All News | Boards | Chips | Devices | Software | Archive | About | Contact | Subscribe
Follow LinuxGizmos:
Twitter Facebook Pinterest RSS feed
*   get email updates   *

Radxa Teases Upcoming AICore SG2300x Module with Octa-core SOPHON SG2300x SoC

Apr 15, 2024 — by Giorgio Mendoza 376 views

Radxa has recently previewed a new compact embedded module, the AICore SG2300x, built around the SOPHON SG2300x System-on-Chip. This module promises significant computational power for AI applications, featuring dual PHYs and a high-performance Tensor Processing Unit.

The AICore SG2300x features an octa-core Arm Cortex-A53 CPU, optimized for efficient processing. It includes a TPU capable of achieving up to 32 TOPS (INT8), 16 TFLOPS (FP16/BF16), and 2 TFLOPS (FP32). 


Additionally, the module supports prominent deep learning frameworks, including TensorFlow, Caffe, and PyTorch, facilitating its integration into a variety of AI projects and development environments.

SOPHON SG2300x block diagram
(click image to enlarge)

For video processing, the module can handle up to 32 channels of H.265/H.264 1080p video at 25fps, supporting advanced post-processing tasks such as image CSC, resize, crop, and more.

Storage capabilities include 16GB of LPDDR4X RAM and 64GB of onboard eMMC, enhanced by an additional 16MB SPI Flash. An SDMMC interface supports further storage expansion and operating system installation.

AICore SG2300x interfaces
(click image to enlarge)

The module’s design facilitates the integration of multiple peripherals through its PCIe X4 interface, which is capable of supporting high-speed devices like SATA, USB, RS232, and RS485. This adaptability makes the AICore SG2300x ideal for complex AI systems that require extensive peripheral connectivity.

Connectivity options include two onboard Gigabit Ethernet PHYs, one PCIe 3.0 X4 Root Complex, and one PCIe 3.0 X4 End Point. The module also features up to three UARTs, two PWMs, three I2C buses, and 32 GPIOs for additional interfacing capabilities

Radxa AICore SG2300x
(click image to enlarge)

Radxa has also included support for the SOPHON BMNNSDK toolkit with the AICore SG2300x, which offers tools for AI project development, including drivers, compilers, and inference deployment tools.

Specifications listed for the Radxa AICore SG2300x:

  • Memory/Storage:
    • 16GB LPDDR4X
    • 64GB Onboard eMMC
    • Onboard 16MB SPI Flash
  • Connectivity:
    • 2x Onboard Gigabit Ethernet PHY
  • Expansion:
    • 1x PCIe3.0 X4 RC
    • 1x PCIe3.0 X4 EP
  • I/O Interfaces:
    • 32x GPIO
    • 3x I2C, 3x UART
    • 2x PWM
    • 144-pin B2B connector
  • Operating Temperature:
    • -20°C to 70°C
  • Mechanical:
    • 59.8 x 55 x 9mm

Further Information

Details on pricing and availability of the AICore SG2300x have not yet been provided by Radxa. For more information, visit the Radxa product page.

(advertise here)

Print Friendly, PDF & Email

Please comment here...