Welcome to Firefly
Switch language
Switch language
Firefly Docsss
Last Updated: 2026-07-31 13:46:27

Compile RK1820/RK1828 Installer

Download SDK

Please contact sales@t-firefly.com to get RK182X SDK download link.

Notice:


1. SDK use cross-compilation, so use SDK in x86_64 PC, do not download SDK to the device
2. We suggest to use Ubuntu20.04 or Ubuntu22.04 (real PC or docker) to build, other OS may cause building failure
3. Do not place or decompress the SDK archive in Virtual Machine share folder or non-english folder
4. Please use the regular user to get/compile the SDK, use root privilege may cause building failure


eg: SDK is RK182X_AI_COPROCESSOR_SDK_ALPHA_V1.0.4.tgz.

mkdir rk182x_sdk
cd rk182x_sdk
tar xf RK182X_AI_COPROCESSOR_SDK_ALPHA_V1.0.4.tgz
.repo/repo/repo sync -l

Config

Use ./build.sh config to config.

Select board type:
1) RK182X EVB1
2) RK182X SODIMM
3) RK182X SODIMM USB
4) RK182X M2
5) Cancel
#?

Select 4

Build

./build.sh

The generated software installation package is located at output/firmware/rknn3_rk182x_m2_installer_arm64.tgz

Install

You need to manually install the RK1820/RK1828 software package, follow the steps below:

  • Copy rknn3_rk182x_m2_installer_arm64.tgz to the main controller (RK3588 or RK3576)
  • Decompress: tar xzf rknn3_rk182x_m2_installer_arm64.tgz
  • Install: ./install.sh
    • After installation and rebooting, the RK3588 or RK3576 system will automatically download the RK182X firmware and start the background service program upon startup.

Others

Version V1.0.4

sudo rknn-smi -v
rknn-smi version              : 1.3.0
PCIe driver version           : 3.3.0
RC chips connect version      : 3.3.1
EP chips connect version      : 0.0.2
PCIe Device 0 firmware version: 1.0.4
rknn3 API version             : NA

