May in Review: AI at the Edge
Author: Dylan Mcgrath
Chipmakers continue to push more AI capability to the IoT edge to reduce the reliance of IoT devices on connectivity to the cloud and the latency involved in AI decision-making while improving privacy and security. So far in 2024, vendors have launched a number of products that pack more neural processing capability into devices that have the footprint and power profile to exist at the edge.
In May, we covered three products that tie perfectly to this trend: Synaptics’ Astra platform and SL-Series processors, Arm’s EthosU85 neural processing unit (NPU), and AMD’s second-generation Versal AI Edge SoCs.
May also saw the continuation of other prominent 2024 trends in microprocessors: the addition of more AI capability in intellectual property (IP) for automotive applications and the dramatic escalation in GPU horsepower in server racks for large AI model training. In addition, we also looked at AMD’s latest Ryzen PRO series models, which cascade new features and capabilities introduced in the company’s Ryzen 8000G series processors launched earlier this year and are now available to enterprise PCs.