FPGA features neural network accelerating software

Mipsology Zebra neural network accelerating software has been integrated into the latest build of Xilinx’s Alveo U50 FPGA data center accelerator card. The card is a single-slot, low profile form factor passively cooled, adaptable accelerator operating up to a 75 W maximum power limit and can compute convolutional neural networks. It supports PCIe Gen3 x 16 or Gen4 x 8 and has 8 GB HBM2 and Ethernet networking capability. This plug-and-play FPGA solution provides broad application flexibility and longer life for deep learning inference acceleration as a CPU or GPU.

It is suited for both data center and large industrial AI applications including robotics, smart cities, imaging processing/video analytics, healthcare, retail, driver-assist cars and video surveillance.