Beilai SMA3588 series is a module based on the powerful Rockchip RK3588J/RK3588 processor (4xCortex-A76 cores + 4xCortex-A55 cores + 3xARM Cortex-M0 core), integrating an ISP and an eIQ Neutron Neural Processing Unit (NPU), with a maximum processing speed of 6 TOPS. This module complies with the latest SMARC 2.2 standard. The core board uses an MXM3 314-pin connector, Fully leveraging the powerful processing capabilities of the RK3588, it is suitable for high-end applications such as industrial control, machine vision, and edge AI computing.
| Item | Specification |
|---|---|
| Processor | Rockchip RK3588J |
| CPU | 4×Cortex-A76 @ 2.4GHz + 4×Cortex-A55 @ 1.8GHz |
| GPU | ARM Mali-G610 MP4 (OpenGL ES 3.2, OpenCL 2.2, Vulkan 1.2) |
| NPU | 6 TOPS (INT8, supports INT4/INT16/FP16/BF16/TF32) |
| Video Codec | 8K@60fps decode (H.265/H.264/VP9), 4K@60fps encode |
| Memory | 2GB/4GB/8GB/16GB LPDDR5/LPDDR5X |
| Storage | 16GB/32GB/64GB eMMC 5.1 |
| Power Input | +5V_SYS |
| Typical Power | Standby <1W, Full Load <5W |
Display Output: Quad-Display Heterogeneous – 1×HDMI 2.1 (4K@120fps) + 1×eDP 1.4 (4K@60fps) + 1×MIPI-DSI (4K@60Hz) + 1×DP (up to 7680×4320@30Hz)
Camera Input: 2×MIPI-CSI (CSI0: 2-lane, CSI1: 4-lane @ 4.5Gbps)
Network Interface: Dual Gigabit Ethernet (RGMII)
High-speed Storage: Native SATA 3.0 Interface
High-speed Expansion: Multi-lane PCIe – Supports PCIe 3.0 (1/2/4 lane) and PCIe 2.0 for Wi-Fi 6, 5G, AI accelerator cards, etc.
USB Interface: 2×USB 3.0 Host (5Gbps) + 1×USB 2.0 OTG + 1×USB 2.0 Host
Expansion Interfaces: 4×UART, 6×I2C, 2×SPI, 2×CAN, Multiple PWM/GPIO
Audio Interface: 1× 5-channel I2S audio interface
No Native LVDS Support: Can be implemented via MIPI-DSI to LVDS conversion chip.
No USB4/USB5 Support: Can be expanded via USB Hub on the carrier board.
Operating Temperature: -40℃ ~ +85℃ (Industrial-grade models)
Thermal Characteristics: θJA=15℃/W, θJC=3℃/W, Max Junction Temp 125℃
Cooling Recommendation: For high-performance applications, heatsink + forced air cooling or heat pipe + forced air cooling is recommended.
| System/Tool | Support Status |
|---|---|
| Linux | Kernel 4.19/5.10 (Fully Supported) |
| Android | 11/12 (Fully Supported) |
| Debian/Ubuntu | Fully Supported |
| Development SDK | RK3588 SDK V1.0+ |
| Driver Support | Full support for all display, camera, network, USB, PCIe, audio, and other peripherals. |
| Application Area | Recommended Config | Core Advantages |
|---|---|---|
| Ultra-HD Smart Cockpit/HMI | 8GB+64GB | Quad heterogeneous display support, powerful GPU for complex 3D UI & multitasking |
| High-end Machine Vision/AI Inspection | 8/16GB + large storage | 6 TOPS NPU + dual high-speed cameras + PCIe for multi-camera/AI card expansion |
| Edge AI Server/Gateway | 8/16GB + large storage | 8-core high-performance CPU + multi-lane PCIe expansion + dual Gigabit Ethernet for complex inference & data aggregation |
| High-end Digital Signage/Conference Display | 4GB+32GB | HDMI 2.1 4K@120Hz + DP 8K output, supports HDR10+ |
| Model | Memory | Storage | Temp Range | Grade |
|---|---|---|---|---|
| SMRC-RK3588-2G16G | 2GB LPDDR5 | 16GB | 0~70℃ | Commercial |
| SMRC-RK3588-4G32G | 4GB LPDDR5 | 32GB | -40~85℃ | Industrial |
| SMRC-RK3588-4G64G | 4GB LPDDR5 | 64GB | -40~85℃ | Industrial High-End |
Performance King: Same 8-core CPU, but GPU upgraded to Mali-G610 MP4, significantly boosting graphics and compute performance with support for modern APIs like Vulkan.
Display Champion: Uniquely supports quad heterogeneous display outputs (HDMI+eDP+DSI+DP), especially the addition of native DP interface supporting 8K output, topping the display system capabilities.
Expansion Master: Strongest PCIe capability, supporting multiple PCIe 3.0/2.0 lanes, providing top-tier expandability for high-speed wireless communication, multi-AI accelerator cards, and high-speed acquisition cards.
Most Comprehensive Interfaces: Builds upon RK3576J, increasing USB 3.0 Host to 2 ports, I2C to 6 channels, pushing interface richness to the limit of the SMARC standard.
Flagship Positioning: As the flagship of the RK35 series, it delivers ceiling-level performance for multi-screen interactivity, high-speed data throughput, complex AI inference, and graphics processing scenarios.
