Tensor Computing Processor
BM1680

Sophon BM1680 is customized AI ASIC for tensor computing acceleration. It is designed for both deep learning training and inference of neural networks like CNN, RNN, DNN etc.

top image for mobile

Specification

Peak performance

2TFlops

Data precision

FP32

Internal memory size

32MBytes

Average power consumption

25W

Download Datasheet >

Features and Advantages

BM1680 performance

Tensor computing acceleration

BM1680 performance

Optimized deep learning architecture

BM1680 performance

Flexible product and easily customized

BM1680 performance

Large size and high speed memory

BM1680 performance

Optimized command set