The FZ5 EdgeBoard AI Box is a rugged and fanless AI edge computing box. With flexibility and stability, algorithms and models, it is designed for data computing at the edge. The computing of data takes place near the ‘edge’ of a network, where the data is being developed, instead of in a centralized data-processing center. On-device processing results in faster performance and response time, lower latency, power efficiency, improved security by retaining data on the device, and cost savings by minimizing data center transports. With a passive cooling enclosure design, the FZ5 EdgeBoard AI Box is capable of operating quietly in temperature ranging from -40°C to +70°C.

FZ5 EdgeBoard AI Box Front View

FZ5 EdgeBoard AI Box Rear View 

The FZ5 EdgeBoard AI Box is using Xilinx Zynq UltraScale+ ZU5EV MPSoC solution. The ZU5EV device features a 1.5 GHz quad-core ARM Cortex-A53 64-bit application processor, a 600MHz dual-core real-time ARM Cortex-R5 processor, a Mali400 embedded GPU, a H.264/H.265 Video Codec Unit (VCU) and rich FPGA fabric. The Box has a mainboard FZ5 Card inside with memory and storage capacities of 4GB/8GB DDR4, 32GB eMMC, 64MB QSPI Flash and 32KB EEPROM as well as a 32GB Micro SD card slot for extension. It has exposed rich peripheral interfaces to the Box enclosure including 4 x USB 3.0, Gigabit Ethernet, RS232, RS485, CAN, Debug, HDMI Input and Mini DisplayPort (DP).
 

FZ5 EdgeBoard AI Box

The FZ5 EdgeBoard AI Box is able to run PetaLinux and supports PaddlePaddle AI framework which is fully compatible to use Baidu Brain’s AI development tools like EasyDL, AI Studio and EasyEdge to enable developers and engineers to quickly leverage Baidu-proven technology or deploy self-defined models, enabling faster deployment. It can support 8- to 16-channel video decoding and 4- to 8-channel intelligent analysis and can be used directly in many fields such as multimedia, automotive ADAS, surveillance, industrial quality inspection, medical diagnosis and more others with creation of customers’ own applications. Its life cycle is as long as 10 years.

Software Architecture of FZ5 EdgeBoard AI Box

The FZ5 EdgeBoard AI Box has Computing Power up to 2.4TOPS and can run at 55 FPS for ResNet-50 in actual measurement. Power consumption is within 10W to 20W under typical working conditions. Measured data for some models are shown as below:

Relative Download and Links

You can download relative chip datasheet, products datasheet, user manual, software package from below. Detailed technical data available on request.

1	FZ5 Card Overview	1.22 MB
2	Zynq UltraScale+ MPSoC Product Selection Guide	1.88 MB
3	FZ5 Card Schematic	1.24 MB
4	FZ5 EdgeBoard AI Box Overview	1.02 MB
5	EdgeBoard AI Box & Accelerator Card (FZ5) Hardware Manual V1.0.0	3.37 MB
6	Baidu Brain EdgeBoard AI Box &Accelerator Card (FZ5) User Manual V1.0	0.97 MB

Other MYIR's Xilinx Products

http://www.myirtech.com/xilinxseries.asp

FZ3 Card - deep learning accelerator card (based on Xilinx Zynq UltraScale+ ZU3EG MPSoC)

Z-turn Board Single Board Computer (based on Zynq-7010 / 7020)

Z-turn Lite Single Board Computer (based on Zynq-7007S / Zynq-7010)

MYD-C7Z015 Development Board (MYC-C7Z015 CPU Module as core board)

MYD-Y7Z010/20 Development Board (MYC-Y7Z010/20 CPU Module as core board)

MYD-C7Z010/20 Development Board (MYC-C7Z010/20 CPU Module as core board)

MYD-CZU3EG Development Board (MYC-CZU3EG CPU Module as core board)

MYD-CZU4EV Development Board (MYC-CZU4EV CPU Module as core board)

VECP Starter Kit - a complete Vision Edge Computing Platform (based on Xilinx Zynq UltraScale+ ZU3EG MPSoC)