#rk3576

[forlinx]

Learn how to convert YOLOv8 models to RKNN and deploy on OK3576-C with RK3576 NPU for real-time embedded AI object detection.

🚀 Edge AI Made Easy with OK3576-C & YOLOv8!

With the rapid growth of intelligent vision applications, real-time on-device object detection has become a core capability for industrial vision, smart devices, and surveillance.

At Forlinx Embedded, we’ve demonstrated a full workflow from model to edge deployment using the OK3576-C board with RK3576 NPU:

1️⃣ Convert YOLOv8 pre-trained models (.pt) → RKNN format 2️⃣ Deploy the model on the OK3576-C board 3️⃣ Run real-time object detection using a USB camera (~30 FPS)

This guide helps developers quickly move from model training to edge inference with minimal setup, leveraging the RK3576 NPU’s efficiency and power.

[forlinx]

Solve RK3568 Ethernet packet loss and port failures caused by PCB differences. A 3-step guide: scan delayline, verify with loopback, and sol

Accelerating Multimodal Vision Applications with RK3576 SoM

Forlinx Embedded unveils a next-gen vision assistant built on the powerful Rockchip RK3576 SoM – integrating LLM + VLM multimodal AI and real-time perception capabilities.

🔍 Key Highlights:

✅ Quad-core A76 + Quad-core A55 architecture for high-performance computing

✅ Integrated NPU (6 TOPS) for fast visual inference

✅Seamless support for multi-camera input, display, and edge AI

✅ Ideal for industrial inspection, smart retail, robotics, and automation

Explore how the RK3576-based platform powers real-time human-machine interaction and AI-driven decision-making.

[embedded-sbc]

Why Embedded SBCs Are Revolutionizing Home Automation

The smart home industry is evolving rapidly, driven by the demand for more intelligent, responsive, and energy-efficient devices. At the heart of this transformation lies the embedded Single Board Computer (SBC). From voice assistants and smart thermostats to AI-powered control panels, SBCs have become the essential computing platform that enables seamless home automation. In this article, we’ll examine how SBCs—particularly those based on Rockchip SoCs—are reshaping the smart home landscape with better cost-efficiency, multimedia performance, and AI capabilities.

Introduction to Embedded SBCs in Home Automation

Embedded Single Board Computers (SBCs) are becoming a core component in smart home devices. They offer a compact, energy-efficient, and powerful platform for running Linux or Android, handling tasks like display control, AI processing, and wireless communication.

Compared to traditional microcontrollers or desktop solutions, SBCs strike a balance between performance, flexibility, and cost—making them ideal for always-on home automation scenarios.

In this article, we’ll explore their functions, compare Rockchip vs NXP SBCs, and show why Rockchip is gaining traction in this fast-growing market.

What Is an Embedded SBC?

An embedded SBC (Single Board Computer) is a complete computer built on a single circuit board. It includes a processor, memory, storage, and I/O ports.

Unlike microcontrollers, SBCs run full operating systems like Linux or Android, making them suitable for complex smart home applications.

They are often used in touch panels, voice assistants, gateways, and more.

Why SBCs Are Ideal for Home Automation Applications

SBCs combine computing power, wireless connectivity, and multimedia capabilities in a compact form.

They support 24*7 operation, integrate easily with sensors and actuators, and reduce time-to-market for smart devices.

Their flexibility and low cost make them ideal for building reliable and scalable smart home systems.

Key Functions of SBCs in Smart Homes

Embedded SBCs enable multiple smart home features by combining processing, control, and connectivity on a single board. Below are some of their key roles:

Touch Panels & Smart Displays

SBCs drive user interfaces on wall-mounted screens, smart mirrors, and home hubs.

Voice Control & AI Processing

Support for microphones and AI frameworks allows real-time voice recognition and local processing.

Wireless Connectivity: Wi-Fi, Bluetooth, Zigbee

SBCs integrate wireless modules for seamless communication across devices and sensors.

Integration with Sensors, Relays, and Actuators

SBCs control lights, switches, temperature sensors, and more through GPIO, I2C, and UART.

