This AI artificial intelligence module is based on the core board with the stamp hole full interface developed by Rockchip’s RV1109/RV1126 AI media processing chip. It is small in size and rich in interfaces, which is convenient for customers to evaluate and develop projects quickly.

This AI artificial intelligence module is widely used in wireless screen projection, intelligent visual robots, drones, intelligent IPC, face recognition panel machines, face capture cameras, video doorbells, gate access control, face recognition, driving recorder, high-definition video conference system, etc.

CPU

• ARM Cortex-A7 1.5GHz 32-bit core which integrates NEON and FPU. There is a 32KB I-cache and 32KB D-cache for each core and a 512KB unified L2 cache.

• RSIC-V 400MHz

NPU

• 2Tops, support INT8/ INT16

Memory

• 32bit DDR4 1G

• Support eMMC 8G

• Support fast booting

Display

• MIPI-DSI/RGB interface, up to 1080P60fps

2D Graphics Engine

• Support rotation, x-mirror, y-mirror

• Support alpha blending

• Support scale-down/up

Multi-Media

• 14M ISP 2.0 with 3F HDR(Line-based/Frame- based/DCG)

• Support 2*MIPI CSI /LVDS/sub-LVDS

• DVP interface with BT.656/BT.1120

• 4K H.264/H.265 30fps video encoder

• 3840 x 2160@30 fps+720p@30 fps encoding

• 4K H.264/H.265 30fps video decoder

• 3840 x 2160@30 encoding + 3840 x2160 @30fps decoding

External interface

• RGMII interface with TSO network acceleration

• USB 2.0 OTG and USB 2.0 • Dual SDIO 3.0 interface for Wi-Fi and SD card

• 8ch I2S with TDM/PDM, 2ch I2S

Availability

• 2020 Q4

• SMIC 14nm, QFN 156 1.0

SOC	RV1126, RV1109
RAM	32-bit, 512MB*2
ROM	8GB EMMC
NPU	0TOPS, support RKNNAI frame
USB interface	Two USBs, one independent OTG, one independent HOST
UART serial port	6 TTL3.3V level, one of which is Debug
GPIO Interface	102 GPIOs (Note: there is a multiplexing function, please check the pin definition in this specification>
PWM interface	5-way PWM has 3.3V/1.8V levels respectively (please check the pin definition in this specification)
MIPI DSI shows	1 channel Maximum 1920×1080 60fps
CAN	1 channel
RGB display	1 channel RGB 888 display interface, maximum 1920×1080
DVP camera interface	1 channel 16-bit parallel DVP camera interface
MIPI camera interface	2 channels MIPI CSI RAW data camera interface (one channel 4K/30)
Voice output	1 channel Left and right channel output
voice input	1 PDM digital silicon microphone input, 1 analog LANE_IN input or I2S
SDIO interface	2 SDIO interfaces, one TF card, the other can be connected to WIFI
ADC interface	6 ADC interface
I2C interface	4 I2C interface
SPI Interface	2-way SPI interface
Ethernet	– one-way RGMII PHY interface, Gigabit Ethernet
system upgrade	supports local USB upgrade
Size	38*48mm
PCB layers	FR4-6 layers
Working range	-10℃-7o℃

As an RV1126 Rockchip core board circuit supplier, we also offer to pin out customized second development layout design produce services and products.

This AI artificial intelligence core module is an AI intelligent module designed based on Rockchip RV1126 high-performance AI processor, quad-core RV1126 ARM Cortex-A7@1.5GHz 32-bit core, integrated NEON, and FPU. Each core has a 32KB l-cache and 32KB D-cache and 512KB unified L2 cache. Built-in NPU supports INT8/INT16 mixed operation, computing power up to 2TOPS Built-in RISC-V architecture MCU; supports Linux4.19 operating system, the network model has strong compatibility, supports a series of deep learning such as TensorFlow / MXNet / PyTorch! Caffe framework that is easy to convert. Super performance, fast system startup, and high stability.

Can be adapted to different face recognition algorithms, suitable for embedding in various AI vision products, widely used in binocular face recognition panel machines, AI binocular cameras, smart communities, smart doorbells, smart access control/cat’s eye, face payment, Related industries related to intelligent upgrades such as driving recorders, game interaction, and webcasting.

The carrier board or motherboard for the core board module.