SOPHON BM1684/BM1684X Edge AI computer delivers up to 32 TOPS, decodes up to 32 Full HD videos simultaneously

Firefly EC-A1684JD4 FD and EC-A1684XJD4 FD are nearly identical Edge AI embedded computers based on respectively SOPHON BM1684 and BM1684X Arm AI SoC delivering up to 32 TOPS of AI inference, and capable of decoding up to 32 H.265/H.264 Full HD videos simultaneously for video analytics applications.

The BM1684(X) SoCs are equipped with eight Cortex-A53 cores clocked at 2.3 GHz to run Linux, and the systems come with up to 16GB RAM, 128GB flash, two Gigabit Ethernet ports to receive the video streams, one HDMI output up to 1080p30 for monitoring,  as well as RS232 and RS485 DB9 connectors, and a few USB ports.

Sophon BM1684 BM1684X Edge AI computer

Firefly EC-A1684JD4 FD and EC-A1684XJD4 FD specifications:

  • SoC – SOPHGO SOPHON BM1684/BM1684X
    • CPU – Octa-core Arm Cortex-A53 processor @ up to 2.3GHz
    • TPU
      • BM1684
        • 64 NPU arithmetic units with each NPU containing 16 EU arithmetic units, or 1,024 EU in total
        • Up to 17.6 TOPS (INT8), 2.2TOPS (FP32), or 35.2 TOPS (INT8 with Winograd Convolution Optimization enabled)
      • BM1684X
        • Up to 32TOPS (INT8), 16 TFLOPS (FP16/BF16), 2 TFLOPS (FP32)
    • VPU
      • BM1684
        • Up to 32-channel H.265/H.264 1080p30 video decoding
        • 1080p50 video encoding
        • MJPEG image encoding and decoding up to 1080p480
      • BM1684X
        • Up to 32-channel H.265/H.264 1080p25 video decoding
        • Up to 32-channel 1080p25 HD video processing (decoding + AI analysis)
        • Up to 12-channel H.265/H.264 1080p25fps video encoding
  • System Memory – 6GB, 12GB, or 16GB LPDDR4/LPDDR4X
  • Storage
    • 32GB, 64GB, or 128GB eMMC flash
    • 128MB SPI Flash
    • M.2 SATA3.0 socket (for M.2 2242 SSD)
    • MicroSD card slot
  • Video and audio output – HDMI port up to 1080p30
  • Networking
    • 2x Gigabit Ethernet RJ45 ports
    • 2.4GHz/5GHz dual-band 802.11a/b/g/n/ac WiFi 5
    • Optional 4G LTE (Mini PCIe) or 5G (M.2)
  • USB – 2x USB 3.0 ports, 2x USB 2.0 ports
  • Serial – RS232 and rS485
  • Power Supply – 12V/5A via 5.5 x 2.5mm power barrel jack
  • Dimensions – 182.5 x 150.0 x 66.5mm
  • Weight – Around 1.8kg
  • Temperature Range – Operating: -20°C to 60°C; storage: -20°C to 70°C
  • Humidity – 10%~90% (non-condensing)

SOPHON BM184 Edge AI embedded computer portsFirefly’s latest Edge AI computers run Linux, but the Wiki has not been updated yet. However, the Wiki for the earlier EC-A1684JD4 model is up and we can see the Sophon3 SDK relies on Debian 9, and an Ubuntu 20.04 image is said to be coming soon. The AI part supports a range of frameworks such as TensorFlow, Caffe, PyTorch, Paddle, ONNX, MXNet, Tengine, and DarkNet.


Applications include access control with face recognition, video structuring, corrective and preventive maintenance, and equipment inspection. Firefox says the SOPHON embedded computers can be used for visual computing, edge computing, smart transportation, smart classroom, unmanned supermarkets, video surveillance, and more.

Sophon BM1684 Block Diagram
SOPHON BM1684 Block Diagram

SOPHGO, the company behind the SOPHON SoCs, also shared some performance and efficiency data of the BM1684 and BM1684X SoCs again “N” (NVIDIA), “A” (Alibaba?), “H” (Huawei?) and  “M” (help me there) competitors using standard Resnet and VGG AI models.

BM1684 Resnet 50 VGG AI performanceEC-A1684XJD4 FD vs NVIDIA vs Huawei

We don’t know which exact competitor platforms were used so the results are not overly useful… The bottom illustration probably uses the Jetson Nano seeing how low the score of the “N” platform is.

I can’t see the new EC-A1684JD4 FD and EC-A1684XJD4 FD Edge AI embedded computers for sale just yet, but the EC-A1684JD4 is going for $799 and up on the company’s store, and $969.90 on Aliexpress with unspecified RAM and storage capacities. Further details may be found on the respective product pages.

Share this:

Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress

ROCK 5 ITX RK3588 mini-ITX motherboard
Notify of
The comment form collects your name, email and content to allow us keep track of the comments placed on the website. Please read and accept our website Terms and Privacy Policy to post a comment.
1 year ago

1 device for 32 cameras with face detection. Creepy and efficient. Efficiently creepy

1 year ago

Welcome to the future.

1 year ago

X version would be perfect for Unmanic, and if these video codecs are capable of handling 4K to some extent, then one could argue that Sophon could make some extra money with streaming services or video hosts in general.

Khadas VIM4 SBC