Chipsets

Ascend 310

Ascend 310 is a highly efficient, flexible and programmable AI processor. 16 TOPS@INT8 and 8 TOPS@FP16 can be output in typical configuration, while power consumption remains at just 8 W. Ascend 310 utilizes HUAWEI-developed Da Vinci Architecture, integrates a wide variety of computing units, and extends the application scope of the AI chip. Through the acceleration of all-AI service processes, the AI system's overall performance is greatly improved while deployment costs are reduced significantly.

HUAWEI Da Vinci NPU architecture

16 TOPS@INT8

3D Cube computing engine

Key Features

  • Architecture

    • HUAWEI Da Vinci
  • Computing Engine

    • 3D Cube
  • Performance

    • 16 TOPS@INT8 and 8 TOPS@FP16
  • Max Power

    • 8 W
  • Process

    • 12 nm FFC

View More Products

Support

HV2 Product Support Intro
Technical Support
banner