Edge AI module for deep learning and neural network processing

Impulse Embedded has announced the availability of the AI-MXM-H84A from Aetina. The device is an MXM Embedded Graphics Accelerator for AI processing to assist the development of Deep Learning and Neural Network processing at the edge.

Providing four Hailo-8 edge AI processors supplying a substantial 104 TOPS on a single embedded MXM graphics module, the device is ideal for machine builders and AI solution architects aiming at machine vision systems, autonomous vehicles and robotics.

The Hailo-8 Edge AI processor is the size of a penny and provides 26 TOPS AI inference performance with a typical power consumption of only 2.5W. ASICs, as the acronym suggests, are produced for a specific application, and reduced production costs, power consumption and excellent performance counter this limitation of design flexibility.

MXM, or Mobile PCI Express Module, was originally developed for laptop use. It was intended to offer an industry-standard socket for graphics processors, which would allow the easy upgrade of GPU capabilities with no need to upgrade the whole system or be tied to proprietary hardware for the lifetime of the device.

The MXM modules are developed specifically for harsh environments, and they aim to be integrated into tested and certified systems to handle vibration adequately. These modules are tough and can be exposed to the impacts and tremors seen in harsh industrial environments at the edge. Due to its small-sized form factor, the high-performance MXM 3.1 Type B module can be readily integrated into various embedded systems by developers and system integrators to manage heavy AI inference workloads with low latency and without the increased power consumption of conventional discrete graphics cards.