company blog about Rockchip RK1808 High-Performance Low Power Neural Network Inference Processor

Rockchip RK1808 High-Performance Low Power Neural Network Inference Processor

Product Description
RK1808 is a high-performance, low power processor for neural network inference.Especially, it is one of current leading solution for mobile device by providingcomplementary neural network hardware accelerator.Equipped with one powerful neural network process unit(NPU), it makes RK1808 easyprogramming and compatible with mainstream platforms in the market, such as caffe,tensor flow, and so on.

Key Highlight
Dual Cortex-A35 clock up to 1.6GHz
High performance NPU
2MB system SRAM
1080P@60FPS H.264 decoder, 1080P@30FPS H.264 encoder
Video Input MIPI-CSI/BT.656/BT.1120
Video Output: 1080P MIPI-DSI/RGB
Audio 8-channel I2S/PDM with voice detection VAD
High-speed interface: USB3.0/PCIe2.1/RGMII
RK1808 support cascade

Features
Dual-core Cortex-A35, maximum frequency 1.6GHz
3 TOPs for INT8 / 300 GOPs for INT16 / 100 GFLOPs for FP16
Support OpenCL/OpenVX
Supports INT8/INT16/FP16
Supports TensorFlow, Caffe, ONNX, Darknet models.
800MHz 32-bit LPDDR2/LPDDR3/DDR3/DDR3L/DDR4
Support Serial SPI NOR/NAND Flash, EMMC
4-lane, MIPI-CSI, support Virtual Channel
Support BT.601/BT.656/BT.1120
4-lane, MIPI-DSI, up to 1920*1080
18-bit Parallel RGB panel, up to 1280*720
Support USB3.0/PCIe2.1
Built-in 2-ch & 8-ch I2S & 8-ch PDM, Built-in VAD
Support Gigabit Ethernet
8 x UART/3 x SPI/6 x I2C/11 x PWM/4 x SARADC

Applications
Sweeping robot
Drone
Smart speaker
Automotive products
Smart wear
Security monitoring
AI computing modules