Rockchip designs some of the best SoCs for smart IoT, with high-end SOCs including RK3568, RK3576 and RK358. AI algorithms are embedded in these SOCs, making them easy to run AI audio and video. The AI cores these chips have will make AI analysis more accurate. This means they are ideal for developing network video recorders.

Why use AI in NVR network video recorders?

Artificial intelligence provides video recorders with various intelligent functions without the need for humans to constantly monitor the recordings in the background.

• Facial detection

• Visual fire warning

• Vehicle parking space recognition

• License plate detection

• Human detection Intrusion alarm

Unlike traditional NVRs, modern smart NVRs must have enough AI computing power to run resolution enhancement and intelligent monitoring algorithms, plus regular functions such as decoding, forwarding, storage, access, etc.

Why Rockchip SoCs are suitable for developing security NVR recorders

As mentioned earlier, all three Rockchip SoCs have AI cores, which makes them ideal for developing NVRs. In addition, these chips are equipped with back-end AI algorithms (RockIVA) to handle tasks such as facial detection, smart motion detection (for people, cars, and other objects), and perimeter detection. These help reduce development efforts and shorten time to market.

l  AI NPU enables super-resolution technology, with clearer image quality

The AI NPU in these SoCs performs image scaling or upscaling, which is essentially image stretching. Traditionally, this stretching makes the image blurrier. But the AI core and algorithm can accurately guess the color of the extra pixels, making high-resolution pictures or videos clearer.

l  With clearer images, the RockIVA AI algorithm can more accurately determine objects in the image for facial recognition and smart motion detection.

l  Intelligent search function – faster

l  Support multiple cascading methods – more flexible

l  Ethernet cascading mode

l  PCIe cascading mode

l  Short latency – low latency

How to choose between RK3568 vs RK3588 vs RK3576?

RK3588 SoC has the most powerful features

RK3588’s powerful SOC features are suitable for developing high-end NVRs with optimal upgrade capabilities and AI video/image analysis for face and object detection. RK3576 is second, and RK3568 is last, both of which can be used to develop mid-range or low-end solutions for customers who do not need a powerful NVR.

RK3568 vs. RK3588

RK3588 is the top SoC among the three NVR developments because it features a 6 TOPS NPU and a heterogeneous multi-core system consisting of a quad-core Cortex-A76 and a quad-core Cortex-A55. This AI computing power provides up to 8-way AI recognition capabilities, covering face detection, intelligent mobile detection, perimeter detection, etc.

The SoC can decode up to 32 1080p channels (supporting single-channel 3200W IPC access), and its encoding end processes half (265 channels) at the same resolution under H.16 or H.264 compression. Other notable features include:

• ARM Mali-G610 GPU

• Supports 4 screens

• Two Gigabit Ethernet ports for 1000Mbps access speed

• 64-bit LPDDR4/LPDDR4x/LPDDR5 memory

• Multiple connectivity options (PCIe 3.0, dual RGMII, USB 3.0, SATA 3.0, etc.)

The RK3568 has a less powerful 1 TOPS NPU paired with a quad-core Cortex-A55 CPU, but this combination is still capable of developing NVRs. Its high-speed connectivity options are almost similar to those in the RK3588, but the SoC can only handle 10 1080p (30FPS) decoding channels, two 1080p encoding channels, simultaneous output to two screens, and access speeds of up to 320Mbps.

The chip also has a less powerful GPU, as it has an ARM Mali-G52, and only has 32-bit DDR3 and DDR4 memory

The RK3576 performance is between the RK3568 and RK3588, as it has a similar 3588 TOPS NPU as the RK6, but with slightly less powerful heterogeneous cores (quad-core Cortex-A72 and Cortex-A53).

Its video encoding capabilities are also comparable to the RK3588 (32 channels), but it can only send its output to three screens via HDMI 2.0, HDMI 1.4 (MIPI converter), and VGA (dp converter).

To connect peripherals (such as hard disks for recording), the chip has USB 3.0, two RGMII, and two SATA 3.0, which can be expanded using a PM chip to connect up to 8 hard disks. On Ethernet, the SoC supports access speeds of up to 500Mbps.

The 3 SOCs each have their own advantages and disadvantages, and customers can choose as needed. At the same time, our company has launched the core boards and development boards of these three SOCs, and supports customized PCB boards with various requirements.