RK1820/RK1828

RK3588 Soc(Host) : As the core of the system, responsible for task scheduling, resource allocation, and overall control.
RK1820/RK1828 Coprocessor(Device) : As an AI computing acceleration unit, we focus on high-performance neural network inference tasks.
PCIe : Realize low latency and high bandwidth data exchange between the main control and co processors.

rknn-smi

rknn-smi (System Management Interface) use as RK1820/RK1828 equipment information collection, function configuration, and log management functions.

Software Version: sudo rknn-smi -v
Hardware Version: sudo rknn-smi info -l
Status: sudo rknn-smi info -w
Performance: sudo rknn-smi set -t work_mode -s 2
Power Consumption: This feature is not supported on hardware
- sudo rknn-smi info -t power

How to use, see RK182X SDK: docs/Tools/Rockchip_User_Guide_RKNN-SMI_Tool_EN.pdf

Probability "Failed to initialize rknnsmi"

/lib/systemd/system/rknn3.service add appropriate delay:

[Unit]
Description=rknn3 runtime service
DefaultDependencies=no
After=local-fs.target

[Service]
Type=forking
ExecStartPre=/bin/sleep 3 # delay 3 seconds
ExecStart=/bin/rknn3_startup start
ExecStop=/bin/rknn3_startup stop

[Install]
WantedBy=sysinit.target

RKNN3

RK182X SDK dictionary:

rknn/
├── rknn3-model-zoo
├── rknn3-runtime
├── rknn3-toolkit
└── rknn-gstreamer-plugins

RKNN3 SDK Block Diagram

RKNN3 Model Zoo

Provide deployment examples of classic models on the RK1820/RK1828 platform.

github

RKNN3 Runtime

The RKNN3 C API is the C language interface for the RKNN3 Runtime.

Developers use C/C++ to develop application programs and deploy model inference through RKNN3 C API.

RKNN3 Toolkit

RKNN3 Toolkit is a development kit that provides users with model transformation, inference, and performance evaluation on the PC platform.

RKNN3 Toolkit different from RKNN-Toolkit and RKNN-Toolkit2.