AMD Launches Next-Gen Single-Chip Solutions for AI-Driven Systems
Introducing Versal AI Edge Series Gen 2 and Versal Prime Series Gen 2
Advanced Micro Devices (AMD) is changing the game for AI-driven embedded systems with their latest innovation in adaptive SoCs technology. The company recently announced the launch of two groundbreaking products - the Versal AI Edge Series Gen 2 for AI-focused systems and the Versal Prime Series Gen 2 for classic embedded systems.
A Single-Chip Heterogenous Processing Solution
Traditional multi-chip solutions come with substantial overheads and aren't entirely optimized for the three key AI computational phases - preprocessing, AI inference, and postprocessing. AMD's new adaptive SoCs aim to overcome these obstacles by providing a single-chip solution that delivers streamlined processes and top-notch performance.
Optimized for Safety, Security, and Efficiency
The Versal AI Edge Series Gen 2 brings end-to-end acceleration uniquely designed to improve the safety and security of AI-driven embedded systems. With a high-performance processor that includes Arm CPUs and cutting-edge AI Engines, this product is perfect for applications requiring high-reliability, safety, and long lifecycles. Areas such as autonomous vehicles, industrial PCs, robots, and healthcare imaging are set to benefit immensely from this technology.
AMD's processing system features up to 8x Arm Cortex-A78AE processors, up to 10x Arm Cortex-R52 real-time processors, and support for the latest industry standards in connectivity and data transfer. Additionally, these devices meet rigorous operating requirements such as ASIL D / SIL 3 and are compliant with various safety and security standards, ensuring up to three times the efficiency in AI inference and up to ten times the scalar compute power for postprocessing.
Looking Ahead: Availability and Expectations
Salil Raje, AMD's senior vice president, expressed excitement over the new SoCs, emphasizing the growing need for efficient, all-in-one compute engines for embedded systems. Developers can currently access early documentation and evaluation kits, with the first silicon samples expected to arrive early next year and production commencing in late 2025.
As AMD leads the way in combining multi-engine processing on a single chip architecture, the future of AI-enabled embedded applications looks more efficient and promising than ever.
