Introduction - 1.0 English

DPUCVDX8H for Convolutional Neural Networks LogiCORE IP Product Guide (PG403)

Document ID
PG403
Release Date
2022-02-28
Version
1.0 English

The Xilinx® Deep Learning Processor Unit (DPU) is a series of soft IPs dedicated to convolutional neural networks acceleration. The DPUCVDX8H is a high-performance CNN inference accelerator specially designed for Xilinx® Versal® platform, which has a large AI Engine array. The DPUCVDX8H can be quickly deployed on the VCK5000 Alveo™ data center accelerator card to instantly meet the acceleration tasks of a variety of neural networks. It runs with a set of efficiently optimized instructions and can support most convolutional neural networks, such as VGG, ResNet, GoogLeNet, YOLO, SSD, and MobileNet.