Choosing the Right SoC for Home Automation

When designing smart home devices, selecting the right System on Chip (SoC) is critical. The ideal SoC must strike a balance between performance, power efficiency, cost, and software ecosystem.

Key Selection Criteria

FactorExplanationPerformanceEnough CPU/GPU/NPU capability for your workloadPower EfficiencyImportant for always-on or battery-powered devicesCostDirectly affects final product pricing and market competitivenessI/O & Peripheral Must match display, audio, sensor, or communication needs Software Ecosystem Good Android/Linux SDK, community support, long-term availability

Rockchip vs NXP: Price and Hardware Capability Comparison

FeatureRockchip RK3566NXP i.MX 8M MiniCPUQuad-core Cortex-A55Quad-core Cortex-A53GPUMali-G52 2EEGC7000LiteAI Support (NPU)Optional 0.8 TOPS NPUNo NPUDisplay InterfacesHDMI, MIPI-DSI, LVDSMIPI-DSI onlyMultimedia4K H.265/H.264 decode, 1080p encodeOnly up to 1080p videoAudioRich audio: I2S, TDM, SPDIF, PDMLimited audio interfacesWireless SupportSDIO Wi-Fi/BT, PCIe LTE modulesSDIO onlyMainline Linux SupportGood community & SDK supportStrong Yocto/Linux BSPCost (SBC level)Lower (approx. $30–$45 USD)Higher (approx. $50–$65 USD)AvailabilityStable for large-scale projectsLong-term supply guaranteed

Why Rockchip SBCs Are Gaining Popularity

Rockchip offers better multimedia and AI performance at a lower cost, especially suitable for touch panels, AI interfaces, and camera systems.

NXP is known for long-term support and is suitable for industrial systems where ecosystem stability is critical.

Rockchip SBC Lineup for Different Smart Home Needs

Rockchip offers a diverse range of SoCs tailored to various smart home scenarios—from cost-sensitive voice control devices to high-end AI-powered multimedia terminals. Here’s a comparison of three popular options:FeatureRK3308RK3566RK3576PositioningEntry-level for voice & gatewayMid-range for HMI & smart displaysHigh-end for AI + multimedia interfacesCPUQuad-core Cortex-A35Quad-core Cortex-A55Quad-core Cortex-A72 + Cortex-A53AI AcceleratorNoOptional (NPU: 0.8 TOPS)Built-in 6.0 TOPS NPUDisplay OutputNone or minimal (SPI LCD)HDMI, MIPI-DSI, LVDSHDMI 2.0, MIPI-DSI, eDPAudio Interface4x PDM, I2S, TDMI2S, TDM, HDMI audioRich audio interface, multi-channelConnectivityEthernet, USB 2.0, SPIUSB 3.0, Ethernet, PCIe, SDIOUSB 3.1, Gigabit LAN, PCIe 2.1Typical Use CaseSmart speakers, voice gatewaysTouch controllers, wall panelsAI camera hubs, smart mirrorsOS SupportLinux, RTOSLinux, AndroidLinux, AndroidCost RangeVery lowMediumHigher-end

 RK3308 — Budget-Friendly SBC for Voice Control & Gateways

RK3308 is optimized for audio processing, making it perfect for voice assistants, intercom systems, and smart gateways.

Its ultra-low power consumption and affordable pricing make it a strong choice for high-volume deployments.

RK3566 — Mid-Tier All-Rounder for Smart Displays and HMI

RK3566 strikes a balance between performance and cost. It supports multiple displays, moderate AI tasks, and fast storage.

Ideal for wall-mounted touch panels, smart thermostats, and home dashboards.

RK3576 — High-Performance Choice for Multimedia & AI-Powered Interfaces

With powerful CPU cores and a 6.0 TOPS NPU, the RK3576 handles HD video, AI vision, and interactive UI with ease.

It suits premium smart home products such as AI camera hubs, smart mirrors, and central control panels.

tips：Higher-performance SoCs, such as the RK3588, are currently less suitable for home automation applications due to their high cost and excessive performance for typical smart home tasks.

