Source: Shanghai Securities News · China Securities Network

Shanghai Securities News, China Securities Network (Reporters Li Xingcai and Huang Shuhui) On August 21, Granfi announced on its official website that the company has officially launched its first high-performance system-on-chip (SoC) based on the ARM architecture—GF800.

Granfi said that as the company’s key product in transitioning from GPU chip and graphics card suppliers to a full-stack SoC platform provider, GF800 will further convert the company’s long-term technological capabilities in the fields of graphics rendering, parallel computing, and AI acceleration into platform-based, productized capabilities. It will provide highly integrated, highly scalable, and independently controllable system-level computing power platforms for end-side products, edge computing, and industry-intelligent application scenarios.

Currently, the GPU solution options in the ARM SoC market mainly include ARM’s mobile graphics processing unit (GPU) IP series (Mali/Immortalis), Qualcomm’s mobile GPU in Snapdragon (Qualcomm Adreno), Samsung’s GPU Xclipse, and Imagination’s PowerVR, among others. For the first time, the GF800 integrates GLANFEI’s continuously self-iterated GPU IP into an SoC product, extending the company’s independent GPU capability from a standalone chip to a system-level platform. This provides customers with new differentiated and self-controlled choices.

According to the introduction, the GF800 has eight major highlights: First, it adopts heterogeneous fusion of “ARM CPU + self-developed GPU + dedicated NPU,” integrating general computing, graphics rendering, and AI acceleration under a unified memory architecture, to provide downstream terminal manufacturers and industry partners with a complete, flexible, one-stop underlying platform. Second, the self-developed GPU has strong graphics processing capabilities. Third, it uses the self-developed Glenfly HMUA hardware architecture and Glenfly HUCA software platform to achieve full-stack coordination, meeting needs such as rapid deployment and migration of large AI models. Fourth, it is deeply optimized and adapted for domestic operating systems such as UnionTech and Kylin, supporting mainstream operating systems including Linux and Android, thereby meeting broad ecosystem requirements and reducing the cost of cross-platform migration and secondary development. Fifth, it is equipped with multiple mainstream display interfaces, supporting up to 8K resolution. Sixth, it features a new-generation ISP with powerful video encoding/decoding and image processing capabilities, providing complete multimedia processing support for machine vision, video processing, voice interaction, and multimodal intelligent applications. Seventh, it offers rich interfaces and flexible expansion. Eighth, with a high level of integration, it helps users reduce deployment costs and shorten product development cycles.

The Chief Technology Strategy Officer of GLANFEI, Wang Yuanfeng, said that the significance of the GF800 lies not only in launching a brand-new SoC, but also in deeply integrating the company’s more than 20 years of continuously iterated GPU architecture innovations, driver software R&D, and system engineering practice capabilities into a scalable, sustainably evolvable platform system. The company’s self-developed GPU is the technical foundation of the GF800, while full-stack coordination capability is its core competitive advantage. Building on this platform, the company will continue to deliver stable, efficient, and continuously evolving product value for edge computing and industry intelligent application scenarios.

Reporters learned that thanks to the highly integrated SoC design, support for multiple operating systems, and a unified software platform, the GF800 can cover a wide range of computing terminals, including but not limited to AI terminals, edge servers, industrial computing equipment, robots, and embodied intelligence, further expanding the market application space of domestic SoCs in on-device products, edge computing, and industry intelligent scenario solutions.