AMD Ryzen™ AI Embedded P100 Series processors deliver powerful, highly scalable performance for Edge AI workloads, combining from 4 up to 12 Zen 5/5c CPU cores, RDNA™ 3.5 graphics, and an integrated XDNA™ 2 NPU with up to 50 TOPS of AI acceleration in an energy efficient System-on-Chip (SoC).

Scalable Edge AI Performance

Delivering up to 50 TOPS of AI performance, AMD Ryzen™ AI Embedded P100 processors are ideally suited for AI-enabled edge applications across industrial automation, medical technology, machine vision, smart city infrastructure, and professional gaming. By combining CPU, GPU, and NPU in a single energy-efficient System-on-Chip (SoC), the processors enable real-time AI processing directly at the edge without the need for cloud connectivity. This brings down latency while improving information security as data is processed right at the edge.

Built on advanced Zen 5/5c architecture with up to 12 CPU cores, the AMD Ryzen™ AI Embedded P100 processors set a new benchmark for scalable x86 performance at the edge. AMD combines high-performance cores for demanding workloads with low-power, efficient cores that reduce power consumption for less demanding or background activities. What's unique is that both core types are built on the same microarchitecture.  

This means all cores offer the full Zen 5 feature set. The main difference is the maximum core frequency, bringing several advantages for embedded applications. The consistent execution environment across all cores handles instructions identically, simplifying programming of real-time applications where timing is critical. Task scheduling does not need complex handshaking between different core types, as the OS sees a homogenous pool of computing resources for CPU-centered tasks. Moreover, this approach is one reason for the excellent power scalability of the AMD Ryzen embedded P100 series.  

AMD’s integrated XDNA™ 2 NPU provides up to 50 TOPS of AI performance, enabling low-latency AI inference. It allows demanding AI workloads, such as computer vision, anomaly detection, optical character recognition (OCR), and real-time data analytics to run locally on edge devices. Unlike cloud-connected devices, this allows for high determinism through local processing.

From cost-optimized to high-performance embedded edge designs

The configurable TDP range from 15 to 54 W allows scalable designs from passively cooled systems to high-performance platforms, eliminating the need for additional AI accelerator cards while simplifying overall system design.

By integrating CPU, GPU, and NPU capability on a single SoC with a configurable TDP from 15 to 54 W, developers can implement various designs from passively cooled, SWaP-C (size, weight, power and cost) optimized systems to high-performance edge platforms with one single platform maintaining a streamlined, cost-efficient design with reduced design efforts and bill of material.