Real-World Applications and Case Studies

Rockchip SBCs have been successfully deployed in a wide range of smart home products. Below are some typical use cases based on specific SoCs.

Smart Thermostat with RK3308

RK3308’s low power consumption and strong audio capabilities make it ideal for smart thermostats with voice input and basic display needs.

It can handle voice commands locally, connect to Wi-Fi, and control relays for HVAC systems.

Wall-Mounted Touch Controller with RK3566

With HDMI/MIPI display support, moderate AI power, and multiple I/O options, RK3566 is widely used in wall-mounted touch panels.

These panels control lighting, curtains, temperature, and more—often acting as a central hub.

AI Camera Hub or Smart Mirror with RK3576

RK3576 offers high-performance AI and multimedia capabilities, making it perfect for smart mirrors with face recognition or camera hubs for security and automation.

These devices benefit from the chip’s 6.0 TOPS NPU, HDMI 2.0 output, and rich I/O options.

Conclusion

Why Rockchip-Based SBCs Are a Smart Investment for Home Automation

Rockchip SBCs offer an excellent balance of performance, cost, and features tailored for smart home applications.

With support for Android/Linux, AI acceleration, multimedia, and a wide range of interfaces, they are flexible enough to power everything from voice gateways to advanced AI panels.

Compared to NXP and other competitors, Rockchip provides stronger multimedia and AI capability at a lower price, making it a smart investment for scalable home automation solutions.

Future Trends and What to Expect Next

As AI and local processing become more important in home automation, SBCs with on-device intelligence will become standard.

Future SBCs will likely integrate more powerful NPUs, better power efficiency, and native support for protocols like Matter and Thread.

Rockchip is well-positioned to lead this evolution, offering SoCs that are ready for both current deployments and future upgrades.

[forlinx]

Discover how Forlinx’s RK3576-based SoM powers intelligent lawn mowing robots with AI, precision control, and multi-sensor integration for s

Smarter, Greener, and More Autonomous!

Meet the future of outdoor automation with our intelligent lawn mowing robot solution powered by the Forlinx FET3576-C SoM, based on the Rockchip RK3576.

With an 8-core CPU, 6 TOPS NPU, AI-powered obstacle detection, and precise route planning, this main control platform is built for high-performance, eco-friendly robotic mowers.

Discover how we're transforming yard work with AI and embedded power.

[forlinx]

System Update Alert! Forlinx Embedded has officially launched Forlinx Desktop 24.04, marking full system compatibility for the FET3576-C System on Module (SoM) with Linux 6.1, Android 14, and the new Forlinx Desktop OS!

Powered by the high-performance Rockchip RK3576 with a 6 TOPS NPU, the FET3576-C SoM is now more versatile than ever—ready to tackle challenges in industrial automation, AIoT, edge computing, and more.

✅ Multi-system support ✅ Rigorous reliability testing ✅ Faster time-to-market ✅ Comprehensive tech support

💡 Whether you’re building smart terminals or embedded AI solutions, the FET3576-C SoM delivers power, stability, and flexibility.

🔗 Learn more:

Forlinx Embedded’s FET3576-C SoM (RK3576 processor) supports Linux 6.1, Android 14, and Forlinx Desktop 24.04. 8-core CPU, 6 TOPS NPU, rigor

[tootimetraveltimemachine]

RK3576 with UFS Storage: In-depth Analysis of Performance Advantages and Read-Write Test Data

In embedded storage field, UFS (Universal Flash Storage) is gradually emerging. UFS is a type of flash memory. Similar to eMMC, it integrates a control chip, accesses a standard interface, and undergoes standard packaging on the basis of NAND storage chips, thus forming a highly integrated storage chip. Due to its compact characteristics, UFS is widely used in embedded devices such as mobile phones and tablets. Moreover, since UFS far outperforms eMMC in terms of performance, it is often used in high-end products.

