Qualcomm recently unveiled detailed technical information about its new-generation Adreno GPU, marking another significant advancement following the CPU upgrade plan for its next-generation flagship processor. This GPU introduces a dedicated hardware module, the Adreno Matrix Cores, which optimizes local AI task processing, reduces latency and power consumption, and provides computing power support for neural fusion technology. This enhances image stability and detail expressiveness, with power consumption reduced by up to 40% under equivalent image quality. Both Unreal Engine and Unity now support this technology, allowing game developers to integrate it without extensive low-level modifications. The new architecture also optimizes multithreaded task scheduling, improving rendering efficiency. In terms of hardware, the GPU adopts a three-slice architecture with a core frequency of up to 1.45GHz and retains 18MB of dedicated cache, enhancing performance-per-watt by 12%. Regarding connectivity, the integrated X105 modem supports 5G Advanced, offering theoretical download speeds exceeding 10Gbps. The FastConnect 8800 integrates Wi-Fi 7 and Bluetooth 5.3, reducing latency by 30% during concurrent multi-device connections. Together with the CPU-side innovations, these advancements form a comprehensive technology matrix. The simultaneous architectural innovations in both CPU and GPU signify that smartphone chips have entered a phase of comprehensive AI acceleration. The industry believes that the actual user experience will depend on terminal manufacturers' thermal management and power consumption tuning. How to translate the energy efficiency advantages of the 3nm process into stable high-performance output will be the key to competition for the next generation of flagship chips.
