Datasheet
Table Of Contents
- About This Guide
- Chapter1 Overview
- Chapter2 Pin Specifications
- Chapter3 Functional Description
- 3.1 CPU
- 3.2 Neural Network Processor (KPU)
- 3.3 Audio Processor (APU)
- 3.4 Static Random-Access Memory (SRAM)
- 3.5 System Controller (SYSCTL)
- 3.6 Field Programmable IO Array (FPIOA/IOMUX)
- 3.7 One-Time Programmable Memory (OTP)
- 3.8 AES Accelerator
- 3.9 Digital Video Port (DVP)
- 3.10 FFT Accelerator
- 3.11 SHA256 Accelerator
- 3.12 Universal Asynchronous Transceiver (UART)
- 3.13 Watchdog Timer (WDT)
- 3.14 General Purpose Input/Output Interface (GPIO)
- 3.15 Direct Memory Access Controller (DMAC)
- 3.16 Inter-Integrated Circuit Bus (I²C)
- 3.17 Serial Peripheral Interface (SPI)
- 3.18 Inter-Integrated Sound (I²S)
- 3.19 TIMER
- 3.20 Read Only Memory (ROM)
- 3.21 Real Time Clock (RTC)
- 3.22 Pulse Width Modulation (PWM)
- Chapter4 Electrical Characteristics
- Chapter5 Package information

Chapter3 Functional Description 16
Mode
Maximum fixed point
model size (MiB)
Maximum pre-quantisation
floating point model size
(MiB)
Non-realtime(< 10fps)
*
1
Flash Capacity
*
2
Flash Capacity
The internal structure of the KPU is shown below.
Gs Interface
Gm Interface
Main memory
Parameter memory
Parameter parsing unit
Main control unit
KPU Arithmetic unit
Arithmetic control unit
Write back unit
AXI
Bus
AXI
Bus
3.3 Audio Processor (APU)
The APU pre-processing module is responsible for the pre-processing of voice
direction and voice data output. The functional characteristics of the APU pre-
processing module are:
• 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
• Uses system DMAC to store output data in system memory
*
1
Non-real-time applications generally include audio applications, as such applications generally do
not require neural network output within 33 ms.
*
2
Flash size support:SPI NOR Flash(8MiB,16MiB,32MiB),SPI NAND Flash(64MiB,128MiB,256MiB)