Advantages of UFS

1. Faster response speed for multitasking

Devices using UFS2.0. LVDS (Low-Voltage Differential Signaling) has a dedicated serial interface, allowing read and write operations to be carried out simultaneously. The CQ (Command) queue dynamically allocates tasks without waiting for the previous process to end. It’s like a car getting on the highway, with multiple lanes allowing high-speed and smooth travel. In contrast, mobile phones using EMMC must perform read and write operations separately, and the instructions are also packaged. In terms of speed, EMMC is already at a disadvantage, and it is naturally slower when performing multitasking. It likes traveling on an common two-lane road with speed limits.

2. Low latency, UFS has a 3-times faster response speed

When reading large-scale games and large-volume files, UFS2.0 takes less time. The time required to load a game is one-third of that of EMMC5.0. Correspondingly, when experiencing games, mobile phones with UFS2.0 have lower latency and smoother pictures.

3. Shorter loading time for photo thumbnails in the album

Taking the mobile phone album as an example, many people’s mobile phones are filled with hundreds or even thousands of photos. When you open the photo thumbnails in the album, you can clearly see the loading process. This is caused by the fact that the mobile phone cannot keep up with the refresh speed when reading photos from the flash memory. On a mobile phone with a good screen, the pictures will load smoothly as you scroll, while on a less-capable mobile phone, you can clearly feel the lag during loading.

4. Faster speed and lower power consumption

After the UFS chip improves its speed, it means that it takes less time to complete the same task. Higher efficiency means lower power consumption. When working simultaneously, the power consumption of UFS is 10% lower than that of eMMC, and it can save approximately 35% of power consumption in daily work.

UFS interface read-write performance test

RK3576 CPU also provides a UFS2.0 interface and an emmc5.1 interface.

FET3576-C SoM also reserves a UFS interface.

Refer to Rockchip’s official document “Rockchip_Developer_Guide_UFS_CN_V1.3.0” to conduct read-write tests on the UFS flash memory of OK3576-C.

Sequential write testroot@ok3576-buildroot:/# fio -filename=/dev /sda -direct=1 -iodepth 32 -thread -rw=write -bs=1024k -size=1G -numjobs=8 -runtime=180 -group_reporting -name=seq_100write_1024k seq_100write_1024k: (g=0): rw=write, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=psync, iodepth=32 ... fio-3.34 Starting 8 threads note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 Jobs: 8 (f=8): [W(8)][96.0%][w=359MiB/s][w=359 IOPS][eta 00m:01s] seq_100write_1024k: (groupid=0, jobs=8): err= 0: pid=1296: Thu Jan 1 00:01:32 1970 write: IOPS=332, BW=333MiB/s (349MB/s)(8192MiB/24631msec); 0 zone resets clat (msec): min=2, max=103, avg=23.55, stdev= 9.15 lat (msec): min=2, max=104, avg=23.77, stdev= 9.15 clat percentiles (msec): | 1.00th=[ 12], 5.00th=[ 14], 10.00th=[ 15], 20.00th=[ 16], | 30.00th=[ 18], 40.00th=[ 20], 50.00th=[ 22], 60.00th=[ 25], | 70.00th=[ 27], 80.00th=[ 31], 90.00th=[ 36], 95.00th=[ 41], | 99.00th=[ 53], 99.50th=[ 59], 99.90th=[ 68], 99.95th=[ 73], | 99.99th=[ 105] bw ( KiB/s): min=206590, max=432470, per=100.00%, avg=342387.68, stdev=7157.63, samples=385 iops : min= 196, max= 421, avg=331.98, stdev= 7.14, samples=385 lat (msec) : 4=0.11%, 10=0.49%, 20=42.49%, 50=55.44%, 100=1.45% lat (msec) : 250=0.01% cpu : usr=1.12%, sys=1.83%, ctx=18228, majf=0, minf=0 IO depths : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0% submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% issued rwts: total=0,8192,0,0 short=0,0,0,0 dropped=0,0,0,0 latency : target=0, window=0, percentile=100.00%, depth=32 Run status group 0 (all jobs): WRITE: bw=333MiB/s (349MB/s), 333MiB/s-333MiB/s (349MB/s-349MB/s), io=8192MiB (8590MB), run=24631-24631msec Disk stats (read/write): sda: ios=165/65464, merge=0/0, ticks=178/1074993, in_queue=1075171, util=99.64%

The print information is as described above, from which it can be known that the speed of sequential writing is 349 MB/s.

