Firefly AIO-1684XQ is a motherboard based on SOPHGO SOPHON BM1684X octa-core Cortex-A53 AI SoC delivering up to 32TOPS for AI inference, and designed for computer vision applications and video analytics.
The headless machine vision board is equipped with 16GB RAM, 64GB eMMC flash, and 128MB SPI flash, and comes with a SATA 3.0 port, dual Gigabit Ethernet, optional 4G LTE or 5G modules, four USB 3.0 ports, and a terminal block with two RS485 interface, two relay outputs, and a few GPIOs.
Firefly AIO-1684XQ specifications:
- SoC – SOPHGO SOPHON BM1684X
- CPU – Octa-core Arm Cortex-A53 processor @ up to 2.3 GHz
- TPU – Up to 32TOPS (INT8), 16 TFLOPS (FP16/BF16), 2 TFLOPS (FP32)
- VPU
- 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 – 16GB LPDDR4x
- Storage
- 64GB eMMC flash
- 128MB SPI Flash
- SATA 3.0 port
- MicroSD card slot
- Networking
- 2x Gigabit Ethernet RJ45 ports
- Optional 4G LTE (Mini PCIe + 1x nano SIM card slot) or 5G (M.2 + 2x nano SIM card slots)
- USB – 4x USB 3.0 ports
- Expansion
- 16-pole terminal block with 2x RS232, 2x Relay, 4x GPIOs, GND
- Mini PCIe socket for 4G LTE
- M.2 socket for 5G cellular module
- Power Supply – 12V/5A via 5.5 x 2.5mm power barrel jack
- Dimensions – 183 x 105 x 41.8mm
- Weight – 380 grams with heatsink
- Temperature Range – Operating: -20°C to 60°C; storage: -20°C to 70°C
- Humidity – 10%90% (non-condensing)
The specifications are quite similar to the company’s Firefly EC-A1684XJD4 FD edge AI computer minus the enclosure and support for HDMI output and WiFi. Since there’s no video output, I’d assume there must be a web-based control interface to configure the BM1684X motherboard and monitor up to 16 video streams.
Documentation specific to the AIO-1684XQ motherboard does not seem available yet, but the wiki for the Core-1684JD4 system-on-module, the little brother of the Core-1684XJD4 used there, indicates support for Debian 9 and Ubuntu 20.04. Programming is done through the Sophon3 SDK with support for Caffe, Darknet, MXNet, ONNX, PyTorch, PaddlePaddle, and TensorFlow frameworks, and that leverages open-source projects such as OpenCV and FFmpeg.
Applications include access control, video structuring, corrective and preventive maintenance, and equipment inspection using algorithms such as “person/vehicle/object” recognition, video structuring, and trajectory behavior. Firefly says the SOPHON motherboard can be used for smart surveillance/security, smart transportation, smart cities, industry 4.0, the smart grid, etc…
Firefly did not provide availability and pricing information for the AIO-1684XQ motherboard, but for reference, the EC-A1684XJD4 FD embedded computer is currently selling for $849, so I’d expect a price in this range, maybe a bit lower, for the motherboard. Further information may be found on the product page.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress
Why put so much AI stuff into it and then stick A53s into it? They could have done it with A55s and have a considerably boost.
Might also have been interesting if they could put a small GPU for applications where a display would be good(I assume that the one with the same SoC but has a HDMI port uses LLVMPipe).
The TOPs/Tflops numbers are well, very high. Higher than a lot of very potent iGPUs and NPUs too. Though there is a lot of questions in utilization, operations implemented, memory and etc.
Another thing is that this, unlike other AI dedicated SoCs, just seem dedicated to be a processing endpoint, rather than capture the footage and process it there, this could limit it a bit.
I also wonder if it actually runs passively or if it will throttle hard with no cooler as it is here.