MYiR-Tech
|
www.myirtech.com
6
/
27
FZ3 Deep Learning Accelerator Card
Hardware Manual
Chapter 2 SOC introduction
2.1 SoC features
The XCZU3EG used in this development platform belongs to the Zynq Ultr
MPSoC series SoC, integrating ARM quad-core Cortex-A53 (PS), dual-core Cortex-R5
(PS), Mali-400 MP2 graphics processing unit and Kintex Ultr FPGA (PL). The
quad-core Cortex-A53 has powerful computing capabilities, the dual-core Cortex-R5 can
be used for real-time processing applications, the Mali-400 MP2 can be used to
accelerate graphics processing, and the FPGA is fully programmable. With the
expandable I / O ports, it can adapt to a variety of application AI development scenarios.
The main chip of the AI development platform uses Xilinx XCZU3EG-SFVC784 devices
with a speed grade of -1 (MYC-CZU3EG SOM is designed to support all speed grades of
XCZU3EG-SFVC784 devices). XCZU3EG-SFVC784 supports 1.5GHz (max -1) APU
speed, 600MHz (max -1) RPU speed, 667MHz (max -1) GPU speed, and DDR4 speed
up to 2400Mbps. The XCZU3EG-SFVC784 device has the following resources:
Figure 2-1