Sequential read testroot@ok3576-buildroot:/#fio -filename=/dev/sda -direct=1 -iodepth 32 -thread -rw=read-bs=1024k -size=1G -numjobs=8 -runtime=180 -group_reporting -name=seq_100read_1024k seq_100read_1024k: (g=0): rw=write, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=psync, iodepth=32 ... fio-3.34 Starting 8 threads note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 note: both iodepth >= 1 and synchronous I/O engine are selected, queue depth will be capped at 1 Jobs: 8 (f=8): [R(8)][100.0%][r=756MiB/s][r=755 IOPS][eta 00m:00s] seq_100read_1024k: (groupid=0, jobs=8): err= 0: pid=1329: Thu Jan 1 00:08:54 1970 read: IOPS=754, BW=755MiB/s (791MB/s)(8192MiB/10857msec) clat (usec): min=2331, max=16444, avg=10573.01, stdev=646.85 lat (usec): min=2335, max=16447, avg=10575.10, stdev=646.84 clat percentiles (usec): | 1.00th=[ 9896], 5.00th=[10159], 10.00th=[10159], 20.00th=[10290], | 30.00th=[10290], 40.00th=[10421], 50.00th=[10421], 60.00th=[10421], | 70.00th=[ 10552], 80.00th=[ 10683], 90.00th=[ 10945], 95.00th=[ 12518], | 99.00th=[ 13042], 99.50th=[ 13173], 99.90th=[ 13960], 99.95th=[ 15139], | 99.99th=[16450] bw ( KiB/s): min=762938, max=786629, per=100.00%, avg=772720.14, stdev=979.45, samples=168 iops : min= 740, max= 767, avg=749.19, stdev= 1.02, samples=168 lat (msec) : 4=0.01%, 10=1.65%, 20=98.34% cpu : usr=0.37%, sys=3.81%, ctx=24750, majf=0, minf=2048 IO depths : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0% submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% issued rwts: total=8192,0,0,0 short=0,0,0,0 dropped=0,0,0,0 latency : target=0, window=0, percentile=100.00%, depth=32 Run status group 0 (all jobs): READ: bw=755MiB/s (791MB/s), 755MiB/s-755MiB/s (791MB/s-791MB/s), io=8192MiB (8590MB), run=10857-10857msec Disk stats (read/write): sda: ios=64132/0, merge=0/0, ticks=544319/0, in_queue=544320, util=99.26%

The print information is as described above, from which it can be known that the speed of sequential writing is 791 MB/s.

With the continuous development of embedded storage technology and the increasing richness of application scenarios, embedded storage has become indispensable in many fields such as smart homes, in-vehicle infotainment systems, and mobile devices. In the future, both eMMC and UFS will play irreplaceable roles in different application fields by virtue of their respective characteristics.

[forlinx]

This article introduces a Smart Assistance and AI Visualization Gateway for power substations based on the RK3576 processor, enhancing grid

Introducing the Smart Assistance and AI Visualization Gateway Solution, powered by the RK3576 SoM. Designed to transform traditional power substation monitoring, this innovative gateway ensures real-time, multidimensional management, advanced AI analysis, and enhanced safety protocols.

🔧 Key Features:

4x Cortex-A72 & 4x Cortex-A53 cores 6 TOPS NPU for AI processing Comprehensive monitoring from transformers to environmental factors Linux OS, video processing, and more!

[forlinx]

MIPI-CSI Debugging for RK3576 Processor — Channel Analysis

#HowTo #Debug #MIPICSI for RK3576 Processor

Forlinx OK3576C SBC supports up to five camera inputs simultaneously via 5 x CSI-2 interfaces, so how to configure camera interfaces, whether it's for a single input or all five? Check out our latest blog to get the full instructions.

Learn Now:

This article explains the camera channels of the RK3576 processor and how to configure different output formats for MIPI - CSI cameras using