processors based on the Transformer architecture, among which the AX650 is still the most efficient processor for running Transformer models.

According to Mr. Liu Jianwei, Axera's AI processors have completed the layout in low, medium and high tiers of computing power, and achieved large-scale mass production in two fields: smart cities and assisted driving. The energy efficiency ratio is an order of magnitude higher than that of GPGPU chips. In addition, in general large model applications such as text-to-image search, general-purpose detection, image-to-text generation, and AI Agent, the AxeraNeurons AI Processor also enables AI developers to carry out efficient development at a lower cost.