FAQ

Currently Supported Models

ModelSource
Qwen2.5-0.5Bhttps://huggingface.co/Qwen/Qwen2.5-0.5B
Qwen2.5-3Bhttps://huggingface.co/Qwen/Qwen2.5-3B-Instruct
Qwen2.5-7Bhttps://huggingface.co/Qwen/Qwen2.5-7B-Instruct
Qwen3-0.6Bhttps://huggingface.co/Qwen/Qwen3-0.6B
Qwen3-1.7Bhttps://huggingface.co/Qwen/Qwen3-1.7B
Qwen3-4Bhttps://huggingface.co/Qwen/Qwen3-4B
Qwen3-8Bhttps://huggingface.co/Qwen/Qwen3-8B
HY-MT1.5-1.8Bhttps://huggingface.co/tencent/HY-MT1.5-1.8B
Youtu-LLM-2Bhttps://huggingface.co/tencent/Youtu-LLM-2B
GLM-Edge-1.5B-Chathttps://modelscope.cn/models/ZhipuAI/glm-edge-1.5b-chat
Qwen2.5-VL-3Bhttps://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct
Qwen2.5-VL-7Bhttps://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct
Qwen2.5-Omni-3B (Thinker)https://huggingface.co/Qwen/Qwen2.5-Omni-3B
Qwen3-VL-2Bhttps://huggingface.co/Qwen/Qwen3-VL-2B-Instruct
Qwen3-VL-4Bhttps://huggingface.co/Qwen/Qwen3-VL-4B-Instruct
FastVLMhttps://github.com/apple/ml-fastvlm
InternVL3-2Bhttps://huggingface.co/OpenGVLab/InternVL3-2B
InternVL3_5-4Bhttps://huggingface.co/OpenGVLab/InternVL3_5-4B-Instruct
MiMo-VL-7B-RLhttps://huggingface.co/XiaomiMiMo/MiMo-VL-7B-RL
Gemma-4-E2Bhttps://huggingface.co/google/gemma-4-E2B-it
Gemma-4-E4Bhttps://huggingface.co/google/gemma-4-E4B-it
SmolVLM-500M-Instructhttps://huggingface.co/HuggingFaceTB/SmolVLM-500M-Instruct
SmolVLM2-500M-Video-Instructhttps://huggingface.co/HuggingFaceTB/SmolVLM2-500M-Video-Instruct
UI-TARS-2B-SFThttps://huggingface.co/ByteDance-Seed/UI-TARS-2B-SFT
PaddleOCR VLhttps://huggingface.co/PaddlePaddle/PaddleOCR-VL
Qwen3-Reranker-0.6Bhttps://huggingface.co/Qwen/Qwen3-Reranker-0.6B
Qwen3-Reranker-4Bhttps://huggingface.co/Qwen/Qwen3-Reranker-4B
Qwen3-Embedding-0.6Bhttps://huggingface.co/Qwen/Qwen3-Embedding-0.6B
Qwen3-Embedding-4Bhttps://huggingface.co/Qwen/Qwen3-Embedding-4B
gme-Qwen2-VL-2B-Instructhttps://huggingface.co/Alibaba-NLP/gme-Qwen2-VL-2B-Instruct
Qwen3-ASR-0.6Bhttps://huggingface.co/Qwen/Qwen3-ASR-0.6B
Qwen3-TTS-12Hz-1.7Bhttps://huggingface.co/Qwen/Qwen3-TTS-12Hz-1.7B-Base
VITShttps://github.com/jaywalnut310/vits
Whisperhttps://huggingface.co/openai/whisper-large-v3
SenseVoiceSmallhttps://modelscope.cn/models/iic/SenseVoiceSmall
Zipformerhttps://huggingface.co/pfluo/k2fsa-zipformer-chinese-english-mixed
SigLIPhttps://huggingface.co/google/siglip-so400m-patch14-384
Siglip2-so400mhttps://huggingface.co/google/siglip2-so400m-patch14-384
MetaCLIP2https://huggingface.co/facebook/metaclip-2-worldwide-m16-384
Dinov3https://huggingface.co/facebook/dinov3-vits16-pretrain-lvd1689m
Depth-Anything-V2-smallhttps://huggingface.co/depth-anything/Depth-Anything-V2-Small
GR00T-N1.6-3Bhttps://huggingface.co/nvidia/GR00T-N1.6-3B
MobilenetV1https://ftrg.zbox.filez.com/v2/delivery/data/95f00b0fc900458ba134f8b180b3f7a1/examples/mobilenet_v1/mobilenet_v1_1.0_224.tflite
MobilenetV2https://ftrg.zbox.filez.com/v2/delivery/data/95f00b0fc900458ba134f8b180b3f7a1/examples/mobilenet/mobilenetv2-12.onnx
Resnet50V2https://ftrg.zbox.filez.com/v2/delivery/data/95f00b0fc900458ba134f8b180b3f7a1/examples/resnet/resnet50-v2-7.onnx
YOLOv5shttps://ftrg.zbox.filez.com/v2/delivery/data/95f00b0fc900458ba134f8b180b3f7a1/examples/yolov5/yolov5s_rknn3.onnx
YOLOv6shttps://ftrg.zbox.filez.com/v2/delivery/data/95f00b0fc900458ba134f8b180b3f7a1/examples/yolov6/yolov6s_rknn3.onnx
YOLOv8shttps://ftrg.zbox.filez.com/v2/delivery/data/95f00b0fc900458ba134f8b180b3f7a1/examples/yolov8/yolov8s_rknn3.onnx

For details, refer to SDK_Path/rknn/rknn3-runtime/doc/00_Rockchip_RKNPU3_ReleaseNote_RKNN3_SDK_V1.0.4_EN.pdf in the SDK.

Compile Main Module Firmware

Previous Page

Accessories

Next Page

On this page

Compile RK1820/RK1828 InstallerDownload SDKConfigBuildInstallOthersVersion V1.0.4FAQCurrently Supported Models