- Veröffentlicht: 23. Juni 2020
iWave Systems has launched an artificial intelligence processor that can support convolutional neural networks (CNNs), based around a Xilinx Zynq UltraScale+ FPGA.
“Edge AI devices are expected to run complex neural networks and deep learning algorithms whilst maintaining latency, power efficiency and accuracy. These devices are finding their way into smart security, autonomous driving, robotics and medical imaging,” according to IWave.
Called Corazon-AI, the inference engine comes in a 100mm x 72mm pico-ITX form factor and can support up to eight internet protocol cameras.
Connectivity options include dual Gbit Ethernet, Wi-Fi, Bluetooth LE and 3G/4G, as well as four 12.5Gbit/s transceivers that implement various interfaces including HDMI.
Processing resources include the FPGA’s quad-core (or dual-core) Cortex-A53 (MPCore) running at up to 1.5GHz, dual Cortex-R5s at 600MHz, a Mali-400MP2 GPU, up to 256,000 logic cells and and Xilinx’ Vitis AI stack. The unit can provide “a scalable multi-dimensional parallel architecture capable of performing major convolutional calculations, batch normalisation through deep pipelined computing engines”, said IWave. “They consist of specialised encoders and decoders which squeeze network model size and have a unique instruction scheduler that dynamically adjusts FIFO.”
12V power is needed, through a 2.5mm jack, and operation is over -40°C to +85°C.
Applications including number plate recognition, traffic management systems and automatic video inspection are foreseen.
Autor(en)/Author(s): Steve Bush
Quelle/Source: Electronics Weekly, 16.06.2020