K210

1 Overview

KPU is a general-purpose neural network processor with built-in convolution, batch normalization, activation, and pooling operations.

It can detect faces or objects in real time.

Supports the fixed-point model that the mainstream training framework trains according to specific restriction rules
There is no direct limit on the number of network layers, and each layer of convolutional neural network parameters can be configured separately, including the number of input and output channels, and the input and output line width and column height
Support for 1x1 and 3x3 convolution kernels
Support for any form of activation function
The maximum supported neural network parameter size for real-time work is 5MiB to 5.9MiB
The maximum supported network parameter size when working in non-real time is (flash size - software size)

The APU pre-processing module is responsible for the pre-processing of voice direction and voice data output.

Up to 8 channels of audio input data, ie 4 stereo channels
Simultaneous scanning pre-processing and beamforming for sound sources in up to 16 directions
Supports one active voice stream output
16-bit wide internal audio signal processing
Support for 12-bit, 16-bit, 24-bit, and 32-bit input data widths
Multi-channel direct raw signal output
Up to 192kHz sample rate
Built-in FFT unit supports 512-point FFT of audio data
Using system DMA to store output data